Upload
khangminh22
View
0
Download
0
Embed Size (px)
Citation preview
Master
Reference
The Interest of Controlled Language in Machine Translation A Study
Carried out on the Basis of the UEFA European Football
Championship 2010-12
THODEN, Annika
Abstract
The concrete idea for this study stemmed from the internship I carried out in the German
section of the UEFA Language Services unit (SLAN) in summer 2011. SLAN's translation
workflow is characterised by the use of a variety of CAT-tools (Translation Memory and
terminology software, electronic dictionaries, etc.). Due to my prior knowledge of MT, I had the
idea of examining whether it would be profitable for UEFA to translate certain texts using a MT
system. Within the limited framework of this pilot projet, I narrowed the concrete goal of this
Master's thesis down to evaluating the quality of MT output (direction of translation:
English-German), i.e. examining whether UEFA regulations - after undergoing a control of the
SL text - are suitable for MT using Systran, since the output quality is in any case the crucial
factor for or against the implementation of a MT-based workflow.
THODEN, Annika. The Interest of Controlled Language in Machine Translation A Study
Carried out on the Basis of the UEFA European Football Championship 2010-12.
Master : Univ. Genève, 2012
Available at:
http://archive-ouverte.unige.ch/unige:22853
Disclaimer: layout of this document may differ from the published version.
1 / 1
The Interest of Controlled Language
in Machine Translation
A Study Carried out on the Basis of the
UEFA Regulations of the
UEFA European Football Championship 2010-12
Master’s Thesis
Annika Thoden
Faculté de Traduction et d’Interprétation,
Université de Genève
Département de traitement informatique multilingue
Thesis director: Prof. Pierrette Bouillon
Thesis jury: Florian Simmen, Head of
UEFA Language Services
August 2012
Table of contents
1. Introduction ................................................................................................ 1
2. An introduction to Machine Translation ................................................. 4
2.1 Definition of Machine Translation .................................................... 4
2.1.1 Fully automatic high quality translation (FAHQT) ................ 5 2.1.2 Human-aided machine translation (HAMT) ......................... 6 2.1.3 Machine-aided human translation (MAHT) .......................... 7
2.2 Historical overview and state of the art ........................................... 8 2.3 MT architectures ............................................................................ 11
2.3.1 Direct translation approach ................................................ 12 2.3.2 Indirect translation approaches.......................................... 12 2.3.3 Corpus-based translation approaches............................... 15
2.4 Controlled languages ..................................................................... 16 2.5 Post-editing .................................................................................... 19 2.6 Systran ........................................................................................... 23 2.7 Suitable text types and applications for MT .................................. 25 2.8 Translator’s perception of MT ........................................................ 27 2.9 Conclusion ..................................................................................... 28
3. Context: Union des Associations Européennes de Football (UEFA) 30
3.1 About UEFA ................................................................................... 30 3.2 Overview of the UEFA Language Services unit ............................ 32 3.3 UEFA regulations ........................................................................... 34
3.3.1 Creation and revision of UEFA regulations ....................... 34 3.3.2 Rules for the structuring and drafting of
UEFA’s regulatory instruments .......................................... 36 3.3.3 Translation workflow of UEFA regulations......................... 37
3.4 Conclusion ..................................................................................... 39
4. The study .................................................................................................. 40
4.1 General study setup ....................................................................... 40
4.1.1 Working with Systran Dictionary Manager......................... 42 4.1.2 Control of S1 and creation of MT-oriented writing
rules for UEFA regulations ................................................. 45 4.1.3 Double-checking: Control of S2 ......................................... 47 4.1.4 Post-editing of TC1 ............................................................ 54
4.2 Problems encountered................................................................... 57
4.2.1 Software problems ............................................................. 57 4.2.2 Translation problems remaining despite control................ 60
4.3 Conclusion ..................................................................................... 72
5 Results ...................................................................................................... 74
5.1 Quantitative evaluation .................................................................. 75 5.2 Qualitative evaluations ................................................................... 81
5.2.1 Evaluation of source text material...................................... 82 5.2.2 Evaluations of target text material ...................................... 84
5.3 Overview of results ........................................................................ 98
6 Discussion .............................................................................................. 100
6.1 Critique of the study ..................................................................... 100 6.2 Further steps ................................................................................ 101 6.3 Conclusion ................................................................................... 103
7 Reference list ......................................................................................... 106
8 List of figures ......................................................................................... 113
9 List of tables........................................................................................... 114
8 Annexes .................................................................................................. 115
Annex A: Overview of UEFA divisions and units .................................... 115 Annex B: Overview of UEFA Events SA ................................................. 117 Annex C: S1 ............................................................................................. 118 Annex D: S2 ............................................................................................. 124 Annex E: Table comparing S1 to C1 ....................................................... 129 Annex F: MT-oriented writing rules for UEFA regulations ...................... 146 Annex G: C1 ............................................................................................ 151 Annex H: Table comparing S2 to C2....................................................... 158 Annex I: C2 .............................................................................................. 180 Annex J: PE1 ........................................................................................... 186 Annex K: PE2........................................................................................... 193 Annex L: PE3 ........................................................................................... 200 Annex M: Questionnaire 1: UEFA SLAN – English translation .............. 207 Annex N: Questionnaire 2: UEFA SLAN – German translation.............. 211 Annex O: Questionnaire 3: Translators with different professional backgrounds ....................................................... 213 Annex P: Results of Questionnaire 3, question 2 ................................... 217
1. Introduction
1
1. Introduction
As a result of the information age, the translation industry has undergone
significant changes over recent decades. Whereas translators worked with pen and
paper, typewriters and paper dictionaries until the early 1990s, they benefit today
from a large range of computer tools in order to maximize their output, in both
economic and qualitative terms. The use of computer-aided translation tools, or CAT-
tools, such as Translation Memories (TM), electronic dictionaries, terminology
databases and corpora, has become standard for most translation services, agencies
and freelance translators, not to mention word processors and the internet as
indispensable parts of the modern translator’s working environment. The ever-
increasing demand for translation also entails an expanding use of Machine
Translation (MT) in order to “increase productivity whilst keeping costs as low as
possible” (de Preux 2005: 1). What is more, there is a general “change in
expectations with regard to the type and quality of translated material” (Allen 2003:
300), leading to increased demand for lower quality translations. Translators are
therefore often hostile towards MT, fearing that computers could take over their jobs,
instead of seeing the advantages they could derive from the use of MT. However, it is
indisputable that fully automatic and high speed translation without human
intervention cannot for the foreseeable future be reconciled with high quality
translation output. Translators will therefore not lose their jobs to machines, but rather
see a further change in their working environment, as was the case with the transition
to the information age. For this reason, it is particularly important for future translators
to be trained in modern translation technology, thereby gaining an understanding of
its capabilities and limitations. The University of Geneva offers a Master’s programme
with a specialisation in Multilingual Translation Technology, enabling students to
attend a wide range of seminars dealing amongst others with MT, CAT-tools,
terminology and software localisation. It is against this backdrop that the subject area
of this Master’s thesis was chosen.
However, the concrete idea for this study stemmed from the internship I
carried out in the German section of the UEFA Language Services unit (SLAN) in
summer 2011. SLAN’s translation workflow is characterised by the use of a variety of
CAT-tools including the TM software SDL Trados Studio 2011, the terminology tool
Trados MultiTerm and electronic dictionaries. During the internship, I gained insight
into the different text types taken on by the unit, which are translated with or without
the use of a TM depending on their content. Texts such as match reports,
programmes or strategies which require creativity and distance from the source text
1. Introduction
2
are simply translated in Microsoft Word, whereas more functional, repetitive texts
such as regulations, circular letters and legal texts are generally translated with a TM.
Due to my prior knowledge of MT, I had the idea of examining whether it would be
profitable for UEFA to translate certain texts using a MT system.
After discussing the idea with Florian Simmen, Head of UEFA Language
Services and receiving the green light for the project, the study setup began to
materialise. Within the limited framework of this pilot project, I narrowed the concrete
goal of this Master’s thesis down to evaluating the quality of MT output, i.e. examining
whether UEFA regulations – after undergoing a control of the SL text – are suitable
for MT using Systran, since the output quality is in any case the crucial factor for or
against the implementation of a MT-based workflow. The text type chosen was UEFA
regulations since their translation is rather time-consuming, owing to their length and
periodic revision. The direction of translation taken into consideration within the
limited scope of my pilot study was English-German, given that UEFA regulations are
initially drafted in English. Systran was the software chosen for the study because it is
a user-friendly system which has proven itself in practice in a large variety of
companies and organisations.
When it comes to the companies and organisations working with MT, their
major concern is obviously to improve the quality of the system’s raw output. In that
connection, “[o]ne promising approach seems to be to influence the input text,
especially by constraining the lexical items present and the grammatical constructions
used – in other words, restricting input text to a controlled language” (de Preux 2005:
1). Due to prior knowledge of controlled language acquired in MT classes in the
course of my Master’s degree, I set up the hypothesis that the use of controlled
language would strongly improve the machine translation output of the chosen text
material. As the title of this thesis suggests, the study also aimed at drawing a more
general conclusion concerning the impact of controlled language in machine
translation.
If the evaluations of the controlled MT input and the German MT output yield
positive results and UEFA generally approves the use of controlled language and MT
for the translation of regulations, a second step would be a thorough operational
evaluation on site, comparing the turnaround time for UEFA regulations translated in
a MT-based and the existing CAT-tool-based workflow. By this means, it could then
be established if it would be profitable for UEFA to translate its regulations using the
MT system Systran.
The introductory chapter of this paper will provide the reader with general
background information about MT in order to lay the theoretical foundation of the
1. Introduction
3
study. Chapter two will present the different degrees of human involvement in and
mechanisation of MT, before providing a brief historical overview and state of the art
of MT research. It will also give an overview of the various existing MT architectures.
Subsequent sections will give a thorough introduction to controlled languages, post-
editing and Systran. The chapter will conclude with an overview of suitable text types
for and applications of MT and a discussion of translators’ perception of MT. Chapter
three will be devoted to the context of the study, namely an overview of UEFA, its
Language Services unit (SLAN) and thorough background information about UEFA
regulations, their creation and revision as well as the rules for their structuring and
drafting, followed by a presentation of the translation workflow for regulations within
SLAN and a discussion of the reasons supporting the introduction of MT for this text
type. Chapter four will cover the practical aspects of the study setup, that is: creating
an individual user dictionary and TM with the Systran Dictionary Manager; controlling
the text material then establishing MT-oriented writing rules for UEFA regulations;
double-checking these writing rules against a second source text; and post-editing the
MT output. The last section of chapter four will discuss the various types of software
and translation problems encountered while working with Systran. Chapter five will
then assess the results obtained in chapter four and will be divided into two main
sections. In the interest of being as objective as possible, the assessment was carried
out in two stages: a quantitative evaluation using the analysis function of SDL Trados
Studio 2011 and different types of qualitative evaluations to corroborate the results of
the quantitative evaluation. The former provided quick and convenient objective data
leading to clear results regarding my hypothesis that the use of CL would improve the
MT output of the text material. The latter consisted in an evaluation of the controlled
source text material by an English translator from UEFA SLAN, followed by an
evaluation of the target text material by different subject groups: the German
translators from UEFA SLAN assessed sentences from different MT stages as well as
from the official human translation, while a subject group made up of translators of
different professional backgrounds assessed the overall quality of the controlled MT
output after it had undergone a rapid post-edit. The last chapter will discuss the final
decision made on the basis of the study, present the study’s flaws and make
recommendations for further steps to be undertaken if SLAN decided to implement
MT. Finally, an overall conclusion will be drawn from the pilot project and the results
placed in a broader context, thereby allowing additional general findings to be made.
2. An introduction to Machine Translation
4
2. An introduction to Machine Translation
This introductory chapter aims to provide the reader with general background
information about Machine Translation (MT) in order to lay the theoretical foundation
for the study. It will open with a definition of MT, provide a historical overview of the
development of MT and describe the state of the art of MT research. It will then define
the different types of MT architectures and how they function, discuss controlled
languages (CL) and post-editing (PE), and briefly present Systran, the MT software
used in the study. Lastly, it will briefly expound on suitable text types for and
applications of MT and conclude with translators’ perception of MT.
2.1 Definition of Machine Translation
The following preliminary definitions are mainly based on Hutchins and
Somers (1992: 6-7, 77 and 147-151), Arnold et al. (1994: 1) and Quah (2006: 6-14).
Other sources will be mentioned separately.
According to Quah, the term Machine Translation (MT) has not always been
defined in the same way: Originally, it “referred only to automatic systems with no
human involvement” (Quah 2006: 8). Nowadays, the term is generally seen in a
broader perspective, with Arnold et al. (1994: 1) defining MT as “the attempt to
automate all, or part of the process of translating from one human language to
another.” Since no other term has been created so far, it is still used for fully
automatic systems as well as for systems requiring human involvement (Quah
2006: 9).
Likely the most used and established classification for MT comes from
Hutchins and Somers (see fig. 1), which has been taken up by various authors in the
field, including Quah (2006: 7), Schäfer (2002: 31) and Ramlow (2008: 50). According
to the classification, translation can have quite different degrees of human
intervention and of mechanisation, ranging from fully automatic machine translation to
traditional human translation with pen and paper. According to Hutchins and Somers,
the two most used computer-assisted (or computer-aided) translation types (CAT) are
human-aided (or human-assisted) machine translation (HAMT) and machine-aided
(or machine-assisted) human translation (MAHT). The difference between the two lies
in perspective: whereas the main focus of HAMT is on the machine, that of MAHT is
on the human translator.
2. An introduction to Machine Translation
5
Nevertheless, it is important to bear in mind that, nowadays, many tools
“become multifunctional” (Quah 2006: 6) such that the distinction between HAMT and
MAHT involves a certain “degree of fuzziness at the edges” (Allen 2003: 13). Some
authors consider only MAHT as a synonym of CAT, whereas they see HAMT as a
category on its own. At the same time, the classification established by Hutchins and
Somers “remains useful as a point of reference for classifying translation in relation to
technology” (Quah 2006: 6).
2.1.1 Fully automatic high quality translation (FAHQT)
The term FAHQT was introduced by the first full-time researcher in MT,
Yehoshua Bar-Hillel. As early as the 1950s, he was convinced that a fully automatic
translation process, without any human involvement, generating a high quality output
“undistinguishable from [that] of human translators” (Hutchins and Somers 1992: 7),
was an unrealistic objective and in principle impossible. Rather, he encouraged
researchers in MT to enhance their efforts in the field of human-aided machine
translation (HAMT). One of the reasons for his reluctance towards FAHQT was the
fact that, in his opinion, computers were not able to learn the “real world knowledge”
of which human translators dispose (Schäfer 2002: 31, Hutchins and Somers
1992: 148). To this day, there is consensus on the fact that FAHQT is impossible for
most text types: fully automatic and high speed translation cannot currently – or for
the foreseeable future – be reconciled with high quality translation output (Hutchins
and Somers 1992: 148). MT’s current main objective is “still to generate translation
automatically, but it is no longer required that the output quality is high, rather that it is
fit for purpose” (Quah 2006: 7). Nevertheless, there remain cases of FAHQT systems
designed for specific, highly restricted subject areas. Examples are METEO, a system
developed at the University of Montreal, that was designed for the translation of
weather bulletins between English and French, or TITUS, a system developed by the
Institut Textile de France, used since 1970 for translating specialised texts in the
textile industry (Schäfer 2002:32). Due to the fact that these texts are written in a
certain sublanguage, i.e. in the specialised language of a given sub-domain, they are
Fig. 1: “Human and machine translation” (Hutchins and Somers 1992: 148)
2. An introduction to Machine Translation
6
assumed to be less complex and thus easier to parse by a machine (Schäfer
2002: 32).
2.1.2 Human-aided machine translation (HAMT)
It can be said that the focus of MT has shifted from FAHQT to HAMT. With
HAMT, it is the machine that takes over the task of translating while the human
assists the machine prior to (pre-editing), during (interactive mode) and/or after (post-
editing) the translation process (see fig. 2).
As shown in Quah’s HAMT model, the source text destined to automatic
translation can be unedited, pre-edited or controlled. Pre-editing consists in
preparing a text for MT by detecting and resolving problems with which the machine
cannot cope, such as spelling and grammar mistakes, ambiguities, idioms or words
and proper nouns that are not in the MT dictionary. A further step of pre-editing would
be the use of controlled language rules that restrict the sentence structures, grammar
and lexis of a given text. These rules are usually applied during the writing process,
not as a post-editing instrument (see section 2.4). A pre-edited text should ideally not
contain any ambiguities. Post-editing, on the opposite end of the HAMT process, is
the task of correcting the machine output (the target text), which may be done by a
human translator or editor (see section 2.5). This step is important in order “to bring
the text to a certain pre-determined standard in terms of language style and
appropriate use of terms” (Quah 2006: 11).
If these two steps, pre- and post-editing, are integrated into the translation
process carried out by the machine, it is called an interactive mode, which is
represented in the bottom box of figure 2. In this case, the human translator or editor
can assist the machine during the translation process involving the resolution of
ambiguities by interpreting structures, selecting lexical items (Hutchins and Somers
1992: 151), choosing between parts of speech, etc. When pre-editing is carried out
interactively, the system foresees the problems that may arise and, as a result,
highlights or flags elements, such as unknown words or proper nouns, so that the
Fig. 2: “Human-aided machine translation model” (Quah 2006: 12)
2. An introduction to Machine Translation
7
user can provide them before the translation process begins (Hutchins and Somers
1992: 151). When it comes to interactive post-editing, “the system alerts users to
places in the text where alternatives have been offered and asks for choices to be
made” (Hutchins and Somers 1992: 151). The opposite would be the so-called batch
or non-interventionist mode.
For the purpose of this study, I was interested in human-aided machine
translation, including pre-editing (through controlled language rules) and post-editing.
2.1.3 Machine-aided human translation (MAHT)
Unlike HAMT, where the machine generates the translation, with MAHT it is
human translators who carry out this task. They are supported by a wide range of
machine aids, such as translation memories (TMs), electronic glossaries, electronic
dictionaries, terminological databases, spell-checkers, grammar and style checkers,
corpora, automatic term look-up, etc. (see fig. 3). The use of these computer aids
thus enables translators “to have under [their] own control the production of high
quality translation” (Allen 2003: 29).
Today, MAHT has generally become standard in the translation industry. “The
only real question is the extent to which computers are used” (Furlani 2009: 5).
According to this statement, not only is the left end of figure 1 (FAHQT) irrelevant (as
seen in section 2.1.1), but the other end of the scale – traditional human
translation (HT) – can now be disregarded. In the highly competitive translation
market, it is inconceivable for translators not to take advantage of any computer tool
that would help them save time, and thus money.
Probably the best known MAHT tool for translators is the translation memory
system. “A TM is a type of linguistic database that is used to store source texts and
their translations. The texts are broken down into short segments that often
correspond to sentences” (Bowker 2002: 92). When an exact, i.e. a
100 percent match is found, the segment is automatically inserted into the translation
editor. This way, a translator can “reuse or ‘recycle’ previously translated segments”
(Bowker 2002: 92) or fragments, a method that is also called “collage translation” by
Fig. 3: “Machine-aided human translation model” (Quah 2006: 13)
2. An introduction to Machine Translation
8
Fig. 4: adapted from “Chronology of machine translation development” (Quah 2006: 58)
Mossop (2006: 786). The translator can usually define a sensitivity threshold, which is
normally set between 60 and 70 percent; a match that is not exact is known as a
fuzzy match (Bowker 2002: 92).
In a typical translator’s workbench or workstation, the above mentioned
translation-support tools are integrated into one single system (Quah 2006: 13). This
means that the workbench incorporates “powerful terminology-management systems
and bilingual concordancers, and some TMs can be integrated with machine-
translation systems” (Bowker 2002: 123). These systems can be divided into those
working with existing word processors and those that come with an independent text
editor. Some examples of commercial suppliers of translator’s workbenches are SDL
Trados by SDL International, Across by Across Systems, Transit by Star Group and
Déjà Vu by Atril.
2.2 Historical overview and state of the art
The following historical sketch is based on Arnold et al. (1992: 12-14),
Hutchins and Somers (1992: 5-9), Quah (2006: 58-66) and Schwarzl (2001: 13-20).
The starting point is an adaptation of Quah’s “Chronology of machine translation
development”, which illustrates the three generations of MT research and the different
MT architectures in relation to a time continuum. The latter will be detailed in the next
section (2.3).
The idea of ‘mechanical translation’ can be traced back to the 17th century,
when Descartes, amongst others, had the idea of creating dictionaries on the basis of
universal numerical codes. However, the pioneering years of Machine Translation did
not begin before the second half of the 20th century. It is generally accepted that the
memorandum written by Warren Weaver to the Rockefeller Association in 1949
marked the beginning of MT research. He proposed using methods such as
cryptographic techniques, statistical analysis or information theory in order to
mechanise translation. The main hypothesis of Weaver’s memorandum was the
2. An introduction to Machine Translation
9
existence of a universal basis for language which could be used as an interlingua in
the translation process, “a linguistic no-man’s-land in between various languages”
(Schwarzl 2001: 14). His idea was a precursor of the future ‘interlingua’ approach
(see below) and generated a large amount of interest, not least because the
fascination with gathering intelligence by means of Machine Translation was huge in
the first years of the Cold War. In consequence, a great number of research groups
was created in Europe and in the United States. In 1951, the first full-time MT
researcher, Bar-Hillel (see section 2.1.1), was appointed. Three years later,
Georgetown University and IBM gave the first public demonstration of an MT system.
They jointly created a system that was only composed of 49 sentences and an input
of 250 words and six grammar rules. Nowadays, it is considered part of the first
generation of MT research (‘direct approach’, see fig. 4 and section 2.3.1),
characterised by its lexical orientation and its technique based on word-for-word
substitution. Despite the lack of scientific value, this first demonstration boosted
funding of MT research not only in the United States, but also world-wide.
Unfortunately, the following years went on to demonstrate that the objective of early
MT research, i.e. the creation of systems that were able to produce high quality
translations, had not been achieved: it became obvious that there were fundamental
differences, even in closely related language pairs such as German and English, and
that the technique of lexical substitution did not meet expectations. In a 1959 report,
Bar-Hillel criticised the goal of FAHQT and suggested a shift towards HAMT (see
section 2.1.1), causing disillusionment, especially in funders. However, the major
setback for MT research was the famous ALPAC1 report which was published in
1966. It drew the conclusion that MT was – in comparison to Human Translation – too
slow, too inaccurate and far too expensive. According to this report, there was no
need for further funding in MT since “there [was] no immediate or predictable
prospect of useful Machine Translation” (Hutchins and Somers 1992: 7). In the wake
of Bar-Hillel’s work, the ALPAC was in favour of strenghtening research in the field of
machine aids for translators and in fundamental computational linguistics. Although
the ALPAC report was largely criticized as “narrow, biased and short-sighted”
(Hutchins and Somers 1992: 7) and its conclusion is known today to be “almost
entirely mistaken” (idem), it virtually put an end to MT research, especially in the
United States where funding was cancelled. The result was a geographical shift in MT
research from the United States to Canada and Western Europe. Almost a decade
would pass before the so-called renaissance of MT.
1 Automatic Language Processing Advisory Committee, formed by the US government
sponsors of MT in 1964
2. An introduction to Machine Translation
10
The second generation of MT originated in the late 1970s. These systems are
called ‘indirect’ and comprise two different approaches, ‘interlingua’ and ‘transfer’ (see
fig. 4 and section 2.3.2). The main difference between the first and the second MT
generation is that “the transfer is taking place no longer on just the lexical but the
syntactical level” (Schwarzl 2001: 18); thus it is carried out indirectly via intermediate
representation. Furthermore, the indirect techniques are still in use today, whereas
the purely direct approach has been discarded. In tandem with the interlingua
approach arose the idea of knowledge-based systems, strongly oriented towards the
field of Artificial Intelligence (AI)2. They aimed at involving ‘understanding’ and ‘real
world knowledge’ into the translation process. These developments led to the creation
of a variety of new MT systems. Examples of transfer-based systems include
METEO, the aforementioned sublanguage system, developed in 1976 in Montreal;
the early Systran system used by the Commission of the European Communities
(CEC) from 1976 onwards; the EUROTRA research project funded by the CEC,
which aimed at establishing a multilingual system for all community languages and
built on the work of two transfer-based projects, GETA (Grenoble, France) and SUSY
(Saarbrücken, Germany); METAL at the University of Austin, Texas, funded by th US
Air Force; the Mu transfer system at Kyoto University, Japan; and SPANAM and
ENGSPAN, developed by the Pan American Health Organization. Examples of
interlingua systems created as of the 1980s include knowledge-based interlingua
systems KANT and KANTOO, developed at Carnegie Mellon University;
Mikrokosmos, created at New Mexico State University; Pivot by NEC and HICATS by
Hitachi as well as non-knowledge-based interlingua systems DLT by BSO in Utrecht
and Rosetta by Philips Research Laboratories in Eindhoven.
In the 1990s, MT research entered its third generation (see fig. 4), which was
caracterised by corpus-based systems, including statistical- and example-based
approaches (see section 2.3.3). The best known statistical-based MT is Candide by
IBM. Example-based systems, on the other hand, were mostly experimented with in
Japan. Despite these new approaches, research continued into the older methods.
The rapid technological advance of the last decade of the 20th century,
including the development of the internet and the fact that individuals could now
afford personal computers (PCs), had an enormous impact on the use of CAT-tools.
From 2000 onwards, modern information technologies, in particular the internet, had
a strong influence on MT since the latter created a new medium for translation.
2 „AI is the field concerned with software and hardware systems that display intelligent
behaviour, where intelligence is judged subjectively by humans observing the system” (Trujillo 1999: 6).
2. An introduction to Machine Translation
11
Consequently, the market has seen the emergence of a variety of new MT systems
designed to translate Web pages, e-mails, chat-room messages, etc. Not only was
there renewed demand for translation, there was also a demand for different types of
MT systems, such as “stand-alone and networked personal computers for
professional translators,” “machine translation systems for professional translators or
organizations,” “online machine translation systems for home users and non-
translators” and “portable hand-held audio-visual translation devices for non-
translators” (Quah 2006: 65). Moreover, these systems are today often integrated into
translator’s workstations. Whereas Machine Translation had for a long time been
more research-centred, it has now gained importance in translation practice and
become a commercial product (Ramlow 2008: 36). An example of this change
towards professional use and commercialization is the abovementioned METAL
system, which today is sold under the product name Comprendium and directed at
private users and small- and medium-sized businesses (Schäfer 2002: 25).
As this historical overview has shown, MT research is into its third generation
and has seen the emergence of three main systems, that is, direct, indirect and
corpus-based approaches. According to Schäfer (2002: 24), modern MT research has
discarded the traditional rule-based systems in favour of corpus-based approaches,
while Quah (2006: 66) comes to the conclusion that current MT systems can be
regarded as “hybrid systems that combine the earlier rule-based approaches and the
subsequent corpus-based approaches”.
Regardless, it is important for the purpose of my study to understand how the
different MT architectures function, not least in order to place Systran, the MT system
used in the study, in a broader context. The following section will take a closer look at
the types of MT architecture.
2.3 MT architectures
As seen in the previous section, MT systems can be classified according to
their architectures, and consequently to the different generations of MT research (see
fig. 4). The following classification is based on Hutchins and Somers (1992: 69-76),
Arnold (2003: 119-123 and 136-141), Trujillo (1999: 5-6 and 203-211) and Quah
(2006: 66-85). Figure 5 gives a detailed overview of the different approaches:
2. An introduction to Machine Translation
12
The following subsection will describe the inner workings of MT technology, in
particular how the ‘classical’ indirect MT systems function.
2.3.1 Direct translation approach
As mentioned above, the direct architecture (also known as transformer
system) dates back to the first generation of MT research, though it is used to this
day – at least partly – by some MT systems (e.g. by Systran, see section 2.4). The
direct approach is the simplest one since it is based on word-for-word substitution.
The term ‘direct’ hints at a missing intermediate stage in the translation process: there
is usually a morphological analysis as well as a disambiguation phase, but the system
goes “directly from a source-language text to a target-language text essentially
without assigning any linguistic structure” (Arnold 2003: 122). Figure 6, taken from
Hutchins and Somers, illustrates this process. The missing analysis of syntax or
semantic relationships results in an “extensive string pattern matching” (Trujillo
1999: 5). After this simple bilingual dictionary look-up, the target text is slightly
rearranged by means of some local reordering rules to conform to the word order of
the target language (TL).
The direct approach obviously has considerable limitations. Hutchins and
Somers compare it with a person having a very cheap bilingual dictionary and very
restricted knowledge of TL grammar. Mistranslations resulting from the word-for-word
technique are frequent, as illustrated by the following example, also taken from
Hutchins and Somers: a direct MT system translated the Russian sentence My
trebuem mira, meaning We want peace, by We require world.
2.3.2 Indirect translation approaches
Due to the failure of the direct approach, MT research concentrated on the
development of more sophisticated systems. This led to the creation of so-called
Fig. 5: adapted from “Machine translation architectures” (Quah 2006: 68)
Fig. 6: “Direct MT system” (Hutchins and Somers 1992: 72)
2. An introduction to Machine Translation
13
indirect systems (also known as linguistic knowledge systems). The idea behind
these approaches is that “[h]igh quality MT requires linguistic knowledge of both the
source and the target languages as well as the differences between them” (Arnold et
al. 1994: 66). Indirect systems generate some kind of intermediate or abstract
representation of the source text, which is then used as basis for the generation of the
target text. The analysis of the source text and the generation of the target text
require “the application of morphological, syntactic and/or semantic rules” (Quah
2006: 70f.). Depending on the language (in)dependency of the representation of the
SL text’s meaning, we speak either of ‘interlingua’ or of ‘transfer’ systems.
2.3.2.1 The interlingua approach
Interlingua systems are characterised by an analysis resulting in a language-
independent representation from which the TL text is directly generated. There is no
need for a transfer step since the intermediate representation “includes all information
necessary for the generation of the target text without ‘looking back’ to the original
text” (Hutchins and Somers 1992: 73). Consequently, the interlingua representation
projects the meaning of the source text and serves at the same time as a basis for the
generation of the target text (or, according to the number of language pairs, for the
generation of several TL texts). The generation of the target text then takes place with
the help of TL dictionaries and grammar rules. The role of interlinguas is very aptly
summarised in Furlani (2009: 18):
“The goal is to distill the source text as much as possible to the very essence of all the meaning carried in each given sub-segment, potentially including an assigned semantic concept, deep semantic labels such as ‘agent’ or ‘indirect object’, markers for number, tense, and mode for verbs, and designations of register, to name just a few examples. This essence is then translated into the representation in the target language, from which a target sentence is generated […].”
The main advantage of the interlingua approach is – in contrast to the transfer
system – its high modularity, i.e. the fact that the single modules are independent and
do not affect each other. This means that the addition of a new language to the
system only requires the creation of two new modules, an analysis and a grammar
generation (see fig. 7), entailing less work and lower costs compared to the transfer
approach which, for its part, requires setting up a transfer module for each language
pair and translation direction (see 2.3.2.2 and fig. 8). The missing transfer step in the
interlingua approach also means that the system developers do not need to have
knowledge of both the source and the target language. The number of target
languages increases for each language addition in relation to the existing language
pairs. In figure 7, for example, the addition of a new analysis module would increase
2. An introduction to Machine Translation
14
the number of SLs from three to four, whereas the addition of a new generation
module would increase the number of TLs by one. The addition of a new language,
e.g. Italian, would thus require the creation of two new modules and increase the
number of language pairs from six to twelve (from English into French, German and
Italian; from French into English, German and Italian; from German into English,
French and Italian; from Italian into English, French and German). This system type is
therefore very attractive for multilingual systems.
2.3.2.2 The transfer approach
Transfer-based (or interface) systems are, on the other hand,
language-dependent since each language pair requires the use of three separate
dictionaries, i.e. a monolingual dictionary of the SL and of the TL as well as a bilingual
transfer dictionary (see fig. 8). The translation process consists of the following three
steps: analysis (or parsing), transfer and generation (or synthesis).
After the creation of an intermediate representation of the source text, it is
mapped to a corresponding representation of the TL text, thereby using contrastive
knowledge of the two languages. The TL representation is then mapped to a target
text.
The clear disadvantage of transfer-based MT, compared to interlingua
systems, is the increased difficulty of adding new languages given that both the two
modules for analysis and generation and new transfer modules for each previously
incorporated language have to be created. In the case of the two-language system
shown in figure 8, this means that the addition of a third language would require four
new transfer modules. In order to add a fourth language, six new transfer modules
would have to be created, and so on exponentially. However, the advantage of the
transfer-based approach is that language-dependent modules are easier to design
Fig. 7: “Interlingua model with six language pairs” (Hutchins and Somers 1992: 74)
Fig. 8: “Transfer model with two language pairs” (Hutchins and Somers 1992: 75)
2. An introduction to Machine Translation
15
than language-independent ones. All the more so as it is not even certain that the
creation of a truly language-independent interlingua is possible.
2.3.3 Corpus-based translation approaches
Corpus-based MT methods include statistic-based and example-based
approaches. These gained popularity in the early 1990s when MT research entered
its third generation (see section 2.2). In contrast to the classical approach to MT,
which involves linguistic and other rules, corpus-based systems can be classified as
‘anaological’ since they “do not apply linguistic rules to the analysis of texts or to the
selection of translation equivalents” (Quah 2006: 77). The obvious advantage of these
approaches is their independence from linguistic knowledge, but this does not mean
that they are the panacea for MT. The main problem of corpus-based MT is the fact
that it demands huge amounts of high-quality bilingual corpora, which are rare.
Unfortunately, corpus-based systems can only be as good as the corpora they build
upon.
2.3.3.1 The statistic-based approach
The statistic-based MT approach (SMT) relies on the idea of translating on
probability. Instead of taking into consideration linguistic rules, this method is based
on “finding the most probable translation of a sentence using data gathered from a
bilingual corpus” (Mukesh et al. 2010: 27). SMT systems are therefore able to learn
translating (Estrella 2008: 10). According to the Statistical Translation Model patented
by Koehn and Knight in 2009, the translation process in SMT may be modelled as the
following three separate parts:
“(1) a language model (LM) that assigns a probability P(e) to any target string of words, (2) a translation model (TM) that assigns a probability P (f│e) to any pair of target and source strings, and (3) a decoder that determines translations based on the assigned probabilities of the LM and TM.” (Koehn and Knight 2009: 7)
In other words, the objective of SMT is to extract general translation rules from
large aligned bilingual corpora consisting of SL sentences and their translations in
order to “compute statistics on the frequency with which source and target language
words co-occur in aligned segments” (Mukesh et al. 2010: 26). In practice, this means
that if the verb “consist” is very often followed by the preposition “of”, but almost never
by the preposition “in”, the probability that “consist of” is a common word string in the
English language is very high. The translation correspondences furnished by the
resulting models are therefore not absolute, but more or less probable. As Mukesh et
al. put it, every word of the TL text in the corpus is therefore a potential translation of
each word of the SL text (idem).
2. An introduction to Machine Translation
16
The great advantage of SMT is the fact that it is not designed for a specific
language pair and that it “can be used for translation between any two languages, as
long as a sufficient parallel corpus is available” (Mukesh et al. 2010: 27). The obvious
drawback of this approach is the rareness of high-quality bilingual corpora.
As mentioned above, the classic example of statistical MT is Candide by IBM,
which was based on the Canadian Hansard Corpus consisting of Canadian Paliament
proceedings. Other well-known examples are the free online translators Google
Translate and Bing Translator.
2.3.3.2 The example-based approach
Like the statistic-based approach, the example-based approach (EBMT) does
not consider any linguistic rules. The principal idea of EBMT is “to translate a
sentence, us[ing] previous translation examples of similar sentences, the assumption
being that many translations are simple modifications of previous translations” (Trujillo
1999: 203). This method has a strong similarity to the use of a translation memory
system, and EBMT does equally save time and foster consistency in terminology and
style. The difference between TM and EBMT is that “a fully fledged EBMT system
may retrieve more than one example, identify fragments which match parts of the
input sentence and combine these fragments into a TL expression” (Trujillo 1999:
203f.). In other words, if the database contains the sentences “Peter plays rugby” and
“Andrew likes football”, the system can deduce the correct translations for sentences
like “Peter plays football” or “Peter likes football” because it finds all the necessary
fragments and their translations in the corpus. The scope of EBMT is however rather
limited: even very large corpora do not cover every part of the segments to be
translated (Mukesh et al. 2010: 27).
2.4 Controlled languages
Technical aspects of MT aside, I turn now to the focus of this study, i.e.
controlled languages (CLs). As seen earlier, the HAMT process can integrate a pre-
editing and a post-editing phase (see section 2.1.2). The benefit of a pre-editing
phase rests on the fact that even though language is the main medium for
communication, it is often a primary source of misunderstandings. Syntactic and
semantic complexity can render language ambiguous, which can be a problem for
humans, but most certainly is one for computers (Huijsen 1998: 1). Whereas human
translators can often turn badly written texts into well written translations, MT systems
cannot and will always turn bad input into bad output (Arnold et al. 1994: 25). This is
where controlled languages come into play: they “address this problem by defining
guidelines for and restrictions on the language which is used to author texts” (Nyberg
2. An introduction to Machine Translation
17
et al. 2003: 245). These restrictions are explicitly defined by specifying constraints on
lexicon, grammar and style (idem) and aim at “reduc[ing] or eliminat[ing] the use of
ambiguous and complex sentence structures” (Nyberg et al. 2003: 247). Typical CL
rules would be “keep sentences short”, “prefer active to passive voice” or “always use
determiners”. When it comes to MT, texts should ideally be written “with MT in mind
from the beginning” (Bernth and Gdaniec 2001: 176) since rewriting complex
sentences according to controlled language rules is a very time-consuming and
therefore costly process.
The main difference between CLs and sublanguages
(see section 2.1.1) – literally a subset of the ‘whole’ language, i.e. the specialised
language of a given sub-domain (Somers 2003: 283) – is that “restrictions of
controlled language are imposed on the authors, [whereas] those of a sublanguage
occur naturally” (Somers 2003: 283). In other words, for a sublanguage, it is the
system which is designed to cope with the terminology and the typical sentence
structure of a certain subject area; whereas, in the case of a controlled language, it is
above all the text that has to be adapted to the system (Hutchins and Somers 1992:
152).
The notion of controlled language first arose in the 1930s and 1940s and can
be traced back to Charles Ogden’s “Basic English” consisting of 850 words and a set
of rules. However, this first attempt at CL was not suitable for any practical purpose
and has never been widely used (Nyberg et al. 2003: 250). It was revisited in the
1970s by English-language based international corporations, mostly in Northern
America. Their idea was to avoid translating every manual and document into
different languages by providing very clearly written English texts that could be read
by a target group with limited English skills. Although CLs were created in part to save
on human translation costs, their introduction came with two major additional benefits:
first, they clearly enhanced the readability and clarity of texts, even for native English
readers. Second, MT results proved to be clearly better for controlled source texts
(Arnold et al. 1994: 148). The first CL was Caterpillar Fundamental English (CFE) in
the 1970s, which gave birth to other CL approaches such as J.I. Case’s Clear and
Simple English (CASE) or Perkin’s Approved Clear English (PACE). To this day, a
broad variety of companies and organisations work with controlled languages, for
example, General Motors, Siemens and the French Aerospace Industry. One of the
best-known CLs is the human-oriented Simplified Technical English (STE), developed
by the European Association of Aerospace Industries (AECMA) which aims to
improve “the readability of maintenance documentation in the civilian aircraft industry”
2. An introduction to Machine Translation
18
(Nyberg et al. 2003: 251). It is today an international standard for aircraft-
maintenance documentation (idem).
There is therefore no single CL for each language, but many individual
approaches to CL which are rarely accessible to the public. Sharon O’Brien carried
out a study in order to find out whether different companies’ and organisations’ CL
rules generally overlapped. She came to the conclusion that the eight CL-rule sets
she analysed had only one rule in common, namely the instruction to keep sentences
short (O’Brien 2003). This lack of a ‘core rule set’ makes it particularly difficult for
organisations looking to introduce a CL “without reinventing the wheel” (O’Brien
2003: 105).
In order to comply with CL rules, the use of a CL checker can be very useful
and time-saving. Even though CL rules are generally rather easy to understand and
apply, it can be difficult for an author to decide how to rewrite a given sentence in
order to achieve conformity with the writing rules. That is why CL checkers that
provide automatic feedback are important tools for efficient authoring (Mitamura et al.
2003: 254). They help the author to “determin[e] whether a text conforms to a
particular CL” (Nyberg et al. 2003: 251) and verify “that all words are approved and
that the writing rules are obeyed” (idem). Furthermore, state-of-the-art CL checkers
such as Acrocheck often include a function of not displaying certain modifications of
the SL text: the MT system translates from the controlled SL text version, but the
controlled SL text for purpose of dissemination does not contain certain pedestrian
structures such as excessive use of articles or prepositions.3 Most CL checkers have
been developed in-house; companies such as Alcatel Telecom, Siemens Nixdorf or
Boeing have created their own programmes. There are also commercial CL checkers,
such as MAXit or the aforementioned Acrocheck (Nyberg et al. 2003: 254).
The following paragraphs will provide a brief overview of the benefits and
disadvantages of the use of a CL. According to Nyberg et al. (2003: 248f.), CLs entail
four main advantages. As mentioned above, they improve the readability and
comprehensibility of texts. Second, they imply uniformity of word choice, use of
terminology, sentence structure and style. Third, the risk of misunderstanding, and
hence of liability, is clearly reduced. Fourth, using a CL improves the consistency and
reusability of the source text, which has also a positive impact on the use of
translation memories. Nyberg et al. (idem) mention two disadvantages: the writing
task can be more time-consuming if the text has to conform to a certain rule set, and
“authors may also experience a reduction in the power of expression if words that
3 According to Pierrette Bouillon
2. An introduction to Machine Translation
19
express the meaning they want to convey are unapproved and no good alternatives
are provided” (Nyberg et al. 2003: 249). This could lead to growing dissatisfaction
among authors and translators. They should consequently be involved in the
language definition process, not least because their active involvement in this process
might help prevent potential resistance to the CL approach.
Further, the introduction and maintenance of a CL requires heavy investment.
However, these costs are normally outweighed by the long-term advantages “for
organizations which produce a high volume of documentation per year, and for whom
the gains in consistency, reusability, and translatability are highly significant” (Nyberg
et al. 2003: 249). The benefits of CLs become particularly apparent in MT: controlled
texts need little or even almost no post-editing at all, which speeds up the translation
and revision process. In a MT background, the financial as well as the time factors
become all the more justifiable the higher the number of target languages (Hutchins
and Somers 1992: 152).
A particular important issue in MT-related CL is the acknowledgement that the
input text quality can in no case be surpassed by output text quality. In other words, a
translation will at best have the same stylistic quality as the source text. If the
controlled source text is written in short and concise sentences with redundancy
(resulting from limited use of pronouns, avoiding coordination, etc.), then the target
text will have the same simple and sometimes cumbersome style. This is very
important to bear in mind since “re-introducing a ‘non-controlled style’” (Nyberg et al.
2003: 274) by carrying out excessive post-editing would undo the advantages of
working with a CL. This topic as well as translators’ often excessive expectations will
be discussed in the next section. In any case, it is evident that overly restrictive CL
rules may introduce usability and productivity problems. As Mitamura aptly states: it
is“essential to find a middle ground which is productive and acceptable for authors
and which promotes high-quality translation” (Mitamura 1999: 52).
2.5 Post-editing
As seen in previous sections, post-editing is often combined with CL writing
rules “in order to improve the translatability of technical texts and to speed up the
productivity of the post-editing process” (Allen 2003: 298). Following Allen’s definition,
post-editing (PE) consists of editing, modifying and/or correcting “pre-translated text
that has been processed by an MT system from a source language into (a) target
language(s)” (2003: 297). Post-edition can thus be regarded as the logical
counterpart of the revision of a human translation (Koby 2001: 4). Hutchins points out
that this is one of the main reasons why the ALPAC report (see section 2.2) is now
2. An introduction to Machine Translation
20
known to be mistaken: it simply denied the fact “that most HT, particularly when
produced in translation agencies, is also revised (‘post-edited’) before submission to
clients” (Hutchins 1986: 166). Since FAHQT is still far from being a realistic goal,
there is an obvious need for post-editing, which is today – contrary to the early years
of MT research – widely acknowledged (Koby 2001: 4; Kay, Gawron, and Norvig
1994: 39). As Sager puts it, MT systems do not produce a human language, but
rather an artificial one. It is thus the job of post-editors to convert artificial texts into
natural ones (1994: 275). Post-editing remains therefore “a standard feature of any
serious machine translation system that will be used for commercial purposes” (Koby
2001: 1).
When it comes to the differences between revising a HT and post-editing MT
output, we can state that the principal divergence lays in the focus. Whereas a reviser
“check[s] the translation against the original source text for inadvertent omissions,
misunderstandings, etc.” (Koby 2001: 7), a post-editor “[adjusts] the machine output
so that it reflects as accurately as possible the meaning of the original text”
(Vasconcellos 1987: 411). In PE, the emphasis is thus put “on an ongoing exercise of
adjusting relatively predictable difficulties, rather than on the discovery of any
inadvertent lapse or error. The passages that clearly require corrections, though
many of them are minor and local, are more frequent than in traditional revision”
(idem). In other words, a human translator will probably not constantly mistranslate
the same words or expressions over and over again but rather commit isolated errors,
whereas a MT system is likely to consistently repeat the same mistakes. The error
categories differ also to the extent that humans may misunderstand whole sentences
or passages of the source text, whereas rule-based MT systems make more local
mistakes, mainly at the lexical, grammatical and syntactic level. They are therefore
not likely to omit any part of the source text. Corpus-based MT systems differ to the
extent that they commit fewer local and obvious mistakes, but rather semantic errors
which do not necessarily catch one’s eye. They might even omit sentence parts or
whole sentences, which can entail serious shifts in the meaning. A thorough post-
editing carried out by a professional translator with knowledge of both the source and
the target language is consequently all the more important. Since machines cannot
think and they commit – in the case of rule-based architectures – rather ‘stupid’
mistakes which are obvious to humans, post-editors might get irritated when they
have to continuously correct the same errors. In any case, repeatedly occurring MT
errors should be addressed at the root, i.e. by revising dictionaries. However, Koby is
right when he states that “a point is often reached where a change in a structure in
one place results in false translations somewhere else” (2001: 7). I will come back to
2. An introduction to Machine Translation
21
this point in section 4.2.1. However, some of these minor, local mistakes can rather
easily and efficiently be corrected with the help of certain post-editing tools that
provide more sophisticated applications than standard word processors that offer
basic functions such as ‘search and replace’. Post-editing toolkits can, for instance,
dispose of concordance searchers, allow dictionaries and/or glossaries (user-made or
user-augmented) to be uploaded, offer the choice of displaying both source and
target texts on screen, etc. (Nunes Vieira and Specia 2011: 36). Furthermore, it can
generally be said that over time, post-editors get used to the “typical error patterns”
(Arnold et al. 1994: 33) of a MT system. Familiarity with these error patterns as well
as tuning the system thus plays an important role in reducing post-editing time, and
consequently in reducing cost.
There is also an important psychological difference between traditional
revision and PE. Sampson discusses the implementation of a MT system at Siemens
and points out the inhibitions senior translators might have when it comes to revising
translations of younger colleagues. He claims that Siemens’ productivity gains
following the substitution of junior translators by a MT system “were not due to high
quality machine translation output, but rather to a loss of human-human inhibitions”
(Sampson 1994, as cited in Koby 2001: 16). Revisers’ psychological constraints
obviously disappeared when faced with a MT output; it took them less time to revise
compared with human translations. Translators’ acceptance of post-editing tasks will
be discussed in section 2.8. Let it be noted, however, that, according to Lange, “[i]t is
even possible to roughly predict the amount of time a translator needs for post-editing
a text from their attitude towards MT” (1998: 36).
The psychological factor leads to the main problem when translators carry out
post-editing tasks: expectations of quality. Arnold et al. point out that “[m]ost post-
editors are also translators and are used to producing high quality texts. They are
likely to apply the same sort of output standards to their translations produced
automatically” (1994: 32). Translators are typically striving for perfection, which is not
the objective of MT post-editing. Due to the differences between MT and HT with
regard to the types of error and due to the ‘artificial’ outputs machines produce (see
above), revisers have to keep in mind not to expect “the same level of quality that he
or she can expect from a human translation” (Koby 2001: 7), as this would simply be
unrealistic. Instead, they must “be aware of the quality expectations of the end user,
which may be lower than those of the translator” (Koby 2001: 7). Many clients will
even accept lower translation prices for an inferior product quality. They may merely
want the translator to correct essential mistakes and give the green light for approval
(Koby 2001: 7f.). Therefore, translators have to “learn to edit without complete
2. An introduction to Machine Translation
22
rewriting” (Hutchins 1986: 331); instead of changing the style of texts, this means
ideally preserving the MT output “and edit[ing] only those points in the text that are
necessary to make the sentence function properly” (Koby 2001: 17). This balance
between minimal editing and maintaining certain quality standards can be hard to
find. Therefore, the “primary linguistic skill needed is revision skill – the ability to
decide what to change and how much to change” (idem). It is particularly important
not to rewrite the output in order to not undo the main idea behind MT, i.e. speeding
up the translation process and making it more efficient. This efficiency is especially
dependent on the post-editing process (as well on the pre-editing process), since the
actual Machine Translation process does not take a notable amount of time.4
Vasconcellos sums it up by saying that “[e]xtensive changes make the process too
costly and negate the advantage of MT. A safe motto is ‘When in doubt, don’t’”
(Vasconcellos 1987: 414).
Having underlined the importance of the end user’s need, i.e. the final purpose
of the translation, let us turn to the different levels of post-editing. These depend on
further factors such as the expectation of quality, the translation turn-around time and
the document’s life expectancy (Allen 2003: 301). All these points are linked to the
type of translation demand: it makes a huge difference for the level of PE if the final
translation will be used for assimilation (i.e. ‘gisting’) or for dissemination purposes
(i.e. for publishing; see section 2.7, page 26). Translations for assimilation purposes
may not be post-edited at all, or they may undergo a so-called ‘rapid post-editing’.
Rapid post-editing means performing “a strictly minimal amount of corrections on
documents that usually contain perishable information” (Allen 2003: 302) and that are
usually intended for internal use. Translations for dissemination purposes, on the
other hand, have to be post-edited, either with minimal or full post-editing. Full PE is
only recommended for controlled input texts, as the results using uncontrolled input
are not satisfying enough, entailing that full PE on uncontrolled texts will in most
cases take more time than the human translation process. As pointed out before,
minimal PE consists in making “the least amount of comments possible for producing
an understandable working document, rather than producing a high-quality document”
(Allen 2003: 305). In other words, “information accuracy prevails over stylistic
considerations” (Roturier 2004: 6). This can entail two problems, ‘over-correcting’ and
‘under-correcting’ (Allen 2003: 305). We have seen that most translators are
perfectionists and have difficulties when it comes to deciding whether changes are
necessary or not. Therefore, experience is a significant factor for post-editors. The
4 Naturally, the output quality also determines the PE effort, since “the worse the
output, the greater the post-edit effort” (Arnold et al. 1994: 32).
2. An introduction to Machine Translation
23
factors mentioned above (end user’s need, expectation of quality, translation turn-
around time, document’s life expectancy) have to be strictly kept in mind to apply the
adequate PE level. In order to tackle these issues and to “set appropriate
expectations about output quality” (1999: 49), Mitamura suggests a controlled target-
language definition. As seen in the previous chapter, “the translated document can be
expected to have at best the same stylistic quality as the source document”
(Mitamura 1999: 49). Hence Mitamura considers it important to have not only a CL
specification for the SL, but also for the TL. There are nonetheless no general PE
criteria for the different levels of PE; translation agencies and in-house translation
departments usually create their own guidelines. There is therefore the same problem
as with CL (see previous section), i.e. “the risk of re-inventing the same wheel” (Allen
2003: 298).
2.6 Systran
Turning to the MT system used for my study, the aim of this section is to
provide the reader with some general information about Systran, based on Hutchins
and Somers (1992: 175-189) and Schäfer (2002: 33-76).
Systran is certainly one of the oldest MT systems since it can be traced back
to the very origins of MT research. Its designer, Peter Thoma, was a principal
programmer of the research group at Georgetown University that had given the first
MT demonstration in 1954 (see section 2.2). During the 1960s, he and his research
team developed a Russian to English MT system that resulted in the Systran
prototype, the acronym of system translation. Systran entered into professional use in
the 1970s, when the US Air Force, NASA and Euratom started working with it. In
1976, Thoma and the European Communities Commission (CEC) signed a contract
foreseeing the adaptation of the English to French version to the CEC context as well
as the development of a French to English version. In 1979, English-Italian was
added as a second language pair, and further languages followed in the 1980s. To
this day, Systran is available in fifty-two language pairs and in ten language packs,
with combinations between Arabic, Chinese, Dutch, English, French, German, Greek,
Italian, Japanese, Korean, Polish, Russian, Spanish and Swedish. The system is
today used by a wide range of internationally known clients, such as NATO
headquarters in Brussels, the European Aeronautic Defence and Space Company
(EADS), the Nuclear Research Centre of Karlsruhe and the French Atomic Energy
and Alternative Energies Commission (CEA). Moreover, corporations such as
General Motors and Xerox corporation currently use Systran.
2. An introduction to Machine Translation
24
There is no general consensus in MT research about the classification of
Systran based on its architecture. Some classify it as a direct system, others identify it
as a transfer-based system that has developed from a direct system. As mentioned
above, the first Systran versions were strongly influenced by the direct approach.
There remain to this day some points suggesting more of a first-generation-system
than a modern transfer-based system, such as the fact that it is missing a coherent
linguistic model (Systran uses a rather pragmatic analysis that does not rely on a
methodical linguistic model), that sentence analysis is not as complete as in other
transfer systems and that Systran is highly based on its extensive dictionaries.
Nonetheless, Systran has undergone various modernisations towards a transfer-
based model, the most important being the enhancement of the system’s modularity.
In this regard, Systran now has a common English analysis lexicon for translations
from English into all other languages, as well as of a “common Romance language
analysis ‘trunk’ […] which can be implemented whenever a Romance language
(French, Italian, Spanish, Portuguese) is the source in a system” (Hutchins and
Somers 1992: 177). Furthermore, the language-pair-independent basic software and
the language-pair-dependent linguistic software are separate, and the main
translation process is divided into three steps characteristic of transfer-based
systems, i.e. a SL-dependent analysis, a language-pair-dependent transfer and a TL-
dependent synthesis. However, the main reason for the lack of consensus about
Systran’s classification is the fact that the translation process may in principle be
divided into these three steps, but they often overlap, which is usually not the case in
transfer-based systems.
The following section will provide a brief overview of how the Systran
translation process functions . According to Hutchins and Somers, Systran’s basic
system consists of two main types of programs. On the one hand, there are
language- independent system programs; these “control and utility programs” are
above all responsible for “dictionary look-up routines” (Hutchins and Somers 1992:
177). On the other hand, there are translation programs which can be divided into
various stages with separate program modules for activities such as analys is or
generation. These are for the most part language-pair-independent, i.e. “the analysis
module for a particular language is constant whatever the target language concerned
and the generation modules are likewise constant whatever the source” (idem).
However, Systran’s main component remains the large and complex bilingual
dictionaries that are the pillar of the translation process since they do not only contain
“lexical equivalences but also grammatical and semantic information used during
analysis and generation” (idem). The actual translation process can in turn be broken
2. An introduction to Machine Translation
25
down into four steps, namely into pre-processing or input, analysis, transfer and
synthesis, the first stages involving dictionary look-up and morphological analysis
(see above).5
There are various Systran versions available on the market, namely desktop,
server, mobile and online versions. These exist in different degrees of
professionalization, e.g. as free online translators like Altavista Babelfish powered by
Systran or as Premium versions designed for professionals in the language industry.
Systran Premium Translator (version 6.0.8.0.) was at my disposal for the study, which
is the precursor to the newest version 7. It comprises a translation project manager
(the translation editor), a dictionary manager and a toolbar (a very simplified version
of the translation project manager that is suitable for short texts or for testing). For
details on the Systran dictionaries also see section 4.1.1.
2.7 Suitable text types and applications for MT
It is very important to see MT “in a proper perspective [and] to be aware of its
strong points and shortcomings” (Arnold et al. 1994: 10). With this view, the following
section will focus on suitable text types for and applications of MT. As seen in section
2.1.1, texts written in a certain sublanguage, i.e. in a specialised language of a given
sub-domain, are easier to parse by a machine than others. This entails that functional
texts such as technical documentation, weather reports, patient reports etc. are
especially suited for MT, whereas the automatic translation of literary texts is likely to
provide a very poor result. But why is this so? First of all, it is important to note that
“domain specific language uses a limited syntax and a limited set of terminology”
(Schwarzl 2001: 33). Since the terms belonging to a specific domain are generally
unambiguous, they are much more likely to be correctly translated by a machine than
terms belonging to the general language. In specific domains, one term ideally
designs one concept and vice versa (Pulitano 2010) whereas in general language,
words are fundamentally ambiguous (Melby and Warner 1995: 55). This fundamental
ambiguity means that “it is impossible to predict the meanings of given words before
one has read the text” (Schwarzl 2001: 33). If this even applies to humans, then
computers are all the more unable to distinguish between the different meanings of
ambiguous words. This concept of “fluidity” (Melby and Warner 1995: 55) explains
“why general language cannot be computed” (Schwarzl 2001: 33). Specialised
language, on the other hand, is more likely to be well-translated by a machine since
informative texts of a specific domain are, or should be, “clearly written, objective,
5 For details concerning the translation process, see Hutchins and Somers (1992: 180 -
183) who provide a very full description.
2. An introduction to Machine Translation
26
factual and neutral” (Quah 2006: 89). Their goal is simply to convey information, not
to convey it in a certain way as poems do, for example. Informative texts suffer
therefore a “minimal loss of meaning during translation” (idem). To summarize with a
quote from Donna Durlani, “[t]he success of HAMT is heavily dependent upon the
intelligent judgment of the translator who designates a text for machine translation,
and poor output can often be traced back to a poor choice of input” (Furlani 2009: 27).
Turning to potential applications of MT, the following factors promote the use
of MT systems: an extensive volume of translation, a large number of target
languages, short turn-around times, digitized source texts and stylistic guidelines for
the production of the source text (e.g. controlled languages) (according to Austermühl
2001: 235). But MT output is not always useful in all situations. Hutchins and
Somers (1992: 158) note that “[t]here is no single mode of use for MT, but a whole
gamut of variations, with different qualities, different user profiles, […] and
consequently of course, different costs”. Somers (2003: 161) therefore distinguishes
three basic types of translation demand: dissemination, assimilation and interchange.
Dissemination designates translations of publishable quality like those expected
from human translators. MT for dissemination output is therefore subject to post-
edition in order to reach the required standard of quality (see section 2.5, page 22).
Assimilation, on the other hand, refers to lower quality translations of short-lived
documents intended for users wanting to understand the essential content of a text.
Its goal is “to produce large, inexpensive volumes of rough translation automatically at
a fast rate (Quah 2006: 90). In this regard, low quality MT which originally occurred as
a “by-product of systems designed originally for the dissemination application”
(Hutchins 2003: 162) has become an option for a certain readership. Those who are
familiar with a certain domain and manage to extract the main message out of a
qualitatively poor ‘raw’ MT output may decide upon this basis which texts or passages
they would like to have properly translated (Hutchins and Somers 1992: 157). This
type of use is also known as “gisting” (Hutchins 2003: 162) and has become more
and more popular since cheaper MT systems have become available for a larger
group of users. The third type of translation demand is interchange use, meaning
“translation between participants in one-to-one-communication” (Somers 2003: 162),
e.g. translations of e-mails, discussion groups or chat messages. This “on-the-spot
translation” (Hutchins 2001: 17) demands immediate translation, which human
translators cannot provide. Even though the interchange MT output may be very poor,
it is still suitable if it conveys the main content of the messages.
Now that the different text types and applications suitable for MT have been
assessed, the following section will discuss the impact of MT on translators.
2. An introduction to Machine Translation
27
2.8 Translator’s perception of MT
The following quote from John Hutchins (2001: 5) aptly describes how
translators receive MT:
“Ever since the idea of using computers to translate natural languages was
first proposed in the 1940s and since the first investigations were begun in the 1950s,
translators have watched developments either in scorn or in trepidation. Either they
have dismissed the very notion that anyone could even believe that translation could
be mechanized, or (at the other extreme) they have feared that their profession would
be taken over entirely by machines.”
This fear seems understandable in the context of early MT research since its
goal was to replace human translators with machines providing perfect output (see
sections 2.1 and 2.2). Even though it has become clear that FAHQT is today a
utopian objective, translators often remain hostile towards MT, fearing that computers
could take over their jobs, instead of seeing the advantages they could derive from
the use of MT. It is a fact that the vast majority of paid translations nowadays are of
technical documents, such as legal, economical or medical texts, which can often be
repetitive and tedious (Hutchins and Somers 1992: 1). Technical translation requires
above all accuracy and consistency, whereas creativity is less important or even
insignificant. In this light, translators should see MT as an aid when it comes to the
translation of rather “boring, repetitive translation jobs” (Arnold et al. 1994: 11), such
as minutes or administrative memoranda. MT systems can save time so that human
translators can “concentrate on more interesting tasks, where their specialist skills are
really needed” (idem). These more interesting tasks, e.g. translation of literature,
advertising texts or articles, require skills that computers cannot learn, i.e. cultural
knowledge, creativity and inventiveness. Literary translation will not provide a
satisfying result when it comes for instance to translating a football match report from
English into German: not only do idioms have to be adapted, the point of view from
which the text is written also has to be changed. If Werder Bremen has for example
beaten Manchester United, in an English report the title would be “ManU beaten by
Bremen” and not “Bremen beats ManU” as in a German text. Consequently, it can be
said that computers will never, or not in the near future, “be able to handle all types of
text reliably and accurately” (Arnold et al. 1994: 12). Rather, they produce literary
translations “deemed to be grammatically and translationally imperfect” (Arnold et al.
1994: 12), which proves that translators’ jobs will never be taken over by machines.
On the contrary, MT is creating new jobs: the market’s demand for pre- and post-
editors is steadily growing. Post-edition of imperfect raw output is and will always be
2. An introduction to Machine Translation
28
absolutely necessary in MT, and who, if not translators, are best skilled to carry out
this work? Koby points out that post-editors have to be translators, because only
translators can judge the accuracies of translation (2001: 12). Another point is the
considerable demand for translation for assimilation purposes which would, without
MT, simply not occur. MT for ‘gisting’ does therefore not threaten translators’ jobs.
As a result, translators’ perception of MT has changed in recent years.
However, the main reason for the change lies in the “[p]ressures to produce more and
more translation with less and less time [that] are driving professional translators to
seek solutions in all quarters including MT” (Hutchins 2003: 189). Translators have
therefore realised that MT is not necessarily a threat, but that it can indeed “be a boon
to them” (idem). Above all the fact that MT systems are getting cheaper and
sometimes available on the internet promotes its use by freelancers and small and
medium-sized enterprises. Brian Mossop (2006: 792) draws the conclusion that
“[t]echnologies are being adopted to serve business purposes, and an offshoot of this,
perhaps, is the change in the mental process of translation”.
2.9 Conclusion
As shown in this introductory chapter, the initial goal of early MT research,
namely the creation of computers being able to produce FAHQT, is not for the time
being a realistic objective. Instead, the MT types which are today in use are above all
HAMT and MAHT. MT research is in its third generation and has seen the emergence
of three different main systems, direct, indirect and corpus-based architectures.
Current systems are often hybrids, combining rule-based and corpus-based
approaches.
The subsequent discussion of CL revealed that the use of restrictive writing
rules generally entails more advantages than drawbacks, particularly in a MT context.
CLs are considered to disambiguate and simplify texts, which is a main factor in
acceptable output. The chapter also discussed the differences between revision of HT
and post-editing of MT output as well as at the distinctive demands for and levels of
PE, the latter depending above all on the final purpose of the translation.
There followed a presentation of Systran, the programme used for my study. It
is one of the oldest MT systems and can be classified somewhere between the direct
and the transfer-based approach. With regard to suitable text types for MT, it was
shown that informative texts are particularly appropriate for MT, whereas literary texts
are not a good choice for this purpose. There are three basic types of demand for
translation, that is, dissemination, assimilation and interchange, which each require a
different level of quality of MT output and consequently have an impact on the extent
2. An introduction to Machine Translation
29
of PE. The last section of this chapter demonstrated how the translators’ working
conditions have changed and to what extent they can benefit from MT. Instead of
being afraid of computers taking over their jobs, computers can take on the tedious
translation jobs in order to save translators time and money.
Now that the theoretical foundation of my study has been laid, the next
chapter will present UEFA regulations, the texts that formed the basis of the study,
and UEFA itself. This is key since the study requires a general overview of the
context in which these texts are written and translated.
3. Context: Union des Associations Européennes de Football (UEFA)
30
3. Context: Union des Associations Européennes de Football (UEFA)
This chapter aims to provide the reader with background information about
UEFA, its Language Services unit and, most importantly, UEFA regulations which
served as the textual basis of my study. The first two sections of this chapter will give
a brief overview of UEFA, including its history and internal structure (3.1), and of the
UEFA Language Services unit (3.2). The third section of this chapter will deal with
UEFA regulations, describing their creation and revision process (3.3.1) as well as the
rules for their structuring and drafting (3.3.2). Furthermore, it will present the current
regulations’ translation workflow and discuss the reasons supporting the introduction
of MT for this text type (3.3.3).
3.1 About UEFA
The following description of UEFA’s history and principal objectives is mainly
based on the organisation’s website (UEFA Website, see reference list). The Union of
European Football Associations was founded in Basel, Switzerland in 1954 and its
administrative headquarters are now located in Nyon in western Switzerland. When it
was founded, UEFA covered 25 national associations and had three full-time
employees. Today, the organisation comprises 53 national associations and employs
no fewer than 340 people with their number steadily increasing. In the field of football
administration, the positions at UEFA range from football specific functions to others
existing in most companies, such as finance, human resources, facility management
etc.
UEFA is the governing body of European football and is one of six continental
confederations recognised by FIFA, the world football’s governing body. According to
its statutes, UEFA’s main objective is to “promote football in Europe in a spirit of
peace, understanding and fair play, without any discrimination on account of politics,
gender, religion, race or any other reason” (UEFA Statutes 2010: Art. 2, para. b).
UEFA represents the European national football associations and acts on their behalf.
One of UEFA’s main tasks is organising and conducting international football
competitions and tournaments at the European level. It does not only organise elite
competitions such as the European Football Championship (EURO) or the UEFA
Champions League (UCL), but also tournaments for every type of football, for all
ages, at all levels and for men and women alike. UEFA supports youth, amateur,
grassroots and disability football, amongst others. In sum, according to its website,
3. Context: Union des Associations Européennes de Football (UEFA)
31
UEFA “has developed from a mainly administrative body into a dynamic organisation
that is in tune with the vast requirements of modern-day football”6.
For the purposes of this thesis, it is important to have a general overview of
UEFA’s internal structure because this paves the way for a better understanding of
the creation process of UEFA regulations to be discussed in section 3.3. Figure 9
below shows that UEFA is composed of ten divisions. Each division consists of one or
several units. The unit under consideration here, namely the Language Services unit,
is part of the Services division, along with Travel & Conferences, Human Resources,
Facility Management and Information and Communication Technology. A complete
overview of all divisions and units can be found in Annex A.
In addition to the ten UEFA divisions is UEFA Events SA, which is a wholly-
owned UEFA subsidiary. Its main objective is to carry out ALL work for the benefit of
UEFA and European football as a whole. UEFA Events SA aims at “supporting
football through business” (UEFA intranet, 07/03/2012) and focuses its efforts in the
following three key areas:
“1. Increase revenues from UEFA competitions (both club and national) 2. Deliver UEFA’s events in a cost effective manner to the highest possible
standard 3. Share knowledge with members of the football family (national associations,
clubs, etc.)” (UEFA intranet, 07/03/2012)
The company is a fully integrated marketing and events organisation that is
split into two strategic business divisions, namely marketing and operations. An
overview of the tasks of these two business divisions can be found in Annex B.
6 idem
Fig. 9: Overview of UEFA divisions
3. Context: Union des Associations Européennes de Football (UEFA)
32
3.2 Overview of the UEFA Language Services unit
This chapter is based on information obtained during an interview with Florian
Simmen, Head of UEFA Language Services.
The UEFA Langue Services (SLAN) unit consists of three sections, one for
each of the three official languages (English, French and German). The German and
the French section each employ three translators, whereas the English section
consists of two persons.
The overall mission of the Head of Language Services is to provide the UEFA
administration with high-quality, image-enhancing linguistic support. Since Simmen is
a native German translator and terminologist, he also supports the German
translation section with translation and editing tasks in case it is overwhelmed. In
terms of terminology tasks, he maintains and enriches the existing version of the
UEFA dictionary compiled by SLAN and various higher education institutions.
Furthermore, he is in charge of choosing the dictionaries used at SLAN and UEFA as
a whole.
In the area of human resources, Simmen develops technical expertise and
soft skills by undergoing adequate training on ever-more specific topics (e.g. financial
fairplay). Regarding translation tools, he is responsible for the flawless use of CAT-
tools in SLAN (mainly SDL Trados Studio 2011 and MultiTerm). He also implements
new processes, such as the new version of a translation management system using
the enterprise software SAP (“Systems, Applications and Products in Data
Processing”), in order to improve the internal workflow and facilitate the management
of translation for clients and translators. The translation tasks originating from all
UEFA units are divided between the translators themselves. There is no coordinator
and the management of translation is carried out by all translators, currently with the
help of Microsoft Outlook (as of August 2012). For outsourced translation tasks, SAP
is already in use; the general introduction of SAP is planned for the near future.
Besides the mentioned fields of activity, Simmen is responsible for managing
the needs for interpretation. He also introduced a new interpreting management
process and continues to implement this new procedure.
The texts translated by the UEFA Language Services unit are mostly English
source texts, therefore the English translators dedicate a lot of time to document
editing. The volume of translation into German is the highest, since few source texts
are written in that language. In 2011/12, the volume of English translation amounted
to 1 213 056 words, of French to 1 096 198 and of German to 1 312 006 words7. In
7 According to statistics received from Florian Simmen
3. Context: Union des Associations Européennes de Football (UEFA)
33
the event that SLAN is very busy and/or if the word count of a text is very high,
freelancers are assigned translations into the official UEFA languages. SLAN is also
responsible for translations into other languages, such as Spanish, Italian, Russian,
as well as Polish and Ukrainian due to EURO 2012, and other rare languages. For
translations into non-official UEFA languages, SLAN cooperates with freelance
translators as well as with translation agencies.
The translation quality of SLAN is very high since the ‘four-eyes principle’ is
strictly adhered to. However, SLAN is not responsible for articles on UEFA’s website.
The online articles are written by journalists and coordinated by the UEFA Online &
Publishing unit, which explains the quality difference between most online translations
and those carried out by UEFA in-house translators.
The texts taken on by SLAN are very diverse. According to the style guide
developed by the German section, these can be grouped into six main text types (see
UEFA-Stilführer Deutsch, pp.8-15):
1. Formal texts
Regulations Disciplinary decisions/judgments
Contracts/basic agreements/ general terms and conditions
2. Minutes
Proceedings reports Action lists Summaries of plenary sessions for the members of the
executive committee/Presentations of the commissions
presidents for the congress
3. Keyword-based texts
Agendas and programmes
4. Reports
5. Communication
Circular letters/letters/E-mails Meeting documents/annexes for the executive committee Press releases
Stadium announces
6. Journalistic texts
Publications News from the member associations, published in the monthly journal UEFA•direct
Technical reports/special inserts of UEFA•direct Books
Speeches/welcome addresses
3. Context: Union des Associations Européennes de Football (UEFA)
34
According to the style guide, UEFA regulations are classified as formal texts.
The following section will be devoted to them as they were the textual basis of my
study.
3.3 UEFA regulations
The UEFA administration can issue regulations for different purposes. For
example, regulations are necessary for competitions such as the EURO or the
Champions League as well as for technical issues (e.g. Stadium Infrastructure
Regulations). According to the Collins English Dictionary, a regulation is classified as
“a rule, principle, or condition that governs procedure or behaviour” (Collins 2012).
UEFA regulations govern competition systems, match dates, pitch conditions and
financial or technical provisions, among others. They thus serve as a legal basis for
participating teams, staff, referees and anyone else involved in UEFA matches. They
are written in rather formal legal language; the following extract from the Regulations
of the UEFA European Football Championship 2010-12 gives a foretaste of this text
type:
“3.04 Associations may not, and may not permit any third party to, develop, create, use, sell or distribute any promotional materials or merchandise bearing any representation of the trophy or any replica thereof (including, without limitation, trophy lift images) or use any such representation in a manner that could lead to an association between any third party and the trophy, replica trophy and/or the competition.” (p. 3)
The following three sections will describe the stakeholders participating in the
‘Regulations Core Process’ and the role SLAN plays in this process. They will also
detail the existing rules for structuring and drafting UEFA’s regulatory instruments,
which will be the basis for the controlled language rules for UEFA regulations to be
examined in section 4.1.2. The last section will discuss the reasons supporting the
introduction of a HAMT-based workflow for the translation of this text type.
3.3.1 Creation and revision of UEFA regulations
The Regulations of the UEFA European Football Championship are usually
revised every new cycle, i.e. every four years8. Along with the national competition
regulations, kit regulations are also revised every new cycle. The club competition
regulations (e.g. UEFA Champions League) are currently revised every year, though
UEFA is attempting to introduce a cyclical revision phase similar to the one in use for
national regulations, entailing revision on a triennial basis. Technical regulations are
revised as required, for example when new technical inventions are introduced (e.g.
8 The competition takes place every four years (two-years-rhythm with the
FIFA World Cup)
3. Context: Union des Associations Européennes de Football (UEFA)
35
goal-line technology, which has been approved by the International Football
Association Board in July 2012). The whole core process described below takes
approximately three months, and the translation and revision phase within SLAN, one
to two weeks per regulation.
The creation process of UEFA regulations is clearly defined in the internal
document UEFA legal framework: Regulations core process and legal principles
applicable to the creation of UEFA’s rules and regulations (2010), hereinafter UEFA
legal framework. This document aims to clarify the responsibilities of all parties
involved in the regulations core process by providing them with a “step-by-step
description of ‘who is doing what, by when and how’” (p.5). Furthermore, it covers
UEFA’s legal framework, composed of UEFA statutes, UEFA regulations and UEFA
directives, in order to define the basic legal principles underlying the creation and
revision of UEFA’s rules and regulations. However, this part is of minor interest for my
study and will therefore not be dealt with here.
According to figure 10 above, the regulations core process can be broken
down into six main steps. The regulations revision process begins with a so-called
“kick-off meeting” (phase A) organised by the Regulations Coordinator (RC). At this
meeting, the revision schedule and each party’s specific tasks are established by the
Sports Legal Services unit (SLS) and the Regulations Manager (RM) for the
regulations under consideration. They establish the first draft version of the
regulations, called ‘draft 0’, which is then proofread by an English translator from
SLAN. This draft is then revised by several stakeholders, such as experts on
insurance matters, doping, commercial rights or marketing matters, and lawyers from
different units. The fourth draft, ‘draft 3’, is then sent back to the English translators
Fig. 10: Overview of the phases of the regulations core process,
adapted from “UEFA legal framework ”, p. 9
3. Context: Union des Associations Européennes de Football (UEFA)
36
from SLAN for proofreading. After that, draft 3 is translated into German and French
(phase B). The translation process is coordinated by the RC, who organises an
editorial meeting in the final translation phase. At this meeting, the RM, SLAN and, if
necessary, the Regulation’s Editor (RE) and/or the SLS lawyer discuss and solve any
wording or translation problems discovered during the translation process. If
necessary, the RM will then update the English draft 3 on the basis of this editorial
meeting and provide the RM and SLAN with a final ‘English Committee draft’. The
German and French translations are then sent to the RC and to the RM by SLAN. In
phase C, the RM and his director prepare the ‘Reasoning for key amendments’ in
English, i.e. the amendments made to the previous edition. This document is then
translated into German and French by SLAN and hereinafter sent to the members of
the relevant UEFA committee(s) for review and feedback, along with the committee
draft. Based on committee feedback, these two documents undergo new changes,
which result in an EXCO9 request, the reasoning for key amendments and an EXCO
draft. These three documents have to be examined by the SLS lawyer for final legal
review. SLAN will then revise and translate them into German and French. In phase
D, all nine files (three for each language) have to be approved by the EXCO. In case
of EXCO changes, these are translated by SLAN and then again approved by the
SLS lawyer. The latter confirms the final wording of the regulations in all three
languages to the RC and the RM. Phases E and F consist in printing and publication,
archiving and dispatch to associations, which are of minor interest for the purposes of
this thesis.
3.3.2 Rules for the structuring and drafting of UEFA’s regulatory instruments
The rules for the structuring and drafting of UEFA’s regulatory instruments are
clearly described in section C of the UEFA legal framework (pp. 64-68). They are
divided into general rules, rules concerning specific features of regulatory language
and syntax and style rules.
The general principle pointed out in this chapter is that “[t]he language used is
the first guarantee of the effectiveness and enforceability of a rule” (p. 64). Rules
should thus be formulated or drafted “finding the simplest and most direct
correspondence between the thinking behind the rule and the way to express it
linguistically” (idem). Furthermore, the importance of the role that the translation
process plays in the improvement of UEFA’s regulations is stressed by clearly stating:
“the translation of a text into another language often highlights ambiguities or
inaccuracies in the original version” (idem).
9 UEFA-internal abbreviation for ‘Executive Committee ‘
3. Context: Union des Associations Européennes de Football (UEFA)
37
The proposed rules concerning specific features are foremost based on clear,
precise and direct writing. This includes internal and external coherence of
documents, concision and the consistent use of existing terminology. Particularly
important for my study, however, are the existing writing rules concerning syntax and
style, since I based the controlled language rules on them. Briefly, the syntax and
style rules mentioned in the UEFA legal framework are as follows:
1.) Choice of words: Prefer modern, idiomatic vocabulary Avoid ambiguous words Only use neologisms when absolutely necessary Only use foreign words when there is no correspondent modern term in the language in question
2.) Form of verbs: Verbs should be used in third person (singular or plural) Prefer the use of the present tense Depending on the context, the future tense may be more
appropriate in some cases 3.) Use of numbers:
Always write out cardinal and ordinal numbers, except in some specific cases (not important for my purposes)
4.) Length of sentences: Wherever possible, keep sentences short However, a sentence consisting of a main clause and a relative
clause is often more readable than several very short sentences. Properly constructed long sentences are not necessarily more
difficult to understand than a short sentence.
These rules are very helpful when it comes to simplifying texts for human
readers, in order to prepare texts for MT purposes, they are not detailed or specific
enough. This issue will be addressed in my study (section 4.1.).
3.3.3 Translation workflow of UEFA regulations
Section 3.3.1 presented SLAN’s role in the UEFA regulations core process
while the following paragraphs will examine the actual translation workflow for these
texts within the unit. In this context, I will set out the reasons why I deem UEFA
regulations suitable for MT.
UEFA regulations are currently translated in a machine-aided human
translation workflow (see section 2.1.3), meaning that the translators are supported
by a translation memory (TM), terminology databases and electronic dictionaries, as
well as by grammar and spell-checkers. UEFA introduced CAT-tools in 1999 (Trados
Workbench) and evolved with SDL Trados until today (current version: SDL Trados 10
Studio 2011). Two years ago, SLAN conducted a market analysis which confirmed
10
http://www.trados.com/en/Default.asp (28/05/2012)
3. Context: Union des Associations Européennes de Football (UEFA)
38
that SDL Trados was still the best choice. The programmes included in the market
analysis were Across, Déjà Vu, Transit, Word Fast and MultiCorpora, the latter
shortlisted with SDL Trados.
At SLAN there are one or several TMs used for every translation direction, for
example, English-German and French-English. The single segments are tagged by
information such as author, date, subject (e.g. women’s football, grassroots, coaching
convention, etc.), text type (e.g. regulations, circular letter, presentation, etc.), division
(e.g. Marketing, Finance, etc.) or competition (e.g. UEFA Champions League, EURO,
etc.). There are currently 174 741 segments in the English-German main TM and
20 052 in the English-German ‘regulations TM’, for a total of 194 793 segments. The
threshold used is generally 70% (standard setting in Trados), but it may vary from
translator to translator. As seen above, UEFA regulations are regularly revised,
depending on their content (competition, technical regulation, etc.). Since regulations
always resemble their precursor in a certain way – though it is generally hard to tell to
what extent they differ from one version to another – the use of a TM is highly efficient
for the translation of this text type.
Turning to the potential use of MT for UEFA regulations, the relative
resemblance of these texts is not only a main argument for their translation using TM
software, but also for their MT-suitability: the fewer changes have to be made from
one regulation version to another, the fewer changes will have to be made to the
dictionary of a MT system. Furthermore, regulations are functional texts, meaning that
function overrides form. They are from a certain sub-domain and contain generally
repetitive terminology that can easily be added to the MT dictionary (see section 2.7).
Further advantages of using MT for the translation of UEFA regulations would be the
time and cost factor. As mentioned above, the current MAHT phase including revision
takes one to two weeks per regulation. It would not only be advantageous for UEFA in
general to save costs by cutting the turnaround times for translation of regulations,
but also for the translators. Since the SLAN unit is generally working at full capacity,
translators would probably welcome not being stretched too thin. Furthermore,
regulations can be repetitive and tedious to translate; they mainly call for accuracy
and consistency, whereas creativity is fairly insignificant for this text type. By cutting
the turnaround time of the regulations’ translation process, translators would
potentially have more time for more interesting translation tasks requiring skills
computers cannot learn, such as inventiveness and cultural knowledge, i.e. when it
comes to translating strategies, programmes, articles etc. (see section 2.8).
Before comparing the time and cost-effectiveness of MT and TM in the context
of UEFA regulations, I had to assess the raw MT output in order to find out if it could
3. Context: Union des Associations Européennes de Football (UEFA)
39
compete with a human (machine-aided) translation. However, due to the limited
scope of my study, this paper will be limited to the first step, in other words whether
UEFA regulations are suitable for MT using Systran, a point addressed in chapter 4.
3.4 Conclusion
This chapter provided information about UEFA, its Language Services unit
and UEFA regulations, the text type used in my study. It briefly overviewed the
complex ‘Regulations core process’ and described how many stakeholders are
involved in the creation of this document type: a Regulations Coordinator, a
Regulations Manager, the Sports Legal Services unit, the Language Services unit,
experts on many different matters, such as insurance, doping or marketing, several
lawyers from different units, a Regulations Editor, and many more. This background
knowledge is very important in order to establish the controlled language rules,
bearing in mind that these will above all address experts in other fields than language.
The chapter also defined the rules for the structuring and drafting of UEFA’s
regulatory instruments, which enable the elaboration of rules geared towards MT
needs. The overview of the translation workflow for regulations and the points
supporting the introduction of MT provided the necessary prior knowledge for the
study.
4. The study
40
4. The study
Now that Machine Translation and UEFA regulations have been introduced,
this chapter will turn to the main goal underlying my study, that is, finding out whether
UEFA regulations are suitable for MT using Systran. The focus will be on evaluating
the MT output in order to find out if it can compete with a human (machine-aided)
translation, since the quality of the MT output is the crucial factor speaking for or
against the implementation of a HAMT-based workflow. The main point will be the
comparison of the MT output of controlled source-language (SL) text excerpts with
the MT output of uncontrolled SL text excerpts. As the title suggests, I also aim to
draw a more general conclusion concerning the impact of controlled language in MT.
To this end, I conducted a series of tests, the results of which will then be evaluated.
However, this study can only be considered as a pilot comparison because “limited
scope and resources prevent it from being statistically valid” (Furlani 2009: 49).
If the evaluations of the MT output yield positive results, a second step would
be comparing the turnaround time for UEFA regulations translated in a MAHT-based
and a HAMT-based workflow (see section 6.2). However, due to the limited scope of
my study, this paper will be limited to the first step, in other words whether UEFA
regulations are suitable for MT using Systran.
This chapter will begin with a brief description of the overall study setup in
order to provide a clear overview of the study’s structure. It will then explain in detail
the four steps that were necessary to obtain the results that will be evaluated in
chapter five. The last section of the chapter will outline the problems encountered
while working with Systran.
4.1 General study setup
For the sake of clarity, the following abbreviations will be used for the text
material:
S1 Main SL text material
S2 Second SL text material
C1 Controlled version of S1
C2 Controlled version of S2
C2/1 First part of C2, carried out by subject 1
C2/2 Second part of C2, carried out by subject 2
TC1 Raw MT output of C1
TC2 Raw MT output of C2
HTS1 Official human translation of S1
NCT1 Raw MT output of S1
NCT2 Raw MT output of S2
4. The study
41
PE1 Minimal PE of TC1
PE2 Full PE of TC1
PE3 Minimal PE of TC2
The first step of the study was to create an individual user dictionary and TM
with the Systran Dictionary Manager. This stage preceded the machine translation of
the text material and was necessary in order to tailor Systran to the text material. The
main SL text material (S1, see Annex C) was an excerpt from the UEFA Regulations
of the UEFA European Football Championship 2010-12 (21 311 words, or 910
sentences in all). I chose these regulations since the EURO – besides taking place in
summer 2012 – is arguably the most popular UEFA competition. The excerpt
consisted of chapters III, IV, VI and VIII (2 709 words, 196 sentences) which were
chosen randomly. As seen in chapter three, the revised versions of UEFA regulations
often resemble their precursors in a certain way which, combined with the fact that
these specialised texts are written in quite repetitive legal language, is one of the
reasons why they seem very suitable for MT.
Secondly, I established a controlled version of S1 (C1), enabling me to
subsequently create MT-oriented writing rules for UEFA regulations. Since it has
been proven that the use of CL in MT can greatly enhance translation quality (see
section 2.4), I could formulate the hypothesis that the use of CL would strongly
improve the MT output of the text material.
Then I had to double-check the writing rules in order to make sure they are
generally applicable. To do so, I chose a second SL text (S2), that is, an excerpt of
the UEFA regulations of the UEFA Champions League 2011/12 (see Annex D; 31
682 words, or 1397 sentences in all). S2 contained chapters III, IV, V and VII (2 562
words, 111 sentences) and was chosen based on S1: I chose the chapters dealing
with the same or a similar content and added other articles in order to obtain
approximately the same number of words, as well as to ensure objectivity. In order to
double-check the writing rules, S2 was controlled by two independent subjects. The
control of S2 (C2) would also reveal whether the Systran dictionary produced ‘silence’
or ‘noise’ (see section 4.1.3), in other words whether it had to be reviewed.
Then the raw MT outputs of C1 (TC1) and C2 (TC2) were post-edited. As
seen in section 2.5, translations for assimilation purposes can be minimally or fully
post-edited and full PE is only recommended for controlled input text. For the
purposes of the study, both PE types were carried out for TC1 in order to find out how
much more time PE2 would take compared to PE1. TC2 was only subjected to a
rapid post-edit and will not be analysed in this chapter. It only served as material for
4. The study
42
the quantitative evaluation (see section 5.1). PE1 and PE2 were used for both
evaluation parts of the study (see sections 5.1 and 5.2).
The discussion of the PE will be followed by a presentation of the different
types of software and translation problems encountered while working with Systran,
with the translation problems remaining after the control being discussed according to
monolingual and transfer ambiguities.
4.1.1 Working with Systran Dictionary Manager
Systran allows users to create user dictionaries (UDs), enabling them to
“modify the system in such a way as to be able to produce the best results possible”
(de Preux 2005: 3). For the purposes of this study, I tailored Systran to UEFA
regulations by creating a UEFA UD and a UEFA translation memory (TM).
Thanks to IntuitiveCoding Technology (see Senellart et al. 2001), dictionary
creation is very easy with Systran, even for users who are not at all familiar with MT
systems. This is a huge advantage of Systran over other MT systems that require
intensive training in order to work with them effectively. According to Systran,
“[p]roper dictionary coding is important, since it impacts the way terms are analyzed
during translation. IntuitiveCoding […] automatically enriches your dictionary entries
with information to improve translation quality” (Systran V6 User Guide: 112). In other
words, if a noun is added to the UD, Systran will guess the word category from the
ending and/or the presence of an article, and most probably automatically insert the
category ‘noun’. The green bar (see fig. 11) displays the reliability of the entry. In the
case of the noun ‘visiting club’, the automatic choice of the category ‘noun’ is
relatively reliable. Figure 12 shows an excerpt of Systran’s Coding Reference Table
for English. Once the user has internalised the coding references, they are very easily
applicable.
Fig. 11: Systran IntuitiveCoding Technology
Fig. 12: Excerpt of Systran’s Coding Reference Table for English
4. The study
43
The same technique applies to any other word category such as verbs,
adjectives, adverbs, etc. If the new entry has not been automatically allocated a
category, it can be chosen manually:
Single entries can be further modified by using Systran’s ExpertCoding (see
fig. 15). In addition to the word category, this function allows the user to modify
flexion, transitivity, hyphenation, semantic context, auxiliary verbs, grammatical case
and reflexivity. In this case, the verb “sich beziehen auf” has been added to the
dictionary using IntuitiveCoding (see fig. 14).
In addition to the UDs, Systran also enables the user to create TMs “to be
used for entries of any set phrases that are more effective as TM segments than as
dictionary entries” (Furlani 2009: 62). The TMs have a similar interface to the
dictionaries’ (see fig. 16 and 17). “During the translation process, SYSTRAN matches
TM entries with sentences in the source text, substituting these sentences where
appropriate“ (Systran V6 User Guide: 134). A particularity of the Systran TM is that
only exact 100% matches are recognized. This is why it is all the more important to
keep the source text(s) coherent and to try not to use synonyms for the sake of style.
Fig. 13: Systran word categories
Fig. 14: IntuitiveCoding of “sich beziehen auf
Fig. 16: Systran dictionary entry priority
Fig. 15: Systran ExpertCoding
4. The study
44
Systran comes with several built-in dictionaries, one general and 20
specialised in various subject areas (law, economics, medicine, etc.), and allows the
user to sort these according to a hierarchy which is then applied to the translation.
The user can thus influence the priority assigned to each dictionary and TM
depending on the text to be translated (see fig. 18 and 19). For a text dealing with
safety engineering, the user could for example tell Systran to look up the terms in the
following order: (1) TM Safety engineering, (2) UD Safety engineering, (3) automotive
engineering, (4) engineering, (5) electronic, (6) general. For this study, I created an
individual UEFA UD as well as a UEFA TM which was used according to the
hierarchy shown in fig. 19.
In addition to the various dictionaries, single dictionary entries can also be
prioritised using different values ranging from 1 (highest priority) to 9 (lowest priority).
In practice, this means that if the verb ‘to provide (prep: for)’ has different meanings in
German, each meaning can be assigned a different priority. In the given context, the
verb ‘provide’ has obtained the same priority 3 for all three relevant meaning in
German since they occur equally often in the text excerpts (‘vorlegen’, ‘sicherstellen’,
‘beinhalten’, ‘zur Verfügung stellen’; see fig. 16). In the Systran translation interface, it
is then possible to choose between the different meanings.
Fig. 18: Systran UD/TM priorites
Fig. 17: Systran UEFA TM
Fig. 19: Systran dictionary priorities
4. The study
45
The UD created for the study was assigned priority 4. This priority is
sufficiently high to prioritise the UD over Systran’s general dictionary. My dictionary
contains in total 523 entries that are made up of 224 nouns, 130 word sequences, 88
verbs, 51 adjectives, 14 adverbs, 9 proper names, 5 acronyms and 2 prepositions, of
which the priority level varies between 5 and 1. The TM created for the study contains
34 segments.
4.1.2 Control of S1 and creation of MT-oriented writing rules for UEFA
regulations
The goal of this chapter is to describe the method I employed in order to
obtain a controlled version of S1 and MT-oriented writing rules for UEFA regulations.
First of all, I translated S1 with Systran, using the previously created UD (see
fig. 20). I then simplified each English sentence in the Systran translation editor
following intuition, at the same time observing the impact of the modifications on the
German translation. I thereby established a methodology for the control of UEFA
regulations. For every sentence, I recorded the applied modifications in a table in
order to derive writing rules from it. The table comparing S1 and C1 and showing my
modifications can be found in Annex E.
The writing rules resulting from the control of S1 were regrouped into three
categories: General (10 rules), Lexis (7) and Grammar (13 rules), for a total of 30
rules. These were partly modified after a control of S2 was used to double-check the
rules. For the proper formulation of the rules, I consulted articles dealing with the
creation of controlled language rules by Bernth and Gdaniec (2001), Mitamura (1999)
and Mitamura et al. (2003). The MT-oriented writing rules for UEFA regulations can
be found in Annex F.
Figures 21 and 22 below depict the proportion of rules that were applied in
order to control S1. Figure 21 illustrates the number of applied rules per category,
whereas figure 22 shows the breakdown of the most employed individual rules.
Fig. 20: Systran Translation Project Manager Interface
4. The study
46
Fig. 21: Number of rules applied for the control of S1
Fig. 22: Most applied rules for the control of S111
The main problem stemming from the simplification and control of the text is
the partial loss of the characteristic English legal language. For example, in the
following sentence I applied one rule from each of the sections ‘General’ (‘Try to avoid
synonyms or different paraphrases expressing the same content in order to ensure
uniformity’), ‘Lexis’ (‘Avoid the ambiguous modal verb may’) and ‘Grammar’
(‘Transform passive into active constructions’):
(S1) Exceptionally, matches may be played in the territory of another UEFA member association, i f so decided by the UEFA administration and/or the disciplinary bodies, for reasons of safety or as a result of a disciplinary
measure.
(C1) Exceptionally, matches can be played on the territory of another UEFA member association, i f the UEFA administration and/or the disciplinary bodies decide so (for reasons of safety or as a result of a disciplinary
measure).
In order to obtain acceptable MT quality, it is often necessary to simplify
certain structures that are part of typical legal language. In this example, the
application of the ‘Lexis’ and ‘Grammar’ rules affected the legal language in which the
use of the modal verb ‘may’ and the excessive use of passive constructions is very
characteristic. For the study, simplifying complex sentence structures and ambiguous
words was inevitable, resulting in the loss of the language in which regulations are
typically written. However, I tried to maintain and add to the UD as many expressions
originating from the characteristic legal language as possible (example: […] remains
in UEFA’s keeping and ownership at all times).
11
The rule ‘Be as explicit and precise as possible’ has been revised and enlarged after the control of S2.
129 24
43 General Lexis Grammar
58 25
25
Be as explicit and precise as possible
Avoid long and complex sentences
Avoid synonyms or paraphrases to ensure uniformity
4. The study
47
With regard to the proportional use of the writing rules, it should be mentioned
that they were not always applied when controlling S1. Whenever possible, I added
typical noun clusters or legal expressions to the UD so that a control would not be
necessary, the reason being that I tried to impose as few rules as possible on the
authors. Considering the fact they came from different fields that have nothing or little
to do with language studies (see section 3.3.1), I felt it important in order to obtain the
highest possible acceptance of the rules.
It should also be pointed out that some rules contradict each other. For
example, the rules ‘Be as explicit and precise as possible’ and ‘Avoid redundancies’
from the section ‘General’ might seem inconsistent; however, they are both necessary
and subject to the author’s or post-editor’s judgement in every specific case. For the
following sentence, I applied the rule ‘Avoid redundancies’ and eliminated the
constituent ‘involved in a tie’. Firstly, it is clear for human readers that ‘the two teams’
designates the teams involved in a tie. Secondly, this constituent is not necessary for
the system’s correct parsing:
(S1) For matches played under the knockout system, if the two teams involved in a tie score the same number of goals over the two legs, the team which scores more away goals qualifies for the next stage.
(C1) For matches that are played under the knockout system, the following
criterion is applied: if both teams score the same number of aggregate goals, the team which scores more away goals qualifies for the next stage.
The following example, on the other hand, shows the application of the rule
‘Be as explicit and precise as possible’ in order to facilitate the parsing for the system
by repeating ‘cycle’ and ‘wins’:
(S1) Once a cycle of three successive wins or five in total has been completed,
the association concerned starts a new cycle from zero.
(C1) Once an association completes a cycle of three successive wins or a cycle of five wins in total, the association starts a new cycle from zero.
These examples highlight how difficult it is to establish a general rule that is
widely applicable. For every sentence, the author or post-editor has to decide
individually which rule he deems appropriate in that context.
4.1.3 Double-checking: Control of S2
As mentioned above, after creating the MT-oriented writing rules for UEFA
regulations, they were double-checked for two reasons: firstly, to ensure the rules
were generally applicable by other individuals (section 4.1.3.1), and secondly, to
determine whether the Systran UD produced ‘silence’ or ‘noise’ (section 4.1.3.2).
4. The study
48
4.1.3.1 General applicability of rules
Since the control of S1 and the creation of the writing rules were done
independently, I had to ensure they were not too subjective and biased. For this
purpose, I chose two independent subjects who carried out the control of S2 using the
writing rules.
Subject 1 was a qualified in-house translator at a solar technology company in
Germany. Her company had an internal technical language meaning that she was not
completely unfamiliar with CL. Furthermore, she had attended company-internal CL
seminars. Subject 2 was a student in her last year of a Master’s in Translation
(specialisation: Multilingual Translation Technology) and had some experience with
CL through MT classes taken during her degree and dealt with MT in her Master’s
thesis. Neither of the subjects was a native English speaker (subject 1: native
German speaker; subject 2: native Italian speaker). Subject 1 controlled the first part
of S2 (1 344 words) and subject 2 controlled the second part (1 159 words). Part one
contained more words since it comprised more enumerations that were not subject to
any control.
The independent control of S2 enabled me to subsequently analyse the
usefulness and applicability of the rules. Furthermore, it will be interesting to see to
what extent the control by two subjects with different backgrounds differed. Since
neither of the subjects was familiar with UEFA regulations, errors obviously occured
in their controlled versions owing to missing background knowledge. Therefore, it
would have been very interesting to see how UEFA employees, ideally stakeholders
in the UEFA regulations core process (e.g. experts on insurance matters, lawyers
from various units or employees of the Sports Legal Services unit) would have
controlled the second reference text. Such extensive double-checking by experts was
unfortunately not possible because of the limited scope of the study.
A few statistics are needed to set the stage for the analysis of the applicability
of the writing rules. The most interesting comparison is undoubtedly between the
number of interventions that each subject carried out and the number of interventions
I carried out for the control of S1. Keeping in mind that subjects 1 and 2 each
controlled approximately 1 200 words whereas I controlled approximately 2 500
words, I divided the number of interventions for the control of S1 by two. Figure 23
illustrates the difference between these three numbers. Subject 2 carried out
approximately the same number of interventions for the control of S2/2 as I did for the
control of half of S1. Subject 1, on the other hand, made less than half the number of
interventions for the control of S2/1. .
4. The study
49
The discrepancy observed between the number of interventions carried out for
the control of S1 and S2/2 on the one hand, and S1/1 on the other hand, can partly
be explained by the fact that S2/1 contained more enumerations than S2/2. Since
these enumerations only consisted of approximately 90 words, this number remains
too low to have a significant impact. I therefore compared the number of most applied
rules for C1, C2/1 and C2/2.
98
41
112 S1 divided by two
S2/1
S2/2
58 25
25
Be as explicit and precise as possible
Avoid long and complex sentences
Avoid synonyms or paraphrases to ensure uniformity
5
5
8
7
4
avoid determiner 'any'
avoid active form of 'apply'
use determiners
avoid possesive case and possessive pronouns avoid modal verb 'may'
Fig. 23: Comparison of the number of interventions carried out for the control of S1 and S2
Fig. 24: Most applied rules for the control of S1
Fig. 25: Subject 1 – Most applied rules for the control of S2/1
4. The study
50
These data clearly reveal a difference in the type of rules that were applied.
Whereas the most applied rules for S1 and S2/2 were from the categories ‘General’
and ‘Grammar’, subject 1 mostly used rules from the categories ‘Lexis’ and
‘Grammar’ and only three of the ‘General’ rules, which is clearly too little. The ‘Lexis’
and ‘Grammar’ rules are obviously more precise and easier to apply than the
‘General’ ones, even though it can be said that the ‘General’ rules become easier to
apply the more experience with CL the user has. This has obviously been the case
with the two subjects: subject 2, who had applied CL rules during MT seminars in her
Master’s degree applied my writing rules with ease, particularly the ‘General’ rules
which are especially important for simplifying the text structure. On the downside, she
logically committed more errors than subject 1 (see fig. 27). Subject 1 somewhat
restricted herself to using the precise, more isolated rules and did not really touch the
text and sentence structure out of a concern to not alter the style of the text, including
the characteristic legal language. It can be assumed that subject 1 would have
applied more rules to S1/1 with a broader CL background.
The errors committed by both subjects are partly attributable to the fact that
neither of them was a native English speaker. This issue would also apply to some of
the stakeholders in the UEFA regulations core process, though it can be assumed
that they possess better knowledge of the characteristic style of UEFA regulations
and would thus commit fewer stylistic and terminological errors. Other errors are, as
expected, due to a lack of background knowledge of UEFA and its regulations. The
following errors would certainly not have been committed by UEFA employees:
‘the UEFA’: the acronym UEFA never follows a determiner. This point was not included in the writing rules because it is obvious for UEFA employees;
Term errors (‘goal difference’, ‘group-winner’)
Semantic errors due to missing background knowledge
11 8
28
28
Avoid long and complex sentences Avoid omissions
avoid passive, prefer active
use determiners
4
12
Subject 1
Subject 2
Fig. 26: Subject 2 – Most applied rules for the control of S2/2
Fig. 27: Comparison of errors committed by subject 1 and subject 2
4. The study
51
In sum, during the control of S2, subject 1 used 13 rules, whereas subject 2
used 14. This is barely half of my 26 initially created rules and can be partly explained
by the rather short excerpts each subject worked on. It is clear that not all of the 26
rules can be used for every text excerpt, especially precise ‘Lexis’ rules such as
‘Avoid the term association when it does not designate a football association’.
Nonetheless, certain rules did require revision; in other words, the double-checking
phase served its main purpose of improving the rules. It enabled us to detect that the
rule we mainly employed for the control of S1 (‘Be as explicit and precise as
possible’) was not precise enough. Consequently, I created a second, similar rule
(‘Simplify sentence structures and expressions whenever possible’) in order to obtain
better results. After analysing the control of S2, I also had to add some rules because
the need for them became obvious during the double-checking phase and make
several minor amendments to the rules. The following table shows all additions and
amendments made to the rules; the underlined parts were amended:
Added rules Amended rules Explanation for amendment
3. Every segment has to be independent on its own:
Make sure that every segment can stand alone syntactically, i.e. avoid
splitting up sentences into parts that do not make sense without a preceding part:
This rule applies if
more than one club
from the same country…
is seeded for…
19. Avoid the use of ing-words, if possible: Exception: typical idioms like
including, without limitation or following, according to
Subject 2 systematically replaced ‘according to’ with ‘in compliance with’. This
intervention may not be wrong, but it was not necessary in the context.
16. Avoid redundant conjunctions such as ‘also’, ‘as well’, etc.
If necessary, prefer the use of conjunctions at the beginning of a sentence
(moreover, furthermore, additionally, etc.)
20. Use determiners (the, a) wherever possible Exception: Do not
necessarily use determiners in titles and subtitles
Subject 2 systematically used determiners wherever possible (as initially indicated
in my writing rules). In principle, this should be positive. Since I had made
the observation that Systran actually deals with titles even though they do not contain
determiners, I deemed the use of determiners in titles redundant because it
doubles the time spent in the post-editing process (elimination of determiners in
the translation for reasons of length and style).
4. The study
52
Added rules Amended rules Explanation for
amendment
17. Only use the combination ‘against each
other’ when it means ‘against one another’. In any other case, use
‘against every other’, e.g. against every other team in the group.
26. Do not omit relative pronouns, even though
this seems to be redundant for humans:
27. Repeat the preposition in conjoined constructions where
appropriate, even though this seems to be redundant for humans:
29. Do not reduce relative clauses, always
write the expanded form, even though this seems to be redundant for humans:
The control of subject 1 demonstrated that she was
not always aware of the fact that machines cannot parse certain structures due to their
inability to reason.
Table 1: Added and amended rules
Further tests that include UEFA employees consulting the revised rules would
have to be conducted in order to draw a more reliable conclusion on the applicability
of the MT-oriented writing rules for UEFA regulations. The test revealed very different
results depending on the background of the subject. Given that one of the subjects
satisfied my expectations, the rules can be considered generally applicable, as long
as the writer or post-editor has some practical experience with CL. This observation
entails that UEFA employees would probably need some thorough introduction to the
topic before drafting regulations according to CL rules.
4.1.3.2 Noise and silence issues
The second reason for double-checking was to ensure that the UD did not
generate noise or silence, and if it did, to revise the corresponding entries. Noise, on
the one hand, refers to dictionary entries that are inserted in inadequate places. This
often happens automatically after the creation of an entry. Take the example of
‘association’, which can refer to a football association (‘Verband’) or simply to a
mental association (‘Assoziation’), it is obvious that I either had to create two entries
or control the source text. In this case, I preferred the second option: according to the
writing rules, the noun ‘association’ should only be used when it designates a football
association, while in any other context, the unambiguous noun ‘link’ should be
preferred. Naturally, I tried to prevent noise problems by creating several dictionary
entries for ambiguous terms, but it is also obvious that all such issues cannot be
4. The study
53
divined beforehand12 (especially those regarding prepositions and word sequences
which might be automatically employed in the wrong places). Silence, on the other
hand, refers to the opposite problem, that is, when an existing entry is not
automatically inserted into the translation. The non-employment of certain entries is
usually due to priorities being set too low (especially for verbs) or to other elements
disrupting an entry (especially for word sequences).
Before going through the controlled MT output of S2 (TC2), I added the
missing terminology to the UD. After this step, I had a closer look at the translation in
order to revise the UD. In so doing, I detected one noise and one silence problem.
The noise problem I detected was the following: the preposition entry ‘in first
(context: place)’ = ‘auf dem ersten’ which we created in order to correctly translate the
sentence part ‘the teams in first, third, fourth and fifth place’ worked well for TC1. The
problem in TC2 was the insertion of this entry in a different context (‘the clubs that
were defeated in the first qualifying round’), in spite of my context specification.
Consequently, I tried to create a second preposition entry with the context ‘round’, but
this entry was not recognized. So I deleted both entries and created the word
sequence ‘in first, third, fourth and fifth place’ = ‘auf dem ersten dritten, vierten und
fünften Platz’. The reason for this noise problem was incorrect entry creation, which
taught me that preposition entries do not work with contexts.
The silence problem discovered during the double-checking phase was
caused by the verb entry ‘to be called to account’ = ‘zur Verantwortung gezogen
werden’. Paradoxically, the entry was correctly recognised in C1 (‘The host
association can be called to account and be disciplined for incidents of any kind.’), but
not in C2 (‘The home club […] can be called to account for incidents of any kind and
can be disciplined.‘). The different sentence structure obviously disturbed the
recognition of the entry in C2. The only solution I found, which is obviously not a
satisfactory one, was to paraphrase the sentence in C2.
A further observation made during this phase was the relative uselessness of
the TM in combination with CL rules applied by different users. This is due to the fact
that Systran TMs only recognize 100% matches: given the fact that every person
writes at least in a slightly different, if not in a very different way, these exact matches
almost never occur if different people are controlling the source texts.
In sum, the second part of the double-checking phase proved that the UD was
practically free of noise and silence problems. On the downside, I realised that the TM
did not serve its purpose because of the very high match threshold.
12
Except when working with a limited, standardised dictionary, as is the case with, for example, Simplified English (SE)
4. The study
54
4.1.4 Post-editing of TC1
This section will analyse the PE1 version (rapid post-edition) and the PE2
version (full post-edition) of TC1. Firstly, it will briefly present the PE guidelines I
adhered to. Secondly, it will provide some statistics for comparison purposes and,
amongst others, calculate the time needed for each PE type. The evaluation of the
quality of the two PE levels will be discussed in chapter five. The rapid post-editing of
TC2 (PE3) will not be discussed here as the exclusive goal of this PE analysis is to
compare the time exposure of rapid and full post-editing. PE3 will serve as material
for the quantitative evaluation (see section 5.1).
For the minimal post-editing of TC1, I generally adhered to a set of PE
guidelines set up by Wagner (1985: 7) and adapted by Roturier (2004: 7). Wagner
established her guidelines for the European Commission Translation Service, strongly
focusing on the time-saving factor as the main justification for rapid PE. Wagner’s
rules are of course not the most recent, but for the most part still as relevant as they
were in the 1980s. Some of them, e.g. “Do use a word processor if possible”, are
obviously outdated. Others, such as:
“Do save time by giving a ‘free’ translation or even a colloquial expression which would not normally be acceptable in Commission usage. This is justifiable because the main aim is to convey the information content of the
text” (Wagner 1985: 7)
are not appropriate in this context. Since UEFA regulations are not for
assimilation, but for dissemination purposes (see page 26), the final version must be
of publishable quality like that expected from human translators. The set of minimal
PE guidelines I adhered to is as follows:
1. Read the SL text first (paragraph by paragraph). It is dangerous to correct the raw translation without referring to the original.
2. Make sure that all information is accurately transferred. 3. Retain as much of the raw machine translation output as possible. Resist
the temptation to delete and rewrite too much. 4. Remember that all the words are probably present in the MT output
(possibly in the wrong order). 5. Rectify only what is grammatically wrong and what is lexically essential for
the understanding of the target text (words/phrases that are nonsensical, wrong or ambiguous).
6. There is no need to change words for the sake of elegant variation and originality, even if the style of the translation is repetitive or pedestrian: information accuracy prevails over stylistic considerations.
7. Don’t spend too much time over a problem. Put in a marker to get back to the problem later if necessary.
8. When in doubt, don’t.
For full PE, I adhered to the same set of guidelines, with modifications in
points 5 and 6:
4. The study
55
1. Read the SL text first (paragraph by paragraph). It is dangerous to correct the raw translation without referring to the original.
2. Make sure that all information is accurately transferred. 3. Retain as much of the raw machine translation output as possible. Resist
the temptation to delete and rewrite too much. 4. Remember that all the words are probably present in the MT output
(possibly in the wrong order). 5. Do only reword sentence parts for streamlining purposes when they are
very repetitive or pedestrian. 6. Only change words, sentence parts or whole sentences for stylistic
improvement if it is a small intervention having an important impact. 7. Don’t spend too much time over a problem. Put in a marker to get back to
the problem later if necessary.
8. When in doubt, don’t.
When comparing the number of interventions carried out for PE1 and PE2,
there is a noticeably higher number of interventions for PE2 than for PE1 (see fig. 28).
This difference is due to the less restrictive character of the guidelines for full PE. The
discrepancy between the number of intervention is, however, a lot smaller than I had
assumed beforehand. It would also have been interesting to compare these figures to
the number of interventions necessary for the PE of NCT1, but the uncontrolled raw
output was not post-edited for the following reasons: (1) the PE of poor raw MT output
from an uncontrolled SL text would cost an enormous amount of time, and (2) in a
professional context, a translation from scratch would in this case be much more
likely than spending a lot of time on post-editing MT output that does not meet a
certain minimum quality requirement.
It is also necessary to compare the number of interventions according to two
categories: (1) Interventions that eliminate isolated errors, such as grammatical
errors, punctuation errors or redundant words and (2) Interventions with an impact on
the whole sentence, such as changing the word order or rewording sentences or
phrases. For the sake of a better overview, figure 30 only depicts the 34 interventions
carried out in PE2, i.e. the second PE round after having finished PE1. It is clear from
these comparisons that for PE2, the relation between categories (1) and (2) is a lot
more balanced than for PE1. This is logical keeping in mind that I eliminated all
serious errors, which were often isolated, in PE1. PE2 also focused on style and
streamlining, meaning that the proportion of deeper interventions is bigger than in
PE1.
104 138
PE1
PE2
Fig. 28: Comparison of the number of interventions PE1 and PE2
4. The study
56
The following table underlines the breakdown of the interventions carried out
according to PE level and intervention category; the most common interventions for
PE1 concerned above all rather isolated errors, whereas the most common
interventions for PE2 concerned rather interventions with an impact on the whole
sentence:
PE level and intervention category
Intervention type
Number of interventions
1
Changes to word order 28
Grammatical errors 23
2
Rewording for streamlining purposes
14
Stylistic improvements 13
Table 2: Most common PE interventions for PE1 and PE2
As previously mentioned, PE2 strongly focused on style and streamlining and
therefore contained a bigger proportion of deep interventions than PE1. The time
spent on both PE versions was calculated by measuring the seconds spent on every
intervention type and then multiplying them by the number of interventions carried out
per type. Figure 31 depicts the time needed for each PE type: approximately 18
minutes on PE1 and 8 minutes on the interventions additionally carried out on PE2.
Figures 33 and 34 compare the minutes spent according to categories (1) and
(2) (see page 55). They reveal that the proportion of interventions carried out per
category is very similar for both PE levels.
38
66
Impact on whole sentence
Isolated errors
17 17 Impact on whole sentence
Isolated errors
18 26
PE1
PE2
Fig. 29: PE1 – interventions according to categories
Fig. 30: PE2 – interventions according to categories
Fig. 31: Comparison in minutes between PE1 and PE2
4. The study
57
The analysis of the post-editing of TC1 shows that the additional number of
interventions and the extra time do not differ to a great extent between the full and
the rapid post-editing. The discrepancy in the number of interventions was 34,
whereas the time difference amounted to approximately 8 minutes for the excerpt
consisting of 2 709 words. Given the rather low additional effort necessary for a full
PE of a translation originating from a controlled SL text, it will be all the more
interesting to analyse the results of the surveys on the MT output (raw output, PE1
and PE2). The qualitative evaluation of PE1 and PE2 carried out by the German-
speaking UEFA translators (see section 5.2.2.1) in particular will provide important
indications concerning the acceptance of both PE levels: did UEFA translators
accept the sentences from the rapid PE, or did they prefer the full PE version? Prior
to the evaluation of the results, the next chapter will give an overview of the
problems encountered using Systran.
4.2 Problems encountered
The following section will discuss the software issues as well as the translation
problems encountered while working with Systran Premium Translator (version
6.0.8.0.). The first part of the chapter will briefly present the main software problems I
had to deal with. The second part will discuss translation problems that occurred due
to monolingual ambiguities (including morphological problems, lexical ambiguities,
structural ambiguities and anaphora resolution) and to transfer ambiguities (including
lexical, structural and other differences).
4.2.1 Software problems
Throughout my work with Systran, I encountered several software problems .
First, there were a lot of bugs in the system that affected the translation editor and the
dictionary manager. When clicking on the command ‘translate’ in the translation
editor, the following error message showed up regularly:
10,3 7,7
Impact on whole sentence
Isolated errors
14,7 11,3
Impact on whole sentence
Isolated errors
Fig. 32: PE1 – Comparison in minutes
according to categories
Fig. 33: PE2 – Comparison in minutes
according to categories
4. The study
58
Fig. 34: Translation editor bug
Every time I saved the dictionary after its import into the dictionary manager,
the following message appeared:
Consequently, the dictionary had to be exported and saved under a different
name and then re-imported, which is quite time-consuming. Furthermore, I often
received the following error message when I wanted to delete an entry:
In these cases, I had to delete the relevant entries manually, i.e. individually
delete the content of every column.
Apart from these bugs, I noticed that the larger the UD, the slower the
software became. This is a very annoying and time-consuming side effect. The UD
contained only 523 entries and it is doubtful that a much larger dictionary would be
usable. Saving an entry sometimes took up to a minute, which is unacceptable in a
professional context, especially when considering that MT was invented in order to
cut costs by saving time.
I also encountered problems during the dictionary creation process. As
illustrated in the previous section, some complex, language-specific translation
problems could not be resolved via dictionary entries and therefore required post-
editing. This is due to the limitation of MT when it comes to idioms, completely
different sentence structures and perspectives in two languages, etc. Likewise, some
of Systran’s basic functions, such as deducing all verb forms from the infinitive form,
do not always operate as they should. For example, we added the verbal phrase ‘to
take into consideration’/’berücksichtigen’ to the dictionary since Systran automatically
Fig. 35: Dictionary bug (1)
Fig. 36: Dictionary bug (2)
4. The study
59
translated it as ‘in Erwägung ziehen’ which did not fit the context. The newly created
entry was not always recognised in subsequent iterations, which entailed the need for
further entries for the third person singular form ‘takes into consideration’ and the
passive form ‘to be taken into consideration’. The negated passive form ‘not taken
into consideration’ was not recognised at all due to the adverb ‘thereby’ which
probably perturbed the parsing of the sentence concerned. I then added it to the TM
since the sentence was likely to appear in other regulations. Curiously, the synonym
‘to take into account’ was present in the Systran General Dictionary and had been
correctly inflected every time it occurred in S1. The problem with ‘to take into
consideration’ must therefore have been caused by the creation of the entry.
A further problem was caused by the distinction between ‘each’/’jeweils’ as a
determiner and ‘each’/’jede/r’ as an adverb. The following examples illustrate the
translation of ‘each’ depending on its word category:
determiner Each team plays against every
other team in its group.
Jede Mannschaft spielt gegen jede
andere Mannschaft in ihrer Gruppe.
adverb The defeated semi-finalists each receive 40 bronze medals.
Die unterlegenen Halbfinalisten erhalten jeweils 40 Bronzemedaillen.
I created several dictionary entries in order to distinguish between the different
word categories of ‘each’ (and their different equivalents in German), but Systran did
not take into account more than one entry at a time. As a result, I had to add the word
sequence ‘each receive’/’erhalten jeweils’ to the dictionary in order to obtain the
correct translation. The translation ‘erhalten jeder’ would not have been grammatically
wrong, but stylistically very clumsy. I did not encounter any noise problems due to this
entry in the double-checking phase (see section 4.1.3.2), which does not mean that it
would not cause problems for the translation of other regulations. Therefore, more
thorough tests would have to be carried out in further studies.
The last software deficiency I wish to illustrate here is the issue, mentioned in
section 2.5, of resolving a problem that leads to further problems. In the official UEFA
version of the EURO regulations, the compound ‘host associations for the final
tournament’ is translated as ‘Endrunden-Ausrichter’. I therefore created the necessary
‘noun’ entry. Before the creation of this entry, the sentence had been translated
correctly, except for the official term. After the creation of the entry, the sentence was
not parsed correctly so the translation contains unnecessary, incorrect prepositions:
4. The study
60
(C1) The teams of the host associations for the final tournament,
Poland and Ukraine, qualify automatically for the final tournament.
TC1 before creation of dictionary entry ‘Endrunden-
Ausrichter’ (correct translation, incorrect term)
Die Mannschaften der Ausrichterverbände für die Endrunde, Polen und die Ukraine, qualifizieren sich
automatisch für die Endrunde.
TC1 after creation of dictionary entry ‘Endrunden-
Ausrichter’ (incorrect translation)
Die Mannschaften von Endrunden-Ausrichter, von Polen und von die Ukraine, qualifizieren sich automatisch für die
Endrunde.
PE2 (correct translation, correct term)
Die Mannschaften der Endrunden-Ausrichter, Polen und die Ukraine, sind automatisch für die Endrunde qualifiziert.
In short, Systran Premium Translator (version 6.0.8.0.) does not work
flawlessly. Some basic functions need to be revised, and above all the slowness of
the dictionary and the constantly recurring bugs have to be remedied in order to be
able to fully recommend the software.
4.2.2 Translation problems remaining despite control
The aim of this chapter is to discuss the translation problems which could not
be solved by the control of S1. As mentioned before, controlled languages come into
play in order to “reduce or eliminate the use of ambiguous and complex sentence
structures” (Nyberg et al. 2003: 247). Accordingly, there should not, in theory, be any
translation problems in the MT output of a controlled SL text. However, not all
possible translation problems can be solved by making use of controlled language
rules, unless they are very limited like the previously mentioned Simplified English.
The following analysis of the persistent translation problems is mainly based on the
structure of chapters five and six of Hutchins’ and Somers’ “An introduction to
Machine Translation” (1992: 81-106), which are dedicated to analysis and translation
problems in MT. It is therefore split up into two main subsections, the first dealing with
translation problems due to monolingual ambiguities (including morphology problems,
lexical ambiguities, structural ambiguities and anaphora resolution) and the second
with translation problems due to transfer ambiguities (including lexical, structural and
other differences).
4.2.2.1 Monolingual ambiguities
Translation problems due to monolingual ambiguities arise during the SL text
analysis and can be divided into morphology problems, lexical ambiguities, structural
ambiguities and anaphora resolution.
4. The study
61
Morphology problems
In MT, morphological analysis can be regarded as “a means of helping to
simplify the more difficult problems of lexical, syntactic and semantic analysis and
generation” (Hutchins/Somers 1992: 82). A system carrying out morphological
analyses reduces dictionary size, the time spent on its creation and the computer time
in assessing it (idem). For example, morphological analysis could mean limiting verb
or adjective entries to their ‘root forms’ (e.g. ‘play’ for ‘playing’, ‘plays’, ‘play’, etc.;
‘angesehen’ for ‘angesehener’, ‘angesehenes’ and ‘angesehene’). The main
advantage of this method is the recognition of unknown words, i.e. words the
dictionary does not contain. When Systran has to deal with unknown words, the
system carries out a morphological analysis in order to determine the word category
of the lexeme. This information can be very useful for further syntactic analys is
(Schäfer 2002: 103f.). The following example from TC1 (without using my UEFA UD
and TM) shows how Systran deals with the unknown English word “semi-finalist” in
German. Even though the system does not recognise the word, it deduces that it is a
noun since many nouns in German have the ending -ist (‘Polizist’, ‘Publizist’, etc.).
According to German convention, nouns are also written with a capital letter. It is
particularly interesting to note the way Systran deals with the compound: the part
which is present in the dictionary (‘plaque’) is translated and combined with the
unknown part. In this case, even though the translation is not perfect (‘Plakette für
Halbfinalisten’), it does make sense and can be understood by a German reader.
In most cases, unknown compounds entail bigger problems. Due to the
morphological analysis carried out by Systran, the system does not necessarily treat
compounds as unknown words. As long as the words constituting a compound are
present in the system’s dictionary (or dictionaries), they are translated because their
“meaning and correct translation can often be derived from its component parts”
(Hutchins/Somers 1992: 84). The following example illustrates two significant
mistranslations resulting from the direct translation of the unknown compounds
‘match venue’ and ‘host association’:
(C1) The match venues are fixed by the host associations.
(TC1 without UEFA UD) Die Gleichschauplätze werden durch die Wirtsverbindungen
geregelt.
(TC1 with UEFA UD) Die Spielorte werden von den Ausrichterverbänden festgesetzt.
4. The study
62
The compound ‘match venues’ was translated as ‘same showplace’, whereas
‘host association’ was translated as ‘host link’, using the medical term ‘host’ - ‘Wirt’ as
in ‘host cell’ and not the more general one which is referred to in this sentence
(‘Gast’). The problem of lexical ambiguity will be further discussed below. In the case
of the compounds ‘match venues’ and ‘host associations’, it would have been better
to guess the unknown compounds plus mark them in red, as seen above for ‘semi-
finalist plaque’. This way, the user can identify unknown compounds more easily, add
them to the dictionary and consequently avoid the risk of overlooking a mistranslation
due to time pressure.
The following example shows how important the system’s capability to deal
with unknown words by deducing their category is for the analysis of the whole
sentence, particularly in longer and more complex sentences:
In this sentence, Systran does not recognise the adverb ‘no more’ and does
not translate it. Thanks to the morphological analysis, the system nonetheless detects
that it is dealing with an adverb and identifies its correct position in the German
sentence. Without this function of guessing the categories of unknown words, the
sentence structure would in this case probably not have been correctly translated in
the TL language, possibly leading to an incomprehensible translation.
Lexical ambiguities
Intralinguistic lexical ambiguities include all cases where “one word can be
interpreted in more than one way” (Hutchins/Somers 1992: 85) and can be divided
into category ambiguities, homography and polysemy.
Category ambiguities designate “a given word [which] may be assigned to
more than one grammatical or syntactic category […] according to the context”
(idem). TC1 translated with my UD and TM contained very few cases of translation
errors due to category ambiguities, and these cases could be easily resolved thanks
to Systran’s ‘ambiguities in the source text’ function which allows the user to indicate
the correct category. This was the case in the following example where the
ambiguous word ‘play’ was parsed as a compound (‘group play’/’Gruppenspiel’).
Once I clicked on ‘ambiguities in the source text: verb’, the sentence was correctly
parsed and translated. In this sentence, the verb ‘play’ was translated as ‘bestreiten’
4. The study
63
(v) and not as the standard equivalent ‘spielen’ (v) since it has several equivalents in
German according to context. I will come back to this point in a later section on
transfer ambiguities (see page 65). In this case, due to the context indication in the
dictionary entry, the correct equivalent ‘bestreiten’ was chosen by the system.
(C1) The winners and runners-up in each group play the quarter-finals over one match.
(TC1 before disambiguation)
Die Sieger und die Zweitplatzierten in jedem Gruppenspiel-das Viertelfinale in einem Spiel.
(TC1 after
disambiguation)
Die Sieger und die Zweitplatzierten in jeder Gruppe bestreiten das
Viertelfinale in einem Spiel.
Logically, the non-controlled version of S1 contained far more cases of
category ambiguity than the controlled one. However, not all of these ambiguities can
be solved since there will in most controlled texts13 remain sentences containing
strings of words that a MT system can parse as compounds, as in the example
above.
Other forms of intralinguistic ambiguity are homography, homonymy and
polysemy. They will be discussed together because for the purposes of this study, it
does not matter if two words simply have the same spelling and pronunciation but
different meanings (homonymy, e.g. the verb ‘can’: ‘be able’/‘put sth in a container’);
the same spelling but different pronunciation and meanings (homography, e.g. ‘bow’:
’weapon’/’form of courtesy’); or the same etymology (polysemy, e.g. ‘crane’:
’bird’/’lifting device’) (Thoden 2010: 7f.; Hutchins/Somers 1992: 86; Oxford 2010).
Such cases of lexical ambiguity can often be resolved by assigning semantic features
to the entries, such as ‘human’, ‘non human’ or ‘country’. I simply added the correct
equivalents to the dictionary and indicated them as standard translation, as in the
aforementioned S1 example of ‘association’: ’football organisation’/’link’. A further
example, from S2, is ‘leg’: ’body part’/’match’/’joint [gastr.]’ which, before the correct
translation ‘Spiel’ was added to the dictionary, Systran had translated as the context-
inappropriate gastronomic ‘Keule’.
Structural ambiguities
In contrast to lexical ambiguities, structural ambiguities involve “problems with
the syntactic structures and representations of sentences. Ambiguity arises when
there is more than one way of analysing the underlying structure of a sentence
according to the grammar used in the system” (Hutchins/Somers 1992: 88). As most
MT systems, Systran is “restricted to the sequential analysis of single sentences”
(idem). There are generally two types of ambiguity, ‘real’ and ‘accidental’: real
13
Unless a very restrictive CL is used, such as Simplified English.
4. The study
64
ambiguities are also problematic for humans since they allow space for various
interpretations, whereas accidental ambiguities are – at least at first glance – not
ambiguous for humans but they are for MT systems (Schäfer 2002: 100).
The following sentence from TC1 reveals a real structural ambiguity due to the
unclear relation of the preposition ‘of’:
(S1) 3.04 Associations must not develop, create, use, sell or distribute any
promotional materials or any merchandise that bear a representation of the trophy or a replica of the trophy […].
Possible interpretations
[…] that bear a representation of the trophy or a representation of a replica of the trophy […].
[…] that bear a representation of the trophy or a replica of the trophy […].
(TC1) 3.04 Verbände dürfen keine Werbematerialien oder keine Werbeartikel entwickeln, herstellen, verwenden, verkaufen oder verteilen, die eine
Darstellung des Pokals oder der Nachbildung des Pokals enthalten […]
In this sentence, it is not absolutely clear what is meant by ‘that bear a
representation of the trophy or a replica of the trophy’. A ‘representation of a replica of
the trophy’ could refer to a picture of a replica trophy, whereas a ‘replica of the trophy’
could for instance be a plastic trophy contained in children menus in fast food
restaurants. Systran’s raw translation is correct in this case. I consulted the official
German and French translations to ensure that a ‘representation of a replica of the
trophy’ is the correct interpretation.
Accidental structural ambiguities occur, as mentioned above,
“[…] due to an accidental combination of words having category ambiguities, due to alternative grammatical uses for syntactic constituents, or due to different possible combinations of syntactic constituents. The types of structural ambiguities that occur differ from language to language, and, importantly, from grammar to grammar” (Hutchins/Somers 1992: 89)
A typical example of accidental ambiguity would be the following:
“Gas pump prices rose last time oil stocks fell.” (from (Hutchins/Somers 1992: 90).
The verbs in this sentence are ‘rise’ and ‘fall’: even human readers do
probably firstly struggle with the sentence structure. A MT system has bigger
problems with this sentence. Interestingly, Systran parses the sentence structure
correctly, even though the translation is inacceptable:
GasAbgabepreise stiegen Ölaktien des letzten Males fielen.
Except for category ambiguities, I did not encounter any accidental
ambiguities in the text material, not even in the translation of the uncontrolled source
text. The best method for resolving structural ambiguities would probably be the use
4. The study
65
of an interactive system that consults the user before translating concerned
sentences.
Anaphora resolution
Anaphora refers to “an oblique reference being made to an entity mentioned
explicitly elsewhere in a text” (Hutchins/Somers 1992: 95). These are in most cases
pronouns such as ‘he’, ‘they’, ‘it’, demonstratives like ‘this’, ‘that’ or ‘these’ and
phrases like ‘the latter’. In order to correctly translate an anaphora, the system has to
detect its antecedent, i.e. “the identification of the earlier noun to which [it] refer[s]”
(idem). This is especially important when translating into a language like German
where – in contrast to English – the gender of pronouns is marked. The correct
translation of an anaphora is often a huge problem for MT systems since they
proceed sentence by sentence, which means that information above sentence level is
generally not recorded or processed. This is why I established writing rules 24 (‘Avoid
possessive case (´s) and possessive pronouns, if possible’) and 25 (‘Minimize the use
of personal pronouns’). It is however not possible to completely avoid anaphora,
meaning that my controlled texts still contained problems due to this phenomenon
(even though they can be greatly reduced by strictly adhering to the writing rules). In
the following sentence, the demonstrative ‘that’ poses a problem for Systran:
(C1) Associations must not use such a representation in a manner that could lead to a
link between […].
(TC1) Verbände dürfen solch eine Darstellung nicht in einem Kontext verwenden, die zu eine Assoziation zwischen […]
Systran has obviously not correctly parsed the demonstrative’s antecedent,
which is ‘manner’/’Kontext’ and instead related it to the noun
‘representation’/’Darstellung’. ‘Kontext’ is masculine, whereas ‘Darstellung’ is feminine
in German. Systran translated ‘that’ as ‘die’; the correct translation would have been
‘der’, so the sentence had to be post-edited.
4.2.2.2 Transfer ambiguities
In contrast to translation problems due to monolingual ambiguities, those due
to transfer ambiguities “arise from specific differences between the language pairs
involved” (Hutchins/Somers 1992: 98) and are therefore considered as contrastive
aspects. In other words, they emerge when
“a single source language word can potentially be translated by a number of different target language words or expressions, not because the source language word itself is ambiguous but because it is ‘ambiguous’ from the perspective of another language” (idem: 99).
4. The study
66
Transfer ambiguities include lexical and structural differences as well as
morphological and other problems.
Lexical differences
Lexical differences are mostly based on stylistic, grammatical and conceptual
translational ambiguities. Stylistic translational ambiguities come up “when the
choice of target language lexical equivalent depends on differences of register or text-
type” (Hutchins/Somers 1992: 99). Grammatical translational ambiguities arise
“when there is a lexical choice in the target language which is conditioned by
grammatical context” (Hutchins/Somers 1992: 100), e.g. the distinction in German
and French for the English ‘know’ (‘kennen’/’connaître’; ’wissen’/‘savoir’). In the text
material, this is the case for the English verb ‘play’, which has to be translated
differently depending on the context:
(C1-1) In principle, matches must be played in a stadium on the territory of the
respective host association.
(TC1-1) Prinzipiell, müssen Spiele in einem Stadion auf dem Gebiet des jeweiligen Ausrichterverbands ausgetragen werden.
(C1-2) The seeded teams play the return match at home.
(TC1-2) Die gesetzten Mannschaften bestreiten das Rückspiel zu Hause.
(C1-3) Each team plays against every other team in its group according to a league system […].
(TC1-3) Jede Mannschaft spielt gegen jede andere Mannschaft in seiner Gruppe nach
dem Meisterschaftsmodus […]
The German equivalents ‚spielen‘, ‘bestreiten‘ and ‘austragen‘ do all designate
the same action (‘to play’14
), but they cannot be employed in the same contexts (‘ein
Spiel wird ausgetragen’; ‘ein Spiel bestreiten‘; ‘gegen eine Mannschaft spielen‘; even
a fourth expression, ‘gegen eine Mannschaft antreten‘, would be possible). This
phenomenon could partly be classified as stylistic translational ambiguity (for the
nuance between ‘ein Spiel spielen’/’austragen’/’bestreiten’), but not for example (3)
where *’gegen eine Mannschaft austragen’/’bestreiten’ would definitely be a
grammatical error: in combination with the preposition ‘gegen’/’against’, only ‘spielen’
is correct in German. This is why I classed this translational ambiguity as
grammatical. In this case, it can quite easily be resolved by creating several entries
for the verb ‘play’ and assigning the relevant context to each one of them:
14
In English there is also the verb ‘contest’ which corresponds to the German ‘bestreiten’.
4. The study
67
Conceptual translational ambiguities arise “when a single ‘concept’
represented by one word in one language corresponds to a number of concepts, and
hence words, in another language” (Hutchins/Somers 1992: 101). This would be the
case for the English ‘wall’ which is either ‘Mauer’ (outside) or ‘Wand’ (inside) in
German.
An example from the text material is the modal verb ‘may’ which has no exact
correspondence in German. The English ‘may’ is very vague and can designate
‘must’/’müssen’, ‘must not‘/’nicht dürfen’, ‘can’/’können’ or ‘will’/’werden’:
(S1-1) Associations must comply with any trophy use guidelines that the UEFA administration may issue from time to time.
(C1-1) The associations must comply with trophy use guidelines of any kind that the UEFA administration can issue from time to time.
(S1-2) Additional medals may not be produced.
(C1-2) Additional medals will not be produced.
(S1-3) Associations may not, and may not permit any third party to, develop, create, use,
sell or distribute any promotional materials […].
(C1-3) Associations must not develop, create, use, sell or distribute any promotional materials […].
Theoretically, several entries could be created with the Systran Dictionary
Manager. The user would then have to choose the correct translation for every
occurrence (in the translation editor) by clicking on ‘alternative meanings’ and picking
the appropriate entry. Since ‘may’ appears no fewer than eight times in S1 alone and
12 times in S2, I created CL rule 13 (‘Avoid the ambiguous modal verb may’). It would
be rather time-consuming to decide on the correct equivalent each time the verb
appears. It is definitely easier and quicker to resolve this problem beforehand during
the controlling stage.
A special case are lexical gaps that occur when a concept existing in the SL
does not exist in the TL and so no one-to-one equivalent can be found, e.g. the
Fig. 37: Dictionary entries for the verb ‘play’
4. The study
68
German adverb ‘doch’ which is, like the French ‘si’, used to affirm negative questions.
In English, there is no equivalent concept, negative questions are affirmed by ‘yes’.
But also words which have no single lexical item corresponding in another language
and which have therefore to be translated periphrastically are regarded as lexical
gaps (e.g. the German verb ‘schweigen’ and its English equivalent ‘to keep/remain
silent’). An example from the text material is the German adjective ‘unentschieden
[enden]’ which does not correspond to a simple adjective in English, but to the phrase
‘[end] in a draw’, or three lexical items. However, this is not a problem for Systran
since the adjective can simply be added to the dictionary: f
Structural differences
Turning to structural differences, the most important cause of translational
problems arising from transfer ambiguities, I will not examine all possible related
problems, but illustrate the main ones I had to deal with as part of the study. The main
problem is the different sentence structure and order of information which Systran
cannot cope with, above all when sentences contain temporal indications:
(C1) The teams must arrive at their transfer hotel at least 24 hours before kick-off;
(TC1) Die Mannschaften müssen in ihrem Transferhotel spätestens 24 Stunden vor
Spielbeginn ankommen;
(PE2) Die Mannschaften müssen spätestens 24 Stunden vor Spielbeginn in ihrem Transferhotel eintreffen;
The system abided with the typical grammar rule that in German – unlike in
English – verbs come at the end of a sentence. However, the temporal indication ‘at
least 24 hours before kick-off’ was not inserted in the right place, thus makes post-
editing necessary. The same problem arose with adverbs such as ‘not’/‘nicht’, ‘five
times’/’fünfmal’ or the previously discussed adjective ‘in a draw’/’unentschieden’. In
some cases, the inability of the system to correctly apply the TL text’s sentence
structure was due to ‘word sequence’ dictionary entries. Word sequences are not
assigned any grammatical features (such as the categories ‘verb’, ‘noun’, etc.), which
means the system has to guess in which place to insert them.
Problems can also arise as a result of reflexive verbs. There is no wholesale
rule in German concerning the position of the reflexive pronoun ‘sich’/’oneself’, which
can be placed in different positions, depending on context. For this reason, Systran
sometimes has difficulties correctly positioning reflexive verbs in the TL text:
Fig. 38: Lexical gap
4. The study
69
(C1-1) If no goals are scored during extra time, the team that qualifies for the final
tournament is determined by kicks from the penalty mark […].
(TC2-1) Wenn keine Tore in der Verlängerung erzielt werden, wird die Mannschaft, die für die Endrunde sich qualifiziert, durch Elfmeterschiessen ermittelt […].
(PE2-1) Wenn keine Tore in der Verlängerung erzielt werden, wird die Mannschaft, die sich für die Endrunde qualifiziert, durch Elfmeterschiessen ermittelt […].
(C1-2) […]: if both associations in question agree and if the principles governing the
release of players for association teams as laid out in Annex […] are complied with.
(TC1-2) […]: wenn beide betroffenen Verbände sich einigen und wenn die Grundsätze betreffend […]eingehalten werden.
(PE2-2) […]: wenn sich beide betroffenen Verbände einigen und wenn die Grundsätze
betreffend […] eingehalten werden.
In both examples, the incorrect sentence order in German results from the fact
that the sentences contain inversions, which obviously confused Systran’s reordering
algorithm. It is not possible to indicate Systran sentence structure rules, making post-
editing necessary in these cases.
Structural differences can also result from different systems of singular and
plural use. In S1, this is the case for several nouns, such as ‘a third party’, which in
English is normally singular and preceded by the indefinite article ‘a’, whereas in
German ‘Dritte’ is plural and never preceded by a determiner. The plural form ‘third
parties’ which is not preceded by any article also exists in English, but the use of the
singular form largely prevails in UEFA regulations. This issue can be resolved by
adding the following dictionary entries:
There remains however the problem of inflection. As the following example
shows, the system does not correctly inflect ‘Dritte’: in this context, the preposition
‘zwischen’ entails the use of the dative, whereas Systran simply used the nominative.
Even though I indicated the use of the dative case through the expert coding function,
Systran did not apply my grammar rule:
Fig. 39: Different systems of singular and plural use
4. The study
70
When I replaced ‘a third party’ by another plural noun, the German translation
was correct, meaning that the preposition ‘zwischen’ is correctly coded in the system.
I was not able to resolve this problem. Other examples of different systems of singular
and plural use are the nouns ‘semi-finals’/’Halbfinale’15 and ‘regulations’/’Reglement’.
The final problem resulting from structural differences I will discuss here
concerns different systems of passive and active use. This is an issue which often
arises between two languages. Since I prescribed the use of active constructions
rather than passive ones in the writing rules, the controlled source text naturally
contained very few active constructions. Problems still remained, as in the following
sentence:
(C1) The teams of the host associations for the final tournament, Poland and Ukraine, qualify automatically for the final tournament.
(TC1) Die Mannschaften der Ausrichterverbände für die Endrunde, Polen und die
Ukraine, qualifizieren sich automatisch für die Endrunde.
(PE1) Die Mannschaften der Ausrichterverbände für die Endrunde, Polen und die Ukraine, qualifizieren sich automatisch für die Endrunde.
(PE2) Die Mannschaften der Endrunden-Ausrichter, Polen und die Ukraine, sind automatisch für die Endrunde qualifiziert.
In English, the verb ‘qualify’ is usually employed in active constructions,
whereas in German, it is more natural to use ‘sich qualifizieren’ in passive
constructions (‘qualifiziert sein’). This is a complex issue which cannot be resolved via
dictionary entries. The active construction is not wrong in German, but a human
translator would always employ the passive form of the verb. This is one of the cases
where the human translator’s innate sense of language is necessary in order to obtain
a perfect translation.
Other divergences
Other translation problems were caused by divergences of perspective
between different languages and the ensuing impossibility to literally translate
certain expressions. Idioms are a typical example, but other constructions also
15
In German there also exists a plural form (‘Halbfinalspiele’). The use of the singular or plural form is a nuance: the singular form designates the round, whereas the plural form literally stands for ‘the matches of the semi -final round’.
(C1) Associations must not use such a representation in a manner that could lead to a link between a third party and the trophy, the replica trophy and/or the competition.
(TC1) Verbände dürfen solch eine Darstellung nicht in einem Kontext verwenden, die zu eine Assoziation zwischen Dritte und dem Pokal, der Nachbildung und/oder dem
Wettbewerb führen könnte.
(PE2) Verbände dürfen solche Darstellungen nicht in einem Kontext verwenden, der zu einer Assoziation zwischen Dritten und dem Pokal, der Nachbildung und/oder dem Wettbewerb führen könnte.
4. The study
71
cannot be translated word for word since they are expressed completely differently in
two languages. For example, it is impossible to literally translate the following English
expression originating from characteristic legal language:
(C1) The associations must not approve the use of a replica trophy in a context where a
third party is granted visibility. (TC1) Die Verbände dürfen die Verwendung einer Nachbildung in einem Zusammenhang
nicht genehmigen, in dem Dritte auftreten dürfen.
(PE2) Die Verbände dürfen keine Verwendung der Nachbildung in einem Zusammenhang genehmigen, in dem Dritte auftreten dürfen.
Not only is the perspective different (passive in English, active in German),
the literal translation of ‘grant visibility’, ‘Sicht bewilligen’, does not make any sense
either, making it necessary to create the following dictionary entry:
The infinitive forms ‘to grant visibility’ or ‘to be granted visibility’ are not
recognised in the translation editor. Without the additional information ‘invariable’ in
German, the verb ‘dürfen’ is conjugated in the singular, which is not correct in this
sentence. By checking other regulations, I ascertained that this expression was
always used in conjunction with the noun ‘third party (singular)’/’Dritte (plural)’, thus
ensuring that the dictionary entry did not cause noise in other contexts.
Another interesting divergence between English and German was ‘lift the
trophy’ and ‘den Pokal in den Händen halten’/‘den Pokal in die Höhe
heben/stemmen/recken/strecken’. The literal translation ‘den Pokal stemmen’ is not
idiomatic in German, the expression has to be translated more freely (literally: ‘hold
the trophy in the hands’/‘lift the trophy into the air’). Using Systran, this can be
achieved by creating the following dictionary entry:
In the following example, a whole subordinate clause had to be post-edited
because the perspective was completely different in the two languages. In English, a
match is prolonged by ‘playing an extra time of two periods of 15 minutes at the end
of the match’ whereas is German, the match is literally ‘prolonged by two times 15
minutes’16:
(C1) If this procedure does not lead to a decision, i.e. if both teams score the same number of home and away goals, an extra time of two periods of 15 minutes is
played at the end of the return match.
(TC1) Wenn dieses Verfahren zu keiner Entscheidung führt, d.h., wenn beide Mannschaften gleich viele Heim- und Auswärtstore erzielen, wird eine Verlängerung von zweimal 15 Minuten am Ende des Rückspiels gespielt.
16
Also the expression ‘ein Spiel geht in die Verlängerung‘ (literally ‘a match enters the extra time’) would be correct in German.
4. The study
72
(PE1) Wenn dieses Verfahren zu keiner Entscheidung führt, d.h. wenn beide
Mannschaften gleich viele Heim- und Auswärtstore erzielen, wird das Rückspiel um zweimal 15 Minuten verlängert.
The same problems arise with personifications. The German language makes
very little use of this stylistic device and mostly prefers impersonal or passive
expressions. This is why the following translation also had to be post-edited:
(C1) This standard fixture list will take into consideration relevant factors for the match organisation, such as weather conditions.
(TC1) Dieser Standardspielplan berücksichtigt Faktoren, die für die Spielorganisation
relevant sind, wie klimatische Bedingungen.
(PE2) In diesem Standardspielplan werden Faktoren berücksichtigt , die für die Spielorganisation relevant sind, wie z.B. klimatische Bedingungen.
In short, divergences of perspective between different languages are often far
too complex to be resolved by means of dictionary entries. In these cases, post-
editing is the only way to obtain correct and idiomatic results.
4.3 Conclusion
At the beginning of this chapter, I formulated the hypothesis that the use of CL
would strongly improve the MT output of the text material. In order to test my
hypothesis, I first had to produce results to serve as a basis for the evaluation phase.
The chapter explained in detail the four steps taken to achieve this.
The first step of the study consisted in creating an individual user dictionary
and TM with the Systran Dictionary Manager. This stage preceded the machine
translation of the text material and was necessary in order to tailor Systran to the text
material. Secondly, I established a controlled version of S1 (C1) to enable me to
create MT-oriented writing rules for UEFA regulations. Thirdly, I proceeded to double-
check my MT-oriented writing rules for UEFA regulations on the basis of a second SL
text. In retrospect, I must say that further tests including UEFA employees would have
to be conducted in order to reach a more reliable conclusion. The test revealed very
different results depending on the background of the subjects. Given that one of the
subjects had satisfied my expectations, I can be positive about the general
applicability of the writing rules as long as the writer or post-editor has some practical
experience with CL. This finding entails that UEFA employees would probably need
some thorough introduction to the topic before drafting regulations according to CL
rules. With regard to noise and silence issues, I can conclude that my UD was proven
to be practically free of problems. On the downside, I realised that my TM did not
serve its purpose because of its very high match threshold. The fourth and final step
of the study consisted in carrying out both a rapid and a full post-editing of the raw MT
4. The study
73
output of C1 (TC1). I then analysed the post-editing process and found that the
additional number of interventions made and the extra time spent did not differ to a
great extent between the full and the rapid PE. The discrepancy in the number of
interventions was 34, whereas the time difference amounted to approximately 8
minutes for the excerpt consisting of 2 709 words. Given the rather minor additional
effort needed for a full PE of a translation originating from a controlled SL text, the
analysis in chapter five of the results of the surveys on the controlled MT output (TC1,
PE1 and PE2) will be all the more interesting. .
Once the scene had been set for the evaluation of results, I discussed the
different types of software and translation problems encountered while working with
Systran. I observed that Systran Premium Translator (version 6.0.8.0.) did not work
flawlessly. Some basic functions need to be revised, namely the slowness of the
dictionary and the constantly recurring bugs have to be remedied in order to be able
to fully recommend the software. Regarding translation problems remaining despite
the control of the SL text, I observed how sophisticated language can be. This
becomes all the more visible in translation and can lead to problems which are often
too complex to be resolved by means of MT dictionary entries or a controlled source
text, except when working with a very limited, standardised dictionary (e.g. Simplified
English). In these cases, post-editing is the only way to obtain correct and idiomatic
results. Since FAHQT is still far from being a realistic goal, it is widely acknowledged
that post-editing is necessary in MT, as “most HT, particularly when produced in
translation agencies, is also revised (‘post-edited’) before submission to clients”
(Hutchins 1986: 166) (see section 2.5).
In the next chapter, I will analyse the results of the study by means of different
evaluation stages, thereby testing my hypothesis that the use of CL would
significantly improve the MT output of the text material.
.
5. Results
74
5 Results
After having provided a detailed description and analysis of the study setup
and discussed the translation problems I faced, I will now proceed with the evaluation
of the results. For this purpose, I tested my hypothesis (‘The use of CL will
significantly improve the MT output of the text material’) by comparing the controlled
MT output (TC1) to the non-controlled (NCT1). The evaluation was conducted in a
black box, in other words, only the MT inputs and outputs were taken into account,
and not the mechanics of the translation engine (White 2003: 225).
In the interest of maximum objectivity, the evaluation was carried out in two
stages. Since it may generally be agreed that “human evaluation has serious
drawbacks” (Coughlin 2003: 63) because it relies not only on subjective judgements,
but is also time-consuming and costly, I first proceeded with a quantitative evaluation
of the target text material (TC1 and TC2). I did not carry out an error analysis by
counting errors, i.e. the deviations of the MT outputs from a reference translation, and
calculating scores according to weighted schemes since this method is also time-
consuming and subjective. It is often difficult to analyse errors and to “specify where
one error starts and another ends”, which entails that “the assignment of a weighting
to such complex errors is […] a tricky business” (Arnold et al. 1994: 166). Instead, I
used the analysis function of SDL Trados Studio 2009 to quickly and conveniently
obtain objective data providing clear results regarding my hypothesis (4.3.1).
Secondly, I carried out different types of qualitative evaluations in order to corroborate
the results of the quantitative evaluation (4.3.2). To this end, I began with an
evaluation of the source text material (C1) by an English translator from UEFA SLAN.
The target text material was then evaluated by different subject groups. The German
translators from UEFA SLAN evaluated sentences from different MT stages as well
as from the official human translation, while a subject group made up of translators
with different professional backgrounds evaluated the overall quality of the controlled
MT output after it had undergone a rapid PE.
Section 5.3 will give an overview of the results, following the description and
analysis of the evaluation process.
As shown in chapter 4 (page 40f.), I will be using the following abbreviations
for the text material:
S1 Main SL text material
S2 Second SL text material
C1 Controlled version of S1
C2 Controlled version of S2
5. Results
75
C2/1 First part of C2, carried out by subject 1
C2/2 Second part of C2, carried out by subject 2
TC1 Raw MT output of C1
TC2 Raw MT output of C2
HTS1 Official human translation of S1
NCT1 Raw MT output of S1
NCT2 Raw MT output of S2
PE1 Minimal PE of TC1
PE2 Full PE of TC1
PE3 Minimal PE of TC2
5.1 Quantitative evaluation
The quantitative evaluation method used in this chapter is based on a study by
Hajič, Homola and Kuboň (2003) for which the authors used Trados Translator’s
Workbench (currently Trados Studio), “one of the most popular TM based commercial
systems […] for a relatively fast and natural method of evaluation of the translation
quality of MT systems” (Hajič, Homola and Kuboň 2003: 157). They propose an
evaluation method which “exploits the matching ability of Trados Translator’s
Workbench for expressing the degree of similarity of a text produced by [a] MT
system and the text postedited [sic!] by a human translator” (idem: 163). In their
study, Hajič et. al provided a human translator with a TM (alignment of raw MT output
and SL text) which the latter used while translating. The translator was “obviously free
to make any changes to the text proposed by the translation memory” (idem). The
translation done by the human translator (in this case, the post-edited version of the
MT output) was subsequently compared to the raw MT output by analysing both texts
in Trados with the previously created TM. “Trados […] evaluate[d] the percentage of
match in the same manner as it normally evaluates the percentage of match of
source text with sentences in translation memory [sic!]“ (idem). The results can
therefore be evaluated, knowing that the higher the number of matches, the closer the
MT output is to the reference translation (i.e. the post-edited MT output).
Given the fact that the study carried out by Hajič et al. did not take into
account the use of a controlled language, I opted for a slightly adapted method in
order to evaluate the various MT outputs (TC1 and NCT1; TC2 and NCT2). Roturier
revisited the aforementioned study in 2004 and used a very similar method: he
created a TM of the controlled source text and the post-edited MT output, the latter
being considered as the reference translation, and subsequently used this TM to
analyse the raw MT output of his controlled and uncontrolled SL sentence sets.
Roturier points out that he deliberately used the post-edited MT output and not a
5. Results
76
human translation as the reference translation because “human-translated reference
translations could be syntactically or stylistically different from excellent MT outputs”
(Roturier 2004: 6). The same idea can be found in Newton (1992: 4), who considers
“direct comparison between a system’s raw output and human translation pointless”
since raw MT output is ”rarely regarded as a finished product; like other raw
materials, it is converted into a finished product only through human agency (i.e.,
post-editing)” (idem).
I adhered to Roturier’s evaluation method and created three translation
memories using Winalign, the alignment tool that came with SDL Trados Translator’s
Workbench 2007: TM1 (alignment of PE1 and C1), TM2 (alignment of PE2 and C1)
and TM3 (alignment of PE3 and C2). Using these TMs, I then analysed TC1 and
NCT1 as well as TC2 and NCT2 with Trados Studio 2009 in order to find out how
much higher the match scores of TC1 would be compared to NCT1 (idem for TC2
and NCT2). I expected to corroborating my hypothesis (‘The use of CL will
significantly improve the MT output of the text material’). Owing to the use of two TMs
for TC1 and NTC1, I also wanted to ascertain the degree of PE effort needed for each
PE level (PE1 and PE2) by examining the segments that needed minimal or no
modification.
The analysis of NCT1 and NCT2 (i.e. the raw MT outputs translated from the
uncontrolled SL texts) using TMs made up of the post-edited MT outputs translated
from the controlled SL texts obviously entailed the problem of syntactic and stylistic
differences. Due to differences in the sentences’ length and complexity, the automatic
evaluation of NCT1 using TM1 and TM2 and of NCT2 using TM3 is not completely
reliable. However, it provides – for the analysis of TC1 and TC2 – an indication of the
number of segments which need minimal or no modification. In any case, it can be
assumed that the MT of particularly long and complex SL text sentences can hardly
provide satisfying results. Therefore, the problem of possibly not encountering a high
number of matches between these very long sentences originating from NCT1 and
NCT2, on the one hand, and the generally simplified and therefore shorter sentences
of the TMs, on the other hand, is less marked.
As to the results of the quantitative evaluation, I will first compare the two
analyses of TC1 by Trados. To this end, I used both TMs made up of the two post-
edited versions of TC1, i.e. TM1 (PE1-C1) and TM2 (PE2-C1). Tables 1 and 2 below
show that using either TM1 or TM2 has a different impact on the matching scores. By
analysing TC1 with TM1, the percentage of new segments is only 8.83% against
15.30% for TM2, which is almost twice as high. However, the 100% matches, i.e. the
5. Results
77
segments not requiring any post-editing at all17, differ to a lesser extent (162
segments, or 41.46%, for TM1 and 159 segments, or 38.67%, for TM2). A high
number of fuzzy matches (between 75% and 99%) for both analyses (63 segments,
or 46.11%, for TM1 and 57 segments, or 41.5%, for TM2) indicates that PE1 and PE2
both contain a high number of segments which underwent only minor modifications.
These results confirm my post-edition analysis (see section 4.1.4), revealing that the
additional number of interventions and the extra time spent did not differ to a great
extent between full and rapid post-editing.
Table 3: Trados analysis of TC1 using TM1
Type Segments Words Characters Percentage
100% 159 915 5923 38.67%
95% - 99% 19 288 1838 12.17%
85% - 94% 22 359 2368 15.17%
75% - 84% 16 335 2217 14.16%
50% - 74% 4 76 467 3.21%
New 23 362 2359 15.30%
Total 249 2366 15369 100% Table 4: Trados analysis of TC1 using TM2
I then carried out the same evaluation for NCT1. Tables 3 and 4 below show
that the matching scores of the analysis with either TM1 or TM2 differ only to a very
small extent, the proportional deviations being similar to the ones obtained from the
analysis of TC1. The percentage of new segments is barely different for either TM
(60.86% for TM1 and 62.94% for TM2); the same being true of 100%-matches (71
segments, or 13.13%, for TM1 and 68 segments, or 11.62%, for TM2) and the fuzzy
matches between 75% and 99% (40 segments, or 19.76%, for TM1 and 42
segments, or 21.97%, for TM2). Only the number of new segments, which increased
by almost half when comparing the analyses of TC1 using TM1 and TM2, is much
lower for the analyses of NCT1 with TM1 and TM2 (almost no deviation).
17
It should however be noted that a high number of perfect matches is partly due to matches in short
and very short segments such as titles (‘Article 1’, ‘III Trophy, Plaques and Medals’) and enumerations
(‘Winner Match 1 vs Winner match 3’), as well as to Systran TM entries which are, as mentioned above,
only taken into account for perfect matches.
Type Segments Words Characters Percentage
100% 162 981 6339 41.46%
95% - 99% 27 458 3025 19.36%
85% - 94% 24 381 2551 16.10%
75% - 84% 12 252 1636 10.65%
50% - 74% 3 53 314 2.24%
New 15 209 1298 8.83%
Total 249 2366 15369 100%
5. Results
78
Type Segments Words Characters Percentage
100% 71 303 2190 13.13%
95% - 99% 11 94 559 4.07%
85% - 94% 20 266 1702 11.53%
75% - 84% 9 96 563 4.16%
50% - 74% 9 142 943 6.16%
New 77 1404 9037 60.86%
Total 198 2307 15004 100% Table 5: Trados analysis of NCT1 using TM1
Type Segments Words Characters Percentage
100% 68 268 1948 11.62%
95% - 99% 9 60 310 2.60%
85% - 94% 21 285 1920 12.35%
75% - 84% 12 162 935 7.02%
50% - 74% 6 73 459 3.16%
New 80 1452 9392 62.94%
Total 198 2307 15004 100% Table 6: Trados analysis of NCT1 using TM2
I then compared the overall results of the analyses of the controlled source
text and the uncontrolled source text (see figure 40) by calculating the respective
average percentages of new segments, 100% matches and fuzzy matches between
75% and 99% for TC1/TM1 and TC1/TM2 as well as for NCT1/TM1 and NCT1/TM2.
Comparing the average results of TC1 and NCT1, I can clearly corroborate my
hypothesis and draw the preliminary conclusion that the use of CL significantly
improved the MT output of the text material.
Fig. 40: Comparison Trados analyses of TC1 and NCT1
Next, I carried out the same evaluations for TC2 and NCT2, using TM3, which
consisted of the post-edited version of TC2 (PE3-C2). The analysis of TC2 and NCT2
was particularly interesting since the control of S2 had been carried out by
independent subjects, whereas the control of S1 was not completely objective as I
had done it myself by directly observing the translation results in the Systran
0,00%
10,00%
20,00%
30,00%
40,00%
50,00%
60,00%
70,00%
New segments
100% matches
High fuzzy matches
12.10%
40.10% 43.80%
61.90%
12.41% 20.87% TC1
NCT1
5. Results
79
translation editor (see section 4.1.2). The following analysis will therefore reveal to
what extent the previous results may be biased.
For TC1 and NCT1 I proceeded with a comparison of the overall results of the
analysis of the controlled source text and the uncontrolled source text, including the
respective average percentages for new segments, 100% matches and fuzzy
matches between 75% and 99%. Figure 41 gives an overview of the data.
Type Segments Words Characters Percentage
100% 69 421 2808 18.57%
95% - 99% 8 63 420 2.78%
85% - 94% 16 200 1354 8.82%
75% - 84% 17 384 2428 16.94%
50% - 74% 11 213 1313 9.40%
New 60 933 6144 41.16%
Total 196 2267 14806 100% Table 7: Trados analysis of TC2 using TM3
Type Segments Words Characters Percentage
100% 55 279 1870 12.42%
95% - 99% 6 41 271 1.82%
85% - 94% 8 82 547 3.65%
75% - 84% 17 223 1459 9.92%
50% - 74% 6 98 642 4.36%
New 85 1441 9602 64.13%
Total 197 2247 14917 100% Table 8: Trados analysis of NCT2 using TM3
These data reveal a difference between the Trados matching scores of TC2
and NCT2, although it is minor, as had been noted for TC1 and NCT1. The control of
S1 entailed a reduction of new segments by 50% and an increase of high fuzzy
matches (between 75% and 99%) by 23%, whereas the percentage of 100% matches
rose by 28%. The direct comparison between TC1 and TC2 as well as NCT1 and
NCT2, illustrated in figures 42 and 43, points out to what extent the quality of the MT
output differs:
0
10
20
30
40
50
60
70
New segments
100% matches
High fuzzy matches
41.16
18.57
28.54
64.13
12.42 15.39
TC2
NCT2
Fig. 41: Comparison Trados analysis of TC2 and NCT2
5. Results
80
Regarding the quality of the MT outputs of the controlled source texts, the
results were clearly better for TC1. In contrast to TC2, TC1 presented a 29%
reduction of new segments, a 21% rise in 100% matches and a 15% increase in high
fuzzy matches. Comparing NCT1 to NCT2, I observed only minor deviances that
probably resulted from the use of different text material. I can therefore conclude that
the control of S2 carried out by two independent subjects had a positive impact on the
quality of the MT output, thereby corroborate once again the previously drawn
conclusion. However, it must be said that the use of CL did not improve the quality of
the MT output to as significant an extent as with S1. This is probably due to the fact
that I controlled S1 in the Systran translation editor following my own CL rules, thus
directly observing the impact of the control on the automatic translation. Neither
subject carrying out the control of S2 had access to Systran or any knowledge of the
specific functioning of the software, e.g. which sentence structures are easily parsed
and which are not. Furthermore, as mentioned in section 4.1.3, they made mistakes
due to a lack of knowledge of football terminology and UEFA regulations.
With a view to drawing an overall conclusion from the quantitative evaluation, I
adjusted my previous hypothesis and determined that the use of CL improved the MT
output of the text material. The fact that the improvement for TC1 was more
significant than for TC2 may indicate that the control of S1 was biased (see above),
but it also hints at the lack of background knowledge of the subjects carrying out the
control of S2. The following restrictions are therefore necessary for further tests: the
control of the text type ‘UEFA regulations’ destined for automatic translation with
Systran must be carried out by persons who have (1) a basic knowledge of UEFA
regulations and, ideally, (2) a basic knowledge of Systran’s functioning. Neither of
these was required of my subjects, which is probably the main reason for the
discrepancies between the results of TC1 and TC2.
0
10
20
30
40
50
New segments
100% matches
High fuzzy matches
12.1
40.1 43.8 41.16
18.57
28.54
TC1
TC2
0
20
40
60
80
New segments
100% matches
High fuzzy matches
61.9
12.41 20.87
64.13
12.42 15.39
NCT1
NCT2
Fig. 42: Comparison Trados analysis of TC1 and TC2
Fig. 43: Comparison Trados analysis of NCT1 and NCT2
5. Results
81
Applying this finding to the UEFA context, it can be said that the scenario of
UEFA employees involved in the regulations core process drafting regulations
according to MT-related writing rules would yield a more acceptable MT output than
the control carried out by my two independent subjects because UEFA employees
have broad background knowledge of the text type in question. Furthermore, linguistic
errors would not be a factor since the final version of regulations are always checked
by English translators from SLAN. However, UEFA employees involved in the
regulations core process are generally familiar with neither translation nor Machine
Translation, entailing that regulations worded based on MT-related writing rules would
not only have to be checked according to the usual procedure, but also according to
their “MTranslatability” (Bernth and Gdaniec: 2001). In other words, persons “having
MT in mind” (Bernth and Gdaniec 2001: 176) would have to check regulations drafted
based on MT-related writing rules. The persons best suited for this task are obviously
translators being trained in CL and MT and who are familiar with the software to be
used. This scenario would have to be compared to the current translation workflow of
UEFA regulations, which is based on machine-aided human translation using SDL
Trados Studio 2011. I will return to this point in my final discussion (section 6.2).
The qualitative evaluations to corroborate the results of the quantitative
evaluation will be covered in the following section.
5.2 Qualitative evaluations
The goal of this section is to corroborate the results obtained in the
quantitative evaluation. As mentioned previously, the human evaluation of
translations is highly subjective because there is no such thing as ‘right’ or ‘wrong’, as
is the case in exact sciences. In this connection, King (1997: 261) points out that “one
cannot imagine artificially constructing the right answer which a machine translation
system should arrive at, any more than one can imagine grading human translations
by comparing them with some single perfect translation”. Furthermore, due to the time
and cost factors, the scale of human evaluation tends obviously to be rather small,
“usually no more than a few hundred sentences [are] examined by a small number of
raters”, meaning that it can be difficult “to draw firm conclusions about system quality”
(Coughlin 2003: 63). Furthermore, “there is evidence to suggest that evaluations are
somewhat compromised by evaluator speed or inattention” (idem: 64). However, I do
not merely want to base my evaluation on the automatic method; I also want to take
into consideration the results of a human quality assessment in order to convince
myself of the reliability of the preliminary results, especially since translators are the
main stakeholders in any MT-based translation workflow.
5. Results
82
For the purpose of this study, I will instead try to measure “whether a
translation is good enough for some specific purpose” (King 1997: 262) and not try to
“define some abstract notion of translation quality” (idem). Hutchins and Somers
(1992:163) assign three factors for the evaluation of MT output: fidelity, intelligibility
and style. For UEFA regulations, the top priorities are definitely fidelity and
intelligibility, whereas style is a minor criterion, making the text type generally suitable
for MT (see section 2.7). Yet, the style factor can not be completely neglected since
the regulations are translated for purpose of dissemination and are accessible to the
general public. The “worst possible scenario would however be inaccurately
reproduced information, since mistranslations even of minor details could have
adverse consequences” (Furlani 2009: 51); for example, articles 4 and 5 of the
Regulations of the UEFA European Football Championship 2010-12 dealing with the
responsibilities of the associations and insurance matters.
In the interest of making the human evaluation as objective as possible, I
established three different questionnaires for different subjects and subject groups. I
began with an evaluation of the source text material (C1), which was carried out by a
translator from UEFA SLAN. Then the target text material was evaluated by different
subject groups (subject group 1 made up of translators from the German translators
from UEFA SLAN and subject group 2 consisting of translators with different
professional backgrounds). Subjects in group 1each evaluated 85 sentences from the
following five texts: the non-controlled MT output of S1 (NCT1), the raw MT output of
C1 (TC1), the minimal and full PE of TC1 (PE1 and PE2) and the official UEFA
translation of S1 (HTS1). Subject group 2, for its part, evaluated the overall quality of
PE1.
5.2.1 Evaluation of source text material
In order to evaluate the overall quality of the controlled source text material
(C1), I designed a questionnaire for UEFA SLAN (see Annex M). This step was
particularly important because if the controlled source text was not accepted by the
professional UEFA English translators, it would be difficult to justify MT combined with
Controlled Language rules. Due to the limited scope of this study and the issues
discussed in section 4.1.3, TC2 was not taken into account for this evaluation. In the
following paragraphs, I will summarise the evaluation results and then discuss the
improvements made by the subject who filled out the questionnaire.
The feedback concerning the overall quality of C1 was generally positive. The
subject agreed with the statement that the text may be stylistically imperfect, but
fulfilled its main objective, which was to accurately transfer all information.
5. Results
83
Furthermore, she agreed that the sentences were short, chopped and repetitive, but
more intelligible than the original long and complicated ones and was, therefore, at
least equal to the latter. According to the subject, it was
“hard to say one is better than the other, as stylistically they’re different and neither is perfect, but because the controlled version is so intelligible, it has the potential to be preferable from a comprehension point of view, in which case the stylistic shortfalls would be justified […]”.
She also stated that C1 (if necessary including minor corrections) met
professional needs and that she would accept this writing style in her position as
English translator and proofreader, on the condition that MT considerably cut
translation costs.
On the downside, the questionnaire also revealed that C1 required minor
modifications to grammar (including illogical clause pairings, inconsistent use of
definite articles, misplaced articles and commas, wrong pronoun use and missing
hyphens). These errors were due to the fact that the control was not carried out by a
native speaker. I corrected all the errors encountered, but will not further discuss
them here; all suggestions for improvement can be found in the questionnaire (see
Annex M).
I also received suggestions for stylistic improvements; for example, the
repeated preposition in the sentence ‘The competition consists of a qualifying
competition and of a final tournament’. This is indeed a very pedestrian construction,
even though the repeated preposition is necessary according to my writing rules
(Rule 27: ‘Repeat the preposition in conjoined constructions where appropriate, even
though this seems to be redundant for humans’). Due to the simple construction of
the sentence, Systran was in this case able to correctly parse the uncontrolled
sentence, so I adopted the subject’s suggestion. For more complex sentences , it is
still in many cases necessary to repeat prepositions in order to ensure a faultless
analysis, which entails stylistic shortfalls that have to be accepted when working with
MT-orientated controlled languages.
I can conclude from the qualitative evaluation of the source text material that
C1 is stylistically imperfect, but still meets professional needs because it accurately
transfers information. The subject’s appraisal of the advantages and drawbacks of
the controlled and non-controlled versions aptly conclude this section:
“Stylistically, even though some sentences are pedestrian and artificial, for regulations the main concern is to be comprehensive and comprehensible. In this respect, the ideal middle ground would be somewhere in between the controlled and non-controlled versions: there could be improvements to both, but the principles applied for the controlled version are sensible principles for
5. Results
84
regulations drafting regardless of translation (short, simple sentences in the active voice).”
5.2.2 Evaluations of target text material
The objective of this section is to analyse the results of the qualitative
evaluations of the target material carried out by two different subject groups. In the
first part of the section, I will assess the results of the questionnaires completed by
the German translators from UEFA SLAN. In the second part, I will analyse the
evaluation carried out by translators from different professional backgrounds.
5.2.2.1 German translators from UEFA SLAN
My questionnaire targeting UEFA SLAN (see Annex N) was completed by all
four native German translators. I drew up two versions of the questionnaire, each
containing 85 sentences to evaluate, which had been extracted randomly, mixed
using the Microsoft Excel filter function ‘random number’ and then divided into two
sentence sets. The sentences came from five different translations: the non-controlled
MT output of S1 (NCT1; 20 sentences), the raw MT output of C1 (TC1; 48
sentences), the rapidly and fully post-edited versions of TC1 (PE1 and PE2; each 40
sentences) and the official UEFA translation of S1 (HTS1; 22 sentences). This way, I
obtained evaluations for 170 sentences, with each sentence evaluated by two
subjects (see tables below). Subjects 1 and 2 evaluated sentence set 1, subjects 3
and 4 evaluated sentence set 2. Due to the limited scope of this study and the issues
discussed in section 4.1.3, the sentences from TC1 and NCT2 were not taken into
account for this part of the evaluation.
Table
9:
Overview of sentence sets
Sentence Set 1 Sentence Set 2
Text type Number of sentences
Number of subjects
Text type
Number of sentences
Number of subjects
NCT1 10 2 NCT1 10 2 TC1 24 2 TC1 24 2 PE1 20 2 PE1 20 2 PE2 20 2 PE2 20 2
HTS1 11 2 HTS1 11 2
5. Results
85
In order to obtain the most objective results possible, the questionnaire
provided only the following information:
“Please have a look at the following sentences. They are all translated from an excerpt of the Regulations of the UEFA European Football Championship 2010-12 using the Machine Translation software Systran. These translated sentences are taken from different post-editing
18 versions (‘raw output’,
‘minimal post-editing’ and ‘perfected post-editing’), which means that the translation quality may strongly vary from sentence to sentence.” (see Annex N)
The subjects therefore did not know that they were not only evaluating MT
output from different PE levels, but also originating from different source texts (S1 and
C1). Furthermore, I added some sentences from the official UEFA translation to the
questionnaire in order to find out if the subjects tended to downgrade their own
translations by believing they were (post-edited) MT output. Thus the evaluation
would also reveal to what extent UEFA SLAN translators might be biased towards
MT. The translated sentences were purposely not accompanied by their respective
SL version since this would have disclosed the use of different SL texts. As UEFA
translators are generally familiar with the regulations, I deemed it appropriate to
provide them with only the translations.
The questionnaire invited the subjects to evaluate the quality of each sentence
by either ticking ‘good’, ‘acceptable’, ‘unacceptable’ or ‘useless’ and to comment on
their choice where appropriate. In retrospect, the instructions given in the
questionnaire were not detailed enough, it would have been necessary to specify the
evaluation criteria as well as the purpose of the translation (in this case
dissemination). A sentence containing minor linguistic errors may be ‘acceptable’ for
assimilation purposes, but not for dissemination (see page 26). Moreover, a ‘good’
translation can be linguistically perfect but contain serious errors at the semantic
level, while ‘unacceptable’ sentences might be full of grammatical errors but still
accurately transfer all information. The lack of specific evaluation criteria was
mentioned by three of the four subjects, thereby indicating that different evaluation
schemes were used. Clearer instructions, such as the following, should have been
provided:
Good No semantic and linguistic errors, no (further) PE necessary
Acceptable No serious errors neither on the semantic, nor on the linguistic level, but quality not good enough for dissemination without (further) PE
18
Following Allen’s definition, post-editing (PE) consists of editing, modifying and/or
correcting “pre-translated text that has been processed by an MT system from a source language into (a) target language(s)” (2003: 297). Post-edition can thus be regarded as the logical counterpart of a human translation’s revision (Koby 2001: 4).
5. Results
86
Unacceptable The sense of the translation is still traceable, but the wording
contains serious errors quality so poor that HT would be more efficient than PE
Useless The sentence does not make any sense at all
Given the lack of such a clear evaluation scheme, I was confronted with the
problem of evaluations following different criteria, which obviously biased the test
results. However, due to the length of the questionnaire and the amount of time
required for its completion, I was unable to repeat the evaluation. That said, I can
roughly deduce the general evaluation scheme the subjects applied from the
comments made on some sentences. In short, sentences were rated as ‘good’ when
there was nothing with which to find fault. The subjects generally chose the rating
‘acceptable’ when the following minor issues were encountered: style (repetitions,
vernacularity, redundancies, word choice), syntax (pedestrian style due to
subordinate clauses, topic-comment/order of information, unclear antecedents, word
order), grammar (wrong prepositions, incoherent use of tenses), semantics (logic),
orthography and punctuation. Sentences were evaluated as ‘unacceptable’ when the
following more serious issues were detected: grammatical errors (wrong
prepositions), style (repetitions. word choice), syntax (logic, word order, wrong
antecedents), punctuation and semantic errors. ‘Useless’ sentences displayed more
or less the same problems as the ‘unacceptable’ ones , but to an even more serious
degree (grammatical errors, syntax, semantic errors, wrong or unclear antecedents).
This evaluation scheme is not objective, however, since every subject applied the
criteria in a different way. Whereas some subjects based a positive evaluation on the
correct transmission of information (by evaluating sentences containing linguistic
errors as ‘acceptable’ when they correctly transferred all information), others focused
more on linguistic issues.
In the following analysis of results, the ratings ‘good’ and ‘acceptable’ will both
be regarded as very satisfying given the following points (see sections 2.4 and 2.5):
1) Quality expectations have to be lowered when working with MT since perfection is
not the goal of machine-translated texts and “information accuracy prevails over stylistic considerations” (Roturier 2004: 6)
2) The acceptance of (albeit post-edited) MT texts can be hard for translators who are
“used to producing high quality texts [and] likely to apply the same sort of output standards to their translations produced automatically” (Arnold et al. 1994: 32)
3) The sentences have been translated from the controlled source text , which means they can by no means be stylistically better than the input text
4) Carrying out excessive post-editing would undo the advantages of working with CL.
Before moving on to the evaluation of the questionnaires, I will discuss the
analysis of the human translation done by Trados using TM1 and TM2 in order to find
out to what extent PE1 and PE2 are identical to the human translation, keeping in
5. Results
87
mind that the number of perfect matches is overestimated owing to short segments
like titles and enumerations. The analyses by Trados resulted in 15.61% perfect
matches between HTS1 and PE1 as well as between HTS1 and PE2 (see tables 9
and 10 below). Fuzzy matches between 95 and 99% totalled 13.27% for PE1 and
14.53% for PE2. These figures reveal that approximately 30% of PE1 and PE2 is
identical or almost identical to HTS1, which gives hope for positive evaluations of the
PE1 and PE2 sentence sets.
Type Segments Words Characters Percentage
100% 81 346 2168 15.61%
95% - 99% 44 294 1969 13.27%
85% - 94% 11 136 975 6.14%
75% - 84% 15 177 1111 7.99%
50% - 74% 3 42 264 1.90%
Nouveau 63 1192 7827 53.79%
Total 222 2216 14496 100% Table 10: Trados analysis of HTS1 using TM1
Type Segments Words Characters Percentage
100% 81 346 2168 15.61%
95% - 99% 46 322 2163 14.53%
85% - 94% 12 152 1065 6.86%
75% - 84% 17 216 1385 9.75%
50% - 74% 5 80 523 3.61%
Nouveau 56 1071 7010 48.33%
Total 222 2216 14496 100% Table 11: Trados analysis of HTS1 using TM2
Turning to the evaluation of the questionnaires, I will first discuss the overall
results of both sentence sets. Figure 44 shows that the sentences from all the
translations generally obtained positive ratings. It was predictable, however, that
NCT1 would contain a high proportion of ‘useless’ and ‘unacceptable’ sentences
(63% in total), bearing in mind that the SL text had not been controlled and the MT
output had not been post-edited. Furthermore, the results of the quantitative
evaluation (see section 5.1) predicted poor results for the qualitative evaluation of
NCT1, which yielded indeed only 38% of ‘acceptable’ or ‘good’ ratings. For TC1, the
proportion of ‘unacceptable’ and ‘useless’ sentences amounted to only approximately
19%. Comparing this figure to the 63% for NCT1, there is a clear corroboration of the
quantitative evaluation and my hypothesis that the use of CLs improves MT results
(under the circumstances defined in section 5.1). 81.3% of the sentences from TC1
were evaluated as ‘good’ or ‘acceptable’, whereas 19% were nonetheless rated as
‘unacceptable’ or poorer, a predictable result since FAHQT is for now an unrealistic
goal (except for certain text types, see section 2.1.1). Given the fact even the best MT
5. Results
88
output does require PE for almost every text type, the result of the evaluation of TC1
is very satisfactory.
The most interesting part of the evaluation is certainly the assessment of the
sentences from PE1 and PE2, which should after the post-editing process be free of
serious errors and therefore not obtain ‘unacceptable’ or even ‘useless’ ratings.
However, 8.8% of the PE1 sentences and 3.6% of the PE2 sentences were evaluated
as ‘unacceptable’. In sum, the analysis of the sentences in question revealed that
none of them contained serious errors, such as grammatical or semantic issues19, but
they did contain clumsy, pedestrian expressions and sentence structures a human
translator would probably not use. Since the sentences rated as ‘unacceptable’ did
not contain any grammatical or semantic errors, and bearing in mind the four points
mentioned on page 86 (lower quality expectations when working with MT, difficulty for
translators of accepting imperfect MT texts, translation from a CL cannot be
stylistically better than the input text, excessive post-editing would undo the
advantages of working with CL), I will disregard the figures for such sentences. When
comparing the evaluations of the PE1 and the PE2 sentences, there is an
improvement of 14.5% for ‘good’ and 10.2% for ‘acceptable’ sentences. These
improvements show that the additional investment for PE2 (eight minutes) was worth
the outcome.
As to the evaluation of the human translation, it is surprising to see that one of
the 22 sentences was rated as ‘unacceptable’. Interestingly, the sentence in question
19
Except for one grammatical error (“*infolge eine Disziplinarmassnahme“), which was overlooked during post-editing and corrected afterwards.
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
NCT1 TC1 PE1 PE2 HTS1
28%
1%
35%
17.7% 8.8%
3.6% 2.3%
25%
44.80% 48,8%
38.6%
18.2%
13%
36.5% 43%
57.5%
80% Good
Acceptable
Unacceptable
Useless
Fig. 44: Results of the qualitative evaluation, sentence sets 1 and 2
5. Results
89
is that discussed in section 4.2.2.1 as an example of monolingual, structural
ambiguities (see page 64). One of the subjects stumbled over the German translation
since ‘einer Nachbildung’ can easily be misunderstood as a grammar mistake when
the sentence is not carefully read. The 18.2% of ‘acceptable’ sentences were mostly
criticised for stylistic reasons. Comparing the overall evaluation of the HTS1
sentences to the PE1 and PE2 sentences, I can conclude that the score for ‘good’
translations is obviously a lot higher for the human translation, but that the sentences
from all three translations were generally positively rated. Moreover, these results
illustrate that the UEFA SLAN translators are not biased towards MT. It is beyond
debate that not all translated sentences, regardless of their origin, can be equally
rated as ‘good’ by four different subjects since translators have different expectations
when it comes to stylistic issues.
The evaluation results of sentence sets 1 and 2 (see table 9, page 84) were
then compared in order to find out if the difference in the sentences yielded diverging
results. According to figures 45 and 46, the results for the raw output of the non-
controlled and of the controlled SL texts differed only a little, whereas the evaluations
of PE1, PE2 and HTS1 (i.e. the versions which were post-edited or translated by
humans) deviated to a greater extent. Sentence set 2 contains far more ‘good’ and
fewer ‘acceptable’ evaluations than sentence set 1 and vice versa. This difference
can partly be explained by the fact that one of the subjects who evaluated sentence
set 1 focused more on the correct transmission of information than on linguistic errors
and style, which probably affected a bigger proportion of sentences rated as
‘acceptable’, assuming that sentences which transmitted the correct meaning
obtained an ‘acceptable’ rather than a ‘good’ evaluation, regardless of their stylistic
quality. The differences between the evaluation of sentence sets 1 and 2 were
probably due to the different evaluation criteria applied by each translator. The results
would presumably have been more in line with each other if I had provided clearer
instructions.
5. Results
90
In short, the first part of the qualitative evaluation yielded positive results not
only regarding PE1 and PE2, but also TC1. These findings clearly corroborate the
results of the quantitative evaluation (section 5.1) and my hypothesis that CL would
improve the MT output of the text material. Nevertheless, the following two limitations
must be kept in mind: (1) the lack of clear instructions regarding the evaluation criteria
and (2) the observations regarding S2 and TC2 made in the quantitative evaluation
(see page 80f.). For these reasons, it must be noted that the results of the first part of
the qualitative evaluation should be understood in light of the imperfect test
conditions. I will revisit this issue in the final discussion (section 6.2).
5.2.2.2 Translators with different professional backgrounds
For the first part of the qualitative evaluation of the target text material (section
5.2.2.1), the German translators from UEFA SLAN each evaluated 85 sentences
extracted from NCT1, TC1, PE1, PE2 and HTS1. For the second part of the
qualitative evaluation, a subject group made up of seven MA students in their final
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
NCT1 TC1 PE1 PE2 HTS1
30%
2,10%
40%
16.7% 10% 7.5%
20%
47.9% 60%
45%
31.8%
10%
33.3% 30% 47.5%
68.2% Good
Acceptable
Unacceptable
Useless
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
NCT1 TC1 PE1 PE2 HTS1
25%
30%
18.7% 7.5% 4.5%
30%
41.6%
37,50%
32.5%
4.5%
15%
39.6%
55% 67.5%
90.9%
Good
Acceptable
Unacceptable
Useless
Fig. 45: Sentence set 1
Fig. 46: Sentence set 2
5. Results
91
year and one employed in-house translator (4 Swiss German and 4 German subjects)
assessed the overall quality of PE1. I chose this text material because of the
advantage of rapid post-editing as the most cost-effective PE method that did not
offset the benefits of working with a CL. It was interesting to see how translators
without background knowledge of UEFA regulations evaluated the target text,
knowing they were faced with (post-edited) MT output. The questionnaire consisted of
a very brief overview of the control applied to the source text and the minimal PE
rules applied to the raw output of TC1 and of five questions (see Annex O). I did not
provide the controlled source text because the post-edited translation was not likely
to contain any terminological and/or semantic errors.
The results of the survey were very satisfying overall. Regarding the general
quality of the translation, six of the eight subjects stated that they would not need the
source text or the official German translation in order to understand the target text,
even though the MT output contained minor grammatical mistakes, such as word
order, word formation and morphology, and that they would not have to consult the
source text in order to correct these mistakes. It is obvious that MT for dissemination
purposes, particularly for a text type that requires utter precision, would never be
carried out without a thorough PE, including a comparison with the SL text. Since the
subjects knew the text had undergone a rapid PE, they assumed there would not be
any semantic errors left in the translation, enabling them to conclude that they could
carry out further PE of linguistic issues without considering the SL text. In that
connection, one subject made the following remark:
“in cases like this one, where there are official rules to be translated, I would be using Machine Translation rather carefully. Given the fact that translations of rules rarely have legal character and that you always have to refer to the original text, Machine Translation might be a solution to be taken into consideration. However, considerable importance should be granted to post-editing including a comparison with the source
text.”
One subject even considered the MT output satisfactory enough without
further modifications because, in spite of stylistic imperfections, the main objective of
accurately transferring all information was met. Only one subject – interestingly the
professional in-house translator – indicated that her understanding of PE1 would be
improved by consulting the source text and that she would have to re-read it a few
times in order to correct the partly significant errors of the MT output.
When asked whether the translation met professional requirements, with
additional minor corrections where necessary, six of the subjects answered ‘yes’,
whereas only two answered ‘no’. The reasons the subjects gave for PE1 not meeting
professional requirements were the following:
5. Results
92
The translation is comprehensible, but at many points unidiomatic and stylistically inadequate
The text type is distorted due to the non-use of typical German nominal style
The often pedestrian and distracting sentence structure puts at risk the comprehensibility of the text
The main target of the text type (accuracy) is not complied with: the translation is partly open to interpretation
Too many short sentences The text is reader-unfriendly due to missing clause linkages The text flow is interrupted due to ‘un-German’ prepositions and
antecedents caused by the static approach of the MT system regarding
the beginning of sentences
One subject made the observation that, in contrast to the first half of the text,
the second part revealed a lot more good translations thanks to many enumerations
and shorter sentence sequences. She indicated that some parts of PE1 would need a
complete retranslation from scratch, whereas the shorter sequences and
enumerations could be left unchanged. Due to the passages she considered as
requiring retranslation from scratch, she deemed that PE1 did not meet professional
requirements.
Concerning the acceptability of MT assuming it considerably cut translation
costs, five of the subjects were open to MT, whereas three indicated that they would
not accept MT if they were head of language services in any company or
organisation. The answers to this question are especially interesting in respect of the
extent to which translators might be biased on the topic of MT. Indeed, one subject
indicated that she deemed the translation compliant with professional requirements
and would accept MT for economic reasons, but still commented that she “would not
recommend MT in general” although without further explaining why. However, this
was an isolated case; all the other subjects that took a negative view of MT gave
clear reasons why. One subject stated that she would not accept MT due to the lack
of stylistic fluency of any MT output, which she set great store by.
Two subjects who were generally open to MT indicated that they had not
expected a system to produce such a good translation. These comments reflect the
common view among many translators that MT output is inevitably of bad quality and
can by no means measure up to a human translation. One of the subjects was
impressed by the high quality of PE1 and stated that she thought MT was a good
option
“for texts where the interaction between form and content is not important, that is to say that [she does not] believe that Machine Translation results would be able to satisfy the needs of a literary text or a press release – Machine Translation might be able to convey the message but not the form.”
5. Results
93
This is a very fitting comment which summarises the characteristics of suitable
text types and applications of MT discussed in the theoretical part of this paper (see
section 2.7). It also demonstrates that working with MT for economic reasons requires
lowering one’s expectations of the linguistic, and consequently stylistic, quality of the
translation (see section 2.5).
The questionnaire also asked the subjects to indicate sentences or passages
they considered as requiring further PE, and in such cases to give suggestions for
improvement. There was a general consensus regarding most grammar and spelling
errors and some striking stylistic issues. However, the subjects’ evaluations regarding
style and sentence structures sometimes revealed great divergences. The most
probable explanation of these variations might simply be the amount of time invested
in answering the questionnaire. Some subjects made a lot of suggestions for
improvement and very detailed comments, whereas others barely made any
recommendations regarding style and semantics.
The suggested improvements can be classified as follows: (1) Errors
encountered (i.e. grammar, terminology), (2) Necessary adjustments arising from the
controlled SL text structure, (3) Necessary adjustments arising from the translation
approach of the system and (4) Criticism of sequences equally occurring in the official
translation. Criticisms resulting from the lack of background knowledge about football
and/or UEFA as well as subjective suggestions for improvement (i.e. when there are
several correct solutions) were disregarded. For the sake of clarity, I created four
detailed tables displaying the sentences in question (respectively the PE1, C1 and
HTS1 version) and the suggestions made. I will, however, discuss only a selection of
representative, telling examples. All of the subjects’ suggestions can be found in
Annex P.
The subjects detected five errors that will not be reviewed in detail here; I
corrected and classified them as grammar mistakes (due to insufficient PE), spelling
mistakes (due to errors in the dictionary), terminology errors or inconsistencies (due
to erroneous dictionary entries or alternatives in the Systran translation editor which
had not been correctly chosen) and inconsistencies (due to incoherent control, e.g.
use of singular and plural).
Many of the subjects’ suggestions for improvement arise from the controlled
SL text’s structure (see table II, page 215). The ten comments made in that regard
can be summed up under the following six categories:
5. Results
94
Syntax (unusual sentence order) Coherence/Style:
- Separation of sentence parts into independent sentences often not convenient
- Separation of if-clauses into several sentences unusual Pedestrian and not idiomatic constructions due to unnecessary relative
clauses requiring streamlining Repetitions and redundancies demanding ellipsis (e.g. pronouns) Pedestrian use of modal verbs Order of information (topic-comment structure)
The following example is representative of pedestrian structures resulting from
the use of CL (n° 3 from table II):
N° PE1 C1 HTS1 Criticism 3 Wenn ein Verband
einen
Trainingsplatz wählt, der nicht Teil dieser
Vorauswahl ist, trägt der Verband alle Kosten, die
entstehen.
Should an association choose a training
ground that is not part of this preselection, the association bears all the
costs that are incurred.
Wählt ein Verband einen Trainingsplatz,
der nicht zu den ausgesuchten Plätzen gehört,
übernimmt er alle dadurch entstehenden
Kosten.
Unnecessary relative
clause, construction too
pedestrian not idiomatic
Due to writing rule 29 which stated that relative clauses should not be
reduced, but always written in the expanded form, the sentence in C1 was expanded
from ‘all costs incurred’ to ‘all the costs that are incurred’. Consequently, the more
pedestrian SL sentence leads to an equally pedestrian translation into German (‘alle
Kosten, die entstehen’), which is correct (unlike the MT output of the uncontrolled SL
sentence: ‘alle Kosten genommen’). Further examples can be found in table II (page
215).
The criticisms expressed by the subjects are of course justified, especially
from the point of view of translators who are used to producing high quality texts.
However, as said before, the quality of the translation can by no means outstrip the
quality of the SL text. If the TL text has been translated from a controlled SL text, as
was the case in this study, one obviously cannot expect a stylistically perfect output.
C1 is characterised by short, simple sentences, many repetitions (which may be
unnecessary for humans, but are necessary for correct parsing by the system, e.g.
pronouns, prepositions etc.), the complete formulation of relative clauses, etc. (see
MT-oriented writing rules for UEFA regulations in Annex F). These stylistic shortfalls
have to be accepted when working with MT and CL, since excessive PE would undo
the advantages of working with a CL (see section 2.5). It is therefore inevitable for
any company, organisation or association planning to introduce CL in combination
5. Results
95
with a MT system to be clear about these reductions in language and style and to
decide whether these are acceptable in view of the money and time saved.
The eight suggestions for improvement relating to the translation approach of
the system can be classified into the following five categories:
Order of information (topic-comment) Nominal style in German often more common than verbal
constructions Pedestrian style Ambiguous sense due to unclear antecedents
Use of tenses
All these are due to the same problem, i.e. the limited ability of a MT system to
change sentence structures. Systran does adhere to certain rules, for instance
placing verbs at the end of German sentences. However, sentence structures are
often translated in the same way as in the SL text. Due to word-for-word translations,
the MT output can partly seem unnatural or artificial, as in the following examples (n°
3 and 4 from table III, page 218):
N° PE1 C1 HTS1 Critique
3 (einschliesslich Bildern, die die Sieger zeigen, die
den Pokal in den Händen halten)
(including, without limitation,
images that show the winners lifting
the trophy)
(einschliesslich Bildern von Pokalübergaben,
auf denen der Pokal zu sehen ist)
Personification Proposition: auf denen zu sehen ist
4 Zusätzliche Medaillen werden
nicht produziert.
Additional medals will
not be produced.
Die Herstellung zusätzlicher
Medaillen ist nicht erlaubt.
Sentence structure Proposition:
nominal style, passive
The criticism of example n°3 is due to the fact that Systran adhered to the
English sentence structure, which in this case is different from the German. In
English, the personification of ‘images’ is not an issue, whereas the German language
is rather wary of using this stylistic device. In HTS1, the sentence structure was
changed into a typical German one, without using personification. Example n°4 was
criticised for its unusual sentence structure, which resulted from a word-for-word
translation by the system. The use of nominal phrasing is very common in German
and would be a more elegant solution here.
As to the criticisms resulting from the controlled SL text structure, stylistic
problems due to the translation approach of the system also have to be accepted. MT
research has not so far produced systems able to change sentence structures to
produce perfectly natural sounding texts. The output of MT systems can be greatly
improved by thorough dictionary maintenance by the user and improvements by the
5. Results
96
software developers, but in general, institutions planning to work with MT must accept
certain shortfalls due to the system’s limitations.
Turning to the last category of suggested improvements, namely criticism of
sequences also occurring in the official translation, the comments can be categorized
as follows:
Differences between Swiss German and German linguistic usage Personifications Overly complicated and therefore incomprehensible sentences Syntax (proposition: nominal style) Pedestrian constructions (above all due to complicated prepositions
and overly long and complex sentences) Use of modal verbs Style/Word order
Grammar
I will discuss three examples occurring equally or similarly in HTS1 (from
table IV, page 219):
N° PE1 Criticism
1 Die übrigen Verbände werden auf der Grundlage der UEFA-Koeffizientenrangliste für Nationalmannschaften klassiert (vgl. Anhang I,
Absatz 1.2.1).
Helvetism
2 […]: wenn sich beide betroffenen Verbände
einigen und wenn die Grundsätze betreffend das Abstellen von Spielern für Auswahlmannschaften der Verbände gemäss
Anhang 1, Artikel 1 des FIFA-Reglements bezüglich Status und Transfer von Spielern eingehalten werden.
Pedestrian construction.
prepositions ‚bezüglich‘, ‘betreffend das Abstellen‘ (proposition: hinsichtlich)
3 Die Nachbildungen dürfen das Land des jeweiligen Verbands nicht ohne die vorherige
schriftliche Genehmigung der UEFA verlassen.
Personification, too active; proposition: außerhalb der
Landesgrenzen gelangen or the like
In example n°1, one subject criticised the use of the verb ‘klassieren’ which
she considered acceptable for a Swiss, but not for a German target group. Indeed,
‘klassieren’ is identified as ‘Swiss’ in arguably the most distinguished German
language dictionary (Duden 2012). An alternative suitable for both target groups could
be the verb ‘klassifizieren’. Example n°2 was criticised for being too complicated and
pedestrian due to the over-use of complex prepositions. Several subjects suggested
avoiding the construction ‘betreffend das Abstellen’ because they considered it too
pedestrian and ‘un-German’. One suggestion was ‘hinsichtlich des Abstellens’.
Example n°3 was criticised by one of the subjects for the personification of the noun
‘Nachbildungen’. As mentioned above, it is rather unusual to use personification in
German. In this case, it would probably be stylistically better to re-word.
5. Results
97
The suggested improvements for sequences equal to the official translation
are very revealing, especially since the subjects did not know they were criticising
parts of the official version. Knowing a translation has been carried out by
professional in-house translators would probably restrain evaluators (particularly
students) from levelling too much criticism, whereas criticising MT output seems to be
psychologically easier (see section 2.5, page 21). The comments made about overly
complicated sentences imply that the language in which the regulations are drafted
might be too complex. All of the subjects are, as translators, language experts,
meaning they should generally not have problems taking on difficult texts. However,
they criticised several passages as being too complicated and pedestrian, even
though the TL text resulted from the controlled SL text. The problem may then arise
that average people, who are neither language experts nor law specialists nor used to
reading highly complex texts, could have problems understanding the regulations. It
might therefore be advisable to draft the regulations in a simpler way in order to
ensure their content is correctly conveyed to the target audience (see section 5.2.1).
In sum, the evaluation carried out by translators with different professional
backgrounds yielded positive results. Most of the eight subjects considered PE1 as
comprehensible without referring to the SL text and indicated that the translation met
professional requirements. Most of the subjects also stated that they would be willing
to accept high quality HAMT for economic reasons. The survey further revealed that
only one of the eight subjects was generally biased when it came to MT. However,
some subjects stated being leery of MT, providing reasonable grounds for their
position. Others expressed their surprise at the high quality of the HAMT output,
reflecting the common view among translators that MT can in no way best a human
translator. Based on the suggestions for improvement made by the subjects, when
working with CL in combination with MT, certain stylistic shortfalls have to be
accepted since excessive PE would undo the advantages of HAMT and instead justify
MAHT. It is therefore inevitable for any company, organisation or association planning
to introduce a CL in combination with MT to be clear on these losses of language and
style and to decide whether these are acceptable in light of the money and time
saved and the text types to be translated. As to some sequences from PE1 being
equal to the official translation, the survey revealed that the regulations might be
drafted in overly complex language and that it might therefore be advisable to word
them in a simpler way in order to ensure that their content is correctly conveyed.
5. Results
98
5.3 Overview of results
This chapter covered the various types of evaluation which proved my
hypothesis that the use of CL improved the MT output of the text material. Both the
quantitative and qualitative evaluations carried out by different subject groups yielded
generally positive results. Moreover, the controlled source text material (C1) was
generally approved by an English translator from UEFA SLAN, who confirmed that it
was stylistically imperfect, but still met professional requirements by accurately
transferring information. A further positive result of the study was the finding that most
of the subjects who took part in the evaluation were not generally prejudiced towards
MT.
However, this favourable overall conclusion has to be qualified due to
limitations encountered during the evaluation process. First, the fact that the
improvement to TC1 against NCT1 was more significant than that to TC2 against
NCT2 may indicate that the control of S1 was biased, but it also hints at the lack of
background knowledge of the subjects carrying out the control of S2. The following
restrictions should therefore be set up for further tests: the control of the text type
‘UEFA regulations’ in view of automatic translation with Systran must be carried out
by persons who have (1) a basic knowledge of UEFA regulations and, ideally, (2)
basic knowledge of Systran’s functioning. Neither of these were required of the
subjects, which is probably the main reason for the discrepancies between the results
of TC1 and TC2.
If this finding were applied to the UEFA context, the scenario of UEFA
employees involved in the regulations core process drafting regulations according to
MT-related writing rules may yield a more acceptable MT output than the control
carried out by the two independent subjects since UEFA employees have broad
background knowledge of the text type at hand. Furthermore, linguistic errors could
be disregarded since the final version of regulations is always checked by the English
translators from SLAN. However, UEFA employees involved in the regulations core
process are generally neither familiar with translation, nor with Machine Translation,
entailing that regulations worded according to MT-related writing rules would not only
have to be checked according to the usual procedure, but also according to their
“MTranslatability” (Bernth and Gdaniec: 2001). In other words, persons “having MT in
mind” (Bernth and Gdaniec 2001: 176) would have to check regulations drafted
according to MT-related writing rules. The persons best suited for this task are
obviously translators being trained in CL and MT and who are alsofamiliar with the
software to be used. This scenario would have to be compared to the current
5. Results
99
translation workflow of UEFA regulations, which is based on machine-aided human
translation using SDL Trados Studio 2009 (see chapter 6.2)
Secondly, it is important to keep in mind the limits of the qualitative evaluation
carried out by the German translators from UEFA SLAN, which resulted from
imperfect test conditions due to the lack of clear instructions regarding the evaluation
criteria.
Based on the suggested improvements gathered through the survey of
translators with different professional backgrounds, it can be said that, when working
with a CL in combination with MT, certain stylistic shortfalls have to be accepted
because excessive PE would undo the advantages of HAMT and, in this case, justify
MAHT instead. It is therefore inevitable for any company, organisation or association
planning to introduce a CL in combination with MT to be clear on these losses of
language and style and to decide whether these are acceptable in light of the money
and time saved and the text types to be translated. As to some sequences in PE1
being equal to those of the official translation, the survey also revealed that the
regulations might be drafted in overly complex language and that it might therefore be
advisable to word them in a simpler way in order to ensure that their content is
correctly conveyed.
The following comment by one of the translators from the subject group that
evaluated the overall quality of the rapid post-edited version offers a fitting close to
this section:
“I have the impression that the use of a TM system consisting of previous regulations and a terminology database would produce better results without involving much more time.”
This critical remark will be the starting point for the final discussion below.
6. Discussion
100
6 Discussion
The aim of this final chapter is to draw an overall conclusion from my study. To
this end, I will first review the flaws of this pilot project and illustrate the measures
needed to eliminate these weaknesses in further experiments. I will then discuss the
necessary follow-up to the study to be carried out if the implementation of Systran is
seriously taken into consideration by UEFA. In the last section of this final chapter, I
will consider the results of my project and try to draw a more general conclusion
concerning the impact of controlled language in Machine Translation.
6.1 Critique of the study
The preceding chapters showed that the results of my study are not entirely
exhaustive due to certain limitations encountered during the evaluation process.
These limitations concern the control of S2 in order to double-check the MT-oriented
writing rules for UEFA regulations as well as the qualitative evaluation carried out by
the German translators from UEFA SLAN.
The results of the control of S2 cannot be considered as conclusive since
neither of the two subjects was familiar with UEFA regulations. Furthermore, both test
participants are professional translators, which also biased the outcome. In order to
yield definitive results, it would be necessary to repeat the double-checking phase
with UEFA employees involved in the regulations core process. The results of this
test would reveal to what extent the scenario of UEFA employees not being familiar
with translation or MT could satisfy expectations regarding the applicability of the
writing rules, and whether a re-control carried out by UEFA SLAN employees (ideally
MT-trained native English translators) would be necessary. Furthermore, the writing
rules are not meant to be applied to a given source text since this is a very time-
consuming task. They should ideally be used as a guideline for drafting new texts,
logically also involving the control of previously drafted paragraphs which do not have
to be revised. The applicability of my MT-oriented writing rules for UEFA regulations
would therefore have to be tested in situ, involving the employees which would have
to work with them should UEFA choose to work with MT. Such tests should take into
account the use of a so-called controlled language checker software. These tools
inform the author of expressions and structures that do not comply with the CL rules
in use. The market leader Acrolinx promotes its CL checker as being able to “cut
writing and editing time by as much as 75%”, thanks to its “real-time conformance
reporting” (Acrolinx website, see reference list). A definitive conclusion regarding the
applicability of my MT-related writing rules for UEFA regulations in a professional
6. Discussion
101
context could not be drawn prior to such in situ tests, and should take into account the
time and costs incurred by this additional step.
The qualitative evaluation carried out by the German translators from UEFA
SLAN cannot be regarded as perfectly valid either. The lack of clear instructions
regarding the evaluation criteria prevents me from drawing an unqualified conclusion
as to the quality of the MT output of the different text material. In order to obtain
meaningful and completely reliable results, this evaluation phase would have to be
carried out again, providing clear evaluation criteria as described in section 5.2.2.1.
Due to the scope of my study, these two steps could not be carried out again.
Having UEFA employees control a text excerpt should, in particular, be revisited if the
implementation of Systran is to be a serious option for UEFA SLAN.
6.2 Further steps
Throughout the study, I gathered positive results regarding the evaluation of
the MT output of the controlled text material. However, the “purchase of an MT
system is in many cases a costly affair and requires careful consideration” (Arnold et
al. 1994: 171). There are consequently further steps to be undertaken before
finalising such an important decision.
My study only considered the translation direction English-German, so the
next step would be to carry out the same quality assessment of the MT output for the
language pair English-French. If English-French MT output quality tests yield similarly
positive results – which is likely to be the case since it is generally agreed that the
translation direction English-French is the easiest – it will then be up to UEFA to pit
potential cost benefits against “a fall in quality of translation” (Arnold et al. 1994: 171).
As seen in chapter 5, some shortfalls regarding language and style have to be
accepted when working with MT since excessive post-editing would undo potential
economic benefits. UEFA will therefore have to assess whether MT-based
translations can be reconciled with SLAN’s task to provide the UEFA administration
with high-quality, image-enhancing linguistic support (see section 3.2). The end-users
of the translations should not be neglected in the evaluation process. The typical
readers of UEFA regulations are unlikely to be language experts, and it will therefore
be crucial to gather their views of the official human translation and the post-edited
MT output.
If UEFA generally approves the use of MT for the translation of regulations,
the next action to be taken would be a thorough operational evaluation. It should be
6. Discussion
102
carried out on site, comparing the costs incurred by an HAMT workflow including CL20
and PE to the costs generated in a MAHT workflow (Arnold et al. 1994: 170). This is a
very important step since “[n]o MT system will be used, however good its quality, if it
cannot compete in terms of cost with human translation” (Hutchins and Somers 1992:
171). As “an operational evaluation conducted by a user will be extremely expensive”
(Arnold et al. 1994: 171), it would be recommendable to “integrate the evaluation
process in the normal production process” (idem). Applied to UEFA, this means that
all the costs involved in the operational environment of a MAHT versus a HAMT
workflow (manpower, extra working time for MT-oriented writing of SL texts, software,
maintenance, etc.) would have to be compared to all the benefits generated by every
approach (speed, productivity, translation consistency, simultaneous output in several
languages, translator satisfaction, etc.), by recording the working time for individual
work steps (e.g. dictionary maintenance, alignments, revision of human translations,
post-editing of MT output). This would establish to what extent translating regulations
using Systran would be economically viable. It must be noted that the implementation
of any MT system does not yield immediate advantages since the system has to be
“tuned towards the texts it is supposed to deal with” (idem), as is the case with any
newly introduced TM software that is advantageous only once the TM contains a
sufficient amount of translation segments. Therefore the operational evaluation period
should ideally last several months given that “the efficiency of the system increases
with use, so initial cost estimates may be considerably misleading” (Hutchins and
Somers 1992: 172).
As translators are the direct users of MT systems, they should “be closely
involved in evaluations of systems for purchase and installation”, even though they
“may not make the final decision” (Hutchins and Somers 1992: 172). If the translators
consider that there will too often be “more work in revising than in translating from
scratch” (idem), they will obviously reject the implementation of MT. That said, my
qualitative evaluation has shown that none of the UEFA SLAN translators has general
prejudices against MT, which is a good basis for the implementation of a HAMT-
based translation workflow. In terms of MT acceptance, it would also be necessary to
ensure that stakeholders in the regulations core process do not reject writing
according to MT-oriented rules.
20
Ideally using a state-of-the-art CL checker such as Acrocheck , that includes a function of not displaying certain modifications of the SL text: the MT system translates from the controlled SL text version, but the controlled SL text for purpose of dissemination does not
contain certain pedestrian structures such as excessive use of articles or prepositions (see section 2.4)
6. Discussion
103
It is only after such an extensive operational evaluation that the final decision
for or against the implementation of MT into the SLAN workflow can be taken.
6.3 Conclusion
The concrete goal of this Master’s thesis was to find out if UEFA
regulations – after undergoing a control of the SL text – are suitable for MT using
Systran, since the quality of MT output is in any case the crucial factor for or against
the implementation of a HAMT-based workflow.
In order to answer the main question underlying my study, I first place laid the
theoretical foundation of MT, controlled languages and post-editing. I then provided
insight into the context of this study and gave an overview of UEFA, the UEFA
Language Services unit and UEFA regulations, discussing the reasons supporting
their suitability for MT. Next, I presented the individual steps of the study, including
the control of the source text material, the subsequent creation of MT-oriented writing
rules for UEFA regulations and finally the post-editing of the MT output.
With a view to testing the applicability of my controlled language rules, I also
undertook a double-checking phase by letting two independent subjects control a
second source text on the basis of these writing instructions. The test yielded
generally positive results as long as the writer or post-editor has some practical
experience with CL. As mentioned in previous sections, this step would have to be
repeated with UEFA employees involved in the regulations core process in order to
provide more reliable results. Furthermore, UEFA employees would probably need
some thorough introduction to the topic before drafting regulations according to CL
rules.
The discussion of the post-editing process then revealed that the additional
number of interventions and the extra time spent between full and rapid PE of the
target text material translated from the controlled source text did not differ to a great
extent. It would however be necessary to carry out further analyses on a greater text
volume in order to draw an unqualified conclusion.
In order to prove my hypothesis that the use of CL would significantly improve
the MT output of the chosen text material, the evaluation part of the study focused on
comparing the MT output of controlled source text excerpts to the MT output of
uncontrolled excerpts. The evaluations provided very conclusive results,
corroborating my assumption that CL would improve the MT output of the text
material. Both the quantitative and qualitative evaluations carried out by different
subject groups yielded generally positive results. The quantitative evaluation using
SDL Trados Studio 2009 empirically supported my hypothesis, although the
6. Discussion
104
discrepancies encountered between the results relating to the two source texts hinted
at the aforementioned limits impeding the pilot project from being entirely valid and
noun. The results of the questionnaires filled out by the four German translators from
UEFA SLAN also supported my hypothesis, although these were also qualified due to
imperfect test conditions. The feedback gathered from the questionnaires answered
by a subject group made up of eight translators with different professional
backgrounds was similarly positive, revealing that the rapid post-edited MT output of
the controlled source text was generally regarded as meeting professional
requirements. Furthermore, the controlled source text material was generally
approved by an English translator from UEFA SLAN who confirmed that it was
stylistically imperfect, but still met professional requirements by accurately
transferring information. A further positive result of the study is the finding that most of
the subjects taking part in the evaluation were not generally prejudiced towards MT.
However, the favourable overall conclusion drawn from the pilot project has to be
qualified due to the aforementioned limitations, namely the applicability test of my MT-
oriented writing rules for UEFA regulations and the results obtained from the
qualitative evaluation carried out by the German translators from UEFA SLAN.
To sum up the results of this pilot project, it can definitely be stated that: (1)
the readability and intelligibility of the main text material was improved by the use of
CL, (2) the use of CL significantly improved the MT output of the chosen text material,
and therefore (3) UEFA regulations – after undergoing a control of the SL text – are
suitable for MT using Systran. With the aim of setting this study against a broader
framework in order to deduce more general findings from it, I must say that its limited
scope and the limits encountered do not allow me to definitively answer the question
as to whether the use of CL does generally improve the MT output of suitable text
material. However, my results underscore the general consensus in MT research that
controlled languages reduce the amount of time needed for post-editing. My study
results also underlined that MT-based workflows are not likely to dispense with
human intervention in the near future, since CL and PE are very important in the
automatic translation process21. “[T]he future for professional translators [therefore]
seems to be secure. [They] may see these changes as threatening, or as a challenge;
their attitude may decide how successful they are in making the transition” (Koby
2001: 14). In the UEFA context, the translators’ attitude is certainly the most decisive
factor for the success of a possible implementation of Systran, since they would be
the main stakeholders of a newly introduced MT-based translation workflow. The
21
Except for very narrow sub-domains, or if the use of a very restricted CL (e.g. Simplified English) forgoes PE.
6. Discussion
105
translators should therefore absolutely be included in the decision-making process for
or against the implementation of MT in the Regulations Core Process.
7. Reference list
106
7 Reference list
Primary literature
UEFA. 2010. “UEFA Regulations of the UEFA European Football Championship
2010-12”.
UEFA. 2010. “Reglement der UEFA-Fussball-Europameisterschaft 2010-12“.
UEFA. 2011. “UEFA regulations of the UEFA Champions League 2011/12”.
UEFA. 2011. “Reglement der UEFA Champions League 2010/11”.
Secondary literature
Allen, Jeffrey. 2003. “Post-editing.” In Computers and Translation: A Translator’s
Guide, ed. Harold Somers. Benjamins Translation Library 35. Amsterdam ;
Philadelphia: J. Benjamins.
Arnold, Doug. 2003. “Why Translation Is Difficult for Computers.” In Computers
and Translation: a Translator’s Guide, pp.119–142. Benjamins Translation Library
35. Amsterdam ; Philadelphia: J. Benjamins.
Arnold, Doug, Lorna Balkan, Siety Meijer, R. Lee Humphreys, and Louisa Sadler.
1994. Machine Translation: An Introductory Guide. Manchester ; Oxford ;
Cambridge: NCC Blackwell.
Austermühl, Frank. 2001. Übersetzen Im Informationszeitalter: Überlegungen Zur
Zukunft Fachkommunikativen Und Interkulturellen Handelns Im “Global Village”.
Trier: WVT.
Bernth, Arendse, and Claudia Gdaniec. 2001. “MTranslatability.” Machine
Translation 16 (3): pp.175–218.
Bowker, Lynne. 2002. Computer-aided Translation Technology: a Practical
Introduction. Didactics of Translation Series. Ottawa: University of Ottawa Press.
Coughlin, Deborah. 2003. “Correlating Automated and Human Assessments of
Machine Translation Quality.” Proceedings of MT Summit IX: 63–70.
Estrella, Paula. 2008. Evaluating Machine Translation in Context: Metrics and
Tools. Doctoral thesis submitted at the École de traduction et interprétation,
Genève.
7. Reference list
107
Furlani, Donna. 2009. Translation Technologies and Medical Translation. Master
Thesis submitted at the École de Traduction et d’Interprétation, Genève.
Hajič, Jan, Petr Homola, and Vladislav Kuboň. 2003. “A Simple Multilingual
Machine Translation System.” Proceedings of MT Summit IX: pp.157–164.
Huijsen, Willem-Olaf. 1998. “Controlled Language - An Introduction.” Proceedings
of CLAW 98: pp.1–15.
Hutchins, John. 2001. “Machine Translation and Human Translation: In
Competition or in Complementation?” International Journal of Translation 13. 1-2,
Jan-Dec: 5–20.
Hutchins, John. 2003. “Commercial Systems: The State of the Art.” In Computers
and Translation: a Translator’s Guide, pp.161–174. Benjamins Translation Library
35. Amsterdam ; Philadelphia: J. Benjamins.
Hutchins, John, and Harold Somers. 1992. An Introduction to Machine Translation.
London ; San Diego [etc.]: Academic Press.
Hutchins, William John. 1986. Machine Translation: Past, Present, Future. Ellis
Horwood Series in Computers and Their Applications. Chichester : New York [etc.]:
E. Horwood ; Halsted press of J. Wiley.
Kay, Martin, Jean Mark Gawron, and Peter Norvig. 1994. Verbmobil: A Translation
System for Face-to-Face Dialog. CSLI. Lectures Notes no 33. Standford: Center
for the Study of Language and Information.
King, Margaret. 1997. “Evaluating Translation.” In Machine Translation and
Translation Theory, ed. Christa Hauenschild and Susanne Heizmann, 251–264.
Berlin: Mouton de Gruyter.
Koby, Geoffrey S. 2001. “Editor’s Introduction - Post-Editing of Machine
Translation Output: Who, What, Why, and How (Much).” In Repairing Texts:
Empirical Investigations of Machine Translation Post-Editing Processes, ed.
Geoffrey S. Koby. Translation Studies 5. Kent Ohio: The Kent State Univ. Press.
Koehn, Philipp, and Kevin Knight. 2009. “Statistical Machine Translation.”
Lange, Andrés C. 1998. “Tying the Knot: How Baan Wed Machine Translation to
Translation Memory - and Survived the Honeymoon.” Language International 10:5:
34–36.
7. Reference list
108
Melby, Alan, and Terry Warner. 1995. The Possibility of Language: A Discussion of
the Nature of Language, with Implications for Human and Machine Translation.
John Benjamins Publishing Company.
Mitamura, Teruko. 1999. “Controlled Language for Multilingual Machine
Translation.” Proceedings of MT Summit VII: pp.46–52.
Mitamura, Teruko et al. 2003. “Source Language Diagnostics for MT.” Proceedings
of MT Summit IX: pp.254–260.
Mossop, Brian. 2006. “Has Computerization Changed Translation?” Meta 51/4:
pp.787–739.
Mukesh, G.S. Vatsa, Nikita Joshi, and Sumit Goswami. 2010. “Statistical Machine
Translation.” DESIDOC Journal of Library & Information Technology Vol. 30, No. 4:
pp. 25–32.
Newton, John. 1992. Computers in Translation: a Practical Appraisal. Psychology
Press.
Nunes Vieira, Lucas, and Lucia Specia. 2011. “A Review of Translation Tools from
a Post-Editing Perspective.” Proceedings of the Third Joint EM+/CNGL Workshop
“Bringing MT to the User: Research Meets Translators”: 33–42.
Nyberg, Eric, Teruko Mitamura, and Willem-Olaf Huijsen. 2003. “Controlled
Language for Authoring and Translation.” In Computers and Translation: A
Translator’s Guide, ed. Harold Somers, 246–282. Benjamins Translation Library
35. Amsterdam ; Philadelphia: J. Benjamins.
O’Brien, Sharon. 2003. “Controlling Controlled English: An Analysis of Several
Controlled Language Rule Sets.” Proceedings of EAMT/CLAW 2003: pp.105–114.
De Preux, Nathalie. 2005. “How Much Does Using Controlled Language Improve
Machine Translation Results?” Translating and the Computer 27: 14pp.
Pulitano, Donatella. 2010. University of Geneva.
Quah, Chiew Kin. 2006. Translation and Technology. Palgrave Textbooks in
Translating and Interpreting. Basingstoke: Palgrave.
7. Reference list
109
Ramlow, Markus. 2008. “Maschinelle Übersetzungssysteme Im Vergleich.” In
Maschinelle Übersetzung Und XML Im Übersetzungsprozess: Prozesse Der
Translation Und Lokalisierung Im Wandel, ed. Uta Seewald. TRANSÜD : Arbeiten
Zur Theorie Und Praxis Des Übersetzens Und Dolmetschens 19. Berlin: Frank &
Timme.
Roturier, Johann. 2004. “Assessing the Set of Controlled Language Rules: Can
They Improve the Performance of Commercial Machine Translation Systems?”
Proceedings of the 26th ASLIB Conference, Translating and the Computer: 14pp.
Sager, Juan C. 1994. Language Engineering and Translation: Consequences of
Automation. John Benjamins Pub Co.
Schäfer, Falko. 2002. Die Maschinelle Übersetzung Von Wirtschaftsfachtexten:
Eine Evaluierung Anhand Des MÜ-Systems Der EU-Kommission, SYSTRAN, Im
Sprachenpaar Französisch-Deutsch. European University Studies. Series 21,
Linguistics 251. Frankfurt am Main [etc.]: P. Lang.
Schwarzl, Anja. 2001. The (im)possibilities of Machine Translation. European
University Studies: Ser. 14, Anglo-Saxon Language and Literature; Vol. 381 Bd.
381. Frankfurt a.M. ; Bern [etc.]: P. Lang.
Senellart, Jean, Péter Dienes, and Tamás Váradi. 2001. “New Generation Systran
Translation System.” MT Summit VIII: Machine Translation in the Information Age,
Proceedings, Santiago De Compostela, Spain, 18-22 September 2001: 311–316.
Somers, Harold, ed. 2003. Computers and Translation: A Translator’s Guide.
Benjamins Translation Library 35. Amsterdam ; Philadelphia: J. Benjamins.
Thoden, Annika. 2010. Wortspiele Im Übersetzungsvergleich Am Beispiel Der
Comicreihe Asterix. 1. ed. Grin Verlag.
Trujillo, Arturo. 1999. Translation Engines: Techniques for Machine Translation.
Applied Computing. London [etc.]: Springer.
UEFA, 2010. “UEFA Legal Framework: Regulations Core Process and Legal
Principles Applicable to the Creation of UEFA’s Rules and Regulations”. Nyon:
UEFA.
UEFA, 2011. “UEFA-Stilführer Deutsch, Ausgabe 2011”. Nyon: UEFA.
7. Reference list
110
Vasconcellos, Muriel. 1987. “A Comparison of MT Postediting and Traditional
Revision.” In Proceedings of the 28th Annual Conference of the American
Translators Association, ed. K. Kummer, 409–415. Medford, NJ: Learned
Information.
Wagner, Emma. 1985. “Post-Editing Systran- A Challenge for Commission
Translators.” Terminologie Et Traduction 3: 1–7.
White, John S. 2003. “How to Evaluate Machine Translation.” In Computers and
Translation: A Translator’s Guide, ed. Harold Somers, 211–244. Benjamins
Translation Library. Amsterdam ; Philadelphia: J. Benjamins.
Digital resources
Hornby, A.S. 2010. Oxford Advanced Learner’s Dictionary of Current English. 8.
ed. Oxford: Oxford University Press.
UEFA website: http://www.uefa.com/uefa/aboutuefa/organisation/history/index.html
(22/02/2012).
Online resources
Collins English Dictionary. http://www.collinsdictionary.com/ (22/05/2012).
Duden. http://www.duden.de/ (22/05/2012).
Simplified Technical English. http://www.asd-ste100.org/ (04/06/2012).
Systran V6 User Guide.
http://download.smartlinkcorp.com/Smart_Link_download//st20109.pdf
(04/06/2012).
UEFA, 2010. “UEFA Statutes.”
http://www.uefa.com/MultimediaFiles/Download/Regulations/uefaorg/General/01/4
7/69/97/1476997_DOWNLOAD.pdf (19/08/2012).
7. Reference list
111
Commercialised tools
Acrolinx: http://www.acrolinx.com/s1000d.html (29/06/2012).
Across: http://www.across.net/en/index.aspx (17/08/2012).
Bing: http://www.microsofttranslator.com/(23/05/2012).
Déjà Vu: http://www.atril.com/ (17/08/2012).
Google Translate: http://translate.google.com/ (23/05/2012).
MAXit: http://www.smartny.com/ (18/08/2012).
MultiCorpora: http://www.multicorpora.com/ (17/08/2012).
SDL Trados: http://www.Trados.com (17/08/2012).
Systran: http://www.systran.co.uk/, (17/08/2012).
Transit: http://www.star-group.net/ENU/group-transit-nxt/transit.html (17/08/2012).
Winalign: http://www.Trados.com (17/08/2012).
Wordfast: http://www.wordfast.net/ (17/08/2012).
7. Reference list
112
Participants of the evaluations
UEFA SLAN, English translation
Catherine Wilson
UEFA SLAN, German translation
Florian Simmen
Doris Egger
Sandra Wisniewski
Frédéric Wyler
Other translators
Anna Leemann (Université de Genève)
Janine Niederberger (Université de Genève)
Kristin Zimmermann (Universität Mainz)
Mireille Tschannen (Université de Genève)
Rafaela Müller (Universität Leipzig)
Sandra Scheuren (Université de Genève)
Viola Trapper (Universität Heidelberg)
Vivien Berg (Université de Genève)
8. List of figures
113
8 List of figures
Fig. 1: “Human and machine translation” (Hutchins and Somers 1992: 148) .......... 5
Fig. 2: “Human-aided machine translation model” (Quah 2006: 12) ........................ 6
Fig. 3: “Machine-aided human translation model” (Quah 2006: 13) ......................... 7
Fig. 4: Adapted from “Chronology of machine translation development”
(Quah 2006: 58) ............................................................................................. 8
Fig. 5: adapted from “Machine translation architectures” (Quah 2006: 68) ............ 12
Fig. 6: “Direct MT system” (Hutchins and Somers 1992: 72).................................. 12
Fig. 7: “Interlingua model with six language pairs”
(Hutchins and Somers 1992: 74) ................................................................. 14
Fig. 8: “Transfer model with two language pairs”
(Hutchins and Somers 1992: 75) ................................................................. 14
Fig. 9: Overview of UEFA divisions ......................................................................... 31
Fig. 10: Overview of the phases of the regulations core process, .......................... 35
Fig. 11: Systran IntuitiveCoding Technology........................................................... 42
Fig. 12: Excerpt of Systran’s Coding Reference Table for English......................... 42
Fig. 13: Systran word categories ............................................................................. 43
Fig. 14: IntuitiveCoding of “sich beziehen auf ......................................................... 43
Fig. 15: Systran ExpertCoding................................................................................. 43
Fig. 16: Systran dictionary entry priority .................................................................. 43
Fig. 17: Systran UEFA TM ....................................................................................... 44
Fig. 18: Systran UD/TM priorites ............................................................................. 44
Fig. 19: Systran dictionary priorities ........................................................................ 44
Fig. 20: Systran Translation Project Manager Interface.......................................... 45
Fig. 21: Number of rules applied for the control of S1 ............................................ 46
Fig. 22: Most applied rules for the control of S1...................................................... 46
Fig. 23: Comparison of the number of interventions carried out............................. 49
Fig. 24: Most applied rules for the control of S1...................................................... 49
Fig. 25: Subject 1 – Most applied rules for the control of S2/1 ............................... 49
Fig. 26: Subject 2 – Most applied rules for the control of S2/2 ............................... 50
Fig. 27: Comparison of errors committed by subject 1 and subject 2..................... 50
Fig. 28: Comparison of the number of interventions PE1 and PE2 ........................ 55
Fig. 29: PE1 – interventions according to categories.............................................. 56
Fig. 30: PE2 – interventions..................................................................................... 56
Fig. 31: Comparison in minutes between PE1 and PE2 ......................................... 56
Fig. 32: PE1 – Comparison in minutes .................................................................... 57
Fig. 33: PE2 – Comparison in minutes according to categories ............................. 57
8. List of figures
114
Fig. 34: Translation editor bug ................................................................................. 58
Fig. 35: Dictionary bug (1) ....................................................................................... 58
Fig. 36: Dictionary bug (2) ....................................................................................... 58
Fig. 37: Dictionary entries for the verb ‘play’ ........................................................... 67
Fig. 38: Lexical gap .................................................................................................. 68
Fig. 39: Different systems of singular and plural use .............................................. 69
Fig. 40: Comparison Trados analyses of TC1 and NCT1 ....................................... 78
Fig. 41: Comparison Trados analysis of TC2 and NCT2 ........................................ 79
Fig. 42: Comparison Trados analysis of TC1 and TC2 ........................................... 80
Fig. 43: Comparison Trados analysis of NCT1 and NCT2 ..................................... 80
Fig. 44: Results of the qualitative evaluation, sentence sets 1 and 2..................... 88
Fig. 45: Sentence set 1 ............................................................................................ 90
Fig. 46: Sentence set 2 ............................................................................................ 90
9 List of tables
Table 1: Added and amended rules ............................................................................. 52
Table 2: Most common PE interventions for PE1 and PE2 ......................................... 56
Table 3: Trados analysis of TC1 using TM1 ................................................................ 77
Table 4: Trados analysis of TC1 using TM2 ................................................................ 77
Table 5: Trados analysis of NCT1 using TM1 .............................................................. 78
Table 6: Trados analysis of NCT1 using TM2 .............................................................. 78
Table 7: Trados analysis of TC2 using TM3 ................................................................ 79
Table 8: Trados analysis of NCT2 using TM3 .............................................................. 79
Table 9: Overview of sentence sets ............................................................................. 84
Table 10: Trados analysis of HTS1 using TM1 ............................................................ 87
Table 11: Trados analysis of HTS1 using TM2 ............................................................ 87
9. Annexes
115
9 Annexes
Annex A: Overview of UEFA divisions and units
Division Unit
CEO Office
Division Management
Communications Digital Media
Division Management
Media Relations
Online & Publishing
TV & Video Production
Competitions Club Competitions
Competitions' Planning & Services
Division Management
Football Content & Legacy
Match Operations
National Team Competitions
Refereeing
Women's & Futsal Competitions
Finance Accounting
Controlling
Division Management
Legal Affairs Business Affairs
Disciplinary Services
Division Management
Event & Corporate Legal Services
Marketing Legal Services
Regulatory Affairs
Sports Legal Services
Marketing Broadcaster Servicing Division Management
Global Sponsor Management
Marketing Activities Media Rights & Services
Sponsorship Sales & Licensing
9. Annexes
116
Division Unit
National Associations
Club Licensing & Financial Fair Play
Division Management
Football Education Services
Football Social Responsibility
National Associations Development
National Associations Services
Stadia & Security
Operations Division Management
Euro 2012 Commercial
Euro 2012 Event Management
Euro 2012 Operations
Euro 2012 Planning. Ticketing & Admission Services
Government Relations & Public Services
Logistics
UEFA Events
President's & Executive Office
Division Management
Executive Office
President's Office
Services Division Management
Facility Management
Human Resources
Information & Communication Technology
Language Services
Travel & Conferences
9. Annexes
117
Annex B: Overview of UEFA Events SA
Marketing Operations
Marketing & Strategic Planning
TV and Digital Media Rights & Services
Sponsorship Sales & Account Management
Licensing & Merchandising
Business Development
Corporate and Competitions Brand & Promotion
Research & Market Intelligence
Agency Management
Interdependent
Collaboration
Tournament Planning & Programme Management
Delivery of UEFA EURO 2012 in conjunction with the Polish & Ukrainian LOCs
UEFA EURO 2012 Commercial & Hospitality Sales
Tournament Venue Delivery
Marketing Rights Delivery
Bidding & Future Host Management
Other Event Services
9. Annexes
118
Annex C: S1
III Trophy, Plaques and Medals Article 3 Trophy
3.01 The original trophy, which is used for the official presentation ceremony at the final and at other official events approved by UEFA, remains in UEFA’s keeping and ownership at all times. A full-size replica trophy, the UEFA European Football Championship winners trophy, is awarded to the winning association. 3.02 Any association which wins the trophy three consecutive times or five times in total receives a special mark of recognition. Once a cycle of three successive wins or five in total has been completed, the association concerned starts a new cycle from zero. 3.03 Replica trophies awarded to winners of the UEFA European Football Championship (past and current) must remain within the relevant association’s control at all times and must not leave the association’s country without UEFA’s prior written consent. Associations must not permit a replica trophy to be used in any context where a third party (including, without limitation, associations’ sponsors and other commercial partners) is granted visibility or in any other way which could lead to an association between any third party and the replica trophy and/or the competition. Associations must comply with any trophy use guidelines that the UEFA administration may issue from time to time. 3.04 Associations may not, and may not permit any third party to, develop, create, use, sell or distribute any promotional materials or merchandise bearing any representation of the trophy or any replica thereof (including, without limitation, trophy lift images) or use any such representation in a manner that could lead to an association between any third party and the trophy, replica trophy and/or the competition.
Commemorative plaque
3.05 Each association that competes in the final tournament receives a commemorative plaque.
Semi-finalist plaque 3.06 Each defeated semi-finalist receives a plaque.
Final plaque
3.07 The finalists receive a plaque.
Medals 3.08 Forty gold medals are presented to the winning team and 40 silver medals to the runner-up. 40 bronze medals are presented to the defeated semi-finalists. Additional medals may not be produced.
IV Responsibilities Article 4
Responsibilities of the associations 4.01 The associations are responsible for the behaviour of their players, officials, members, supporters and anyone carrying out a function at a match on their behalf. 4.02 The association in the territory of which a qualifying match or the final tournament is being staged is considered the host association.
9. Annexes
119
4.03 The host association is responsible for order and security before, during and after the match. The host association may be called to account for incidents of any kind and may be disciplined. 4.04 Matches must, in principle, be played in a stadium within the territory of the host association. Exceptionally, matches may be played in the territory of another UEFA member association, if so decided by the UEFA administration and/or the disciplinary bodies, for reasons of safety or as a result of a disciplinary measure. Additional responsibilities for the final tournament 4.05 The UEFA administration informs the associations participating in the final tournament about any further guidelines, directives or decisions related to the final tournament and provides them with all relevant documents in due time.
VI Competition System Article 6
Competition stages
6.01 The competition consists of a qualifying competition and a final tournament.
Article 7 A. Qualifying competition
Group formation
7.01 The teams of the host associations for the final tournament, Poland and Ukraine, qualify automatically for the final tournament. The remaining teams are drawn into 6 groups of 6 teams and 3 groups of 5 teams. The groups are formed by the UEFA administration by means of a draw, with seeded teams, on completion of the 2010 FIFA World Cup qualifying competition. The reigning European champions are always seeded. The other associations are classified on the basis of the UEFA national team coefficient ranking system (see Annex I, paragraph 1.2.1). The decisions of the UEFA administration are final. 7.02 If any of the teams in question have the same coefficient, the following criteria, relating only to the most recent qualifying competition, are determinant, in the order given: a) UEFA national team coefficient from the matches played; b) average goal difference; c) average number of goals scored; d) average number of away goals scored; e) fair play ranking; f) drawing of lots.
Match system for the qualifying competition
7.03 The matches in the qualifying competition are played in groups according to the league system, with each team playing all opponents in its group in a series of home and away matches. Three points are awarded for a win, one point for a draw, and no points for a defeat. Equality of points after the group matches 7.04 If two or more teams are equal on points on completion of the group matches, the following criteria are applied, in the order given, to determine the rankings: a) higher number of points obtained in the group matches played among the teams in question; b) superior goal difference from the group matches played among the teams in question; c) higher number of goals scored in the group matches played among the teams in question;
9. Annexes
120
d) higher number of goals scored away from home in the group matches played among the teams in question; e) if, after applying criteria a) to d) to several teams, two or more teams still have an equal ranking, the criteria a) to d) will be reapplied to determine the ranking of these teams. If this procedure does not lead to a decision, criteria f) to j) apply; f) superior goal difference in all group matches; g) higher number of goals scored in all group matches; h) higher number of goals scored away from home in all group matches; i) fair play ranking in all group matches; j) drawing of lots.
Qualification for the final tournament
7.05 The nine group winners and the best runner-up qualify directly for the final tournament. 7.06 To determine the best runner-up, only results against the teams in first, third, fourth and fifth place are taken into account, with the following criteria being taken into consideration in the order given: a) higher number of points obtained in these matches; b) superior goal difference from these matches; c) higher number of goals scored in these matches; d) higher number of away goals scored in these matches; e) position in the UEFA national team coefficient ranking system (see Annex I, paragraph 1.2.2); f) fair play ranking in these matches; g) drawing of lots. 7.07 The eight remaining runners-up contest play-off matches. The four ties are determined by means of a draw. The four runners-up with the best position in the UEFA national team coefficient ranking system (see Annex I, paragraph 1.2.2) are seeded for the draw. If any of the teams in question have the same coefficient, the criteria listed in paragraph 7.02 apply. The play-offs are played according to the knockout system, with each team playing each opponent twice, in home and away matches. The seeded teams play the return match at home. The teams which score the greater aggregate of goals in the two matches qualify for the final tournament. Otherwise, the provisions of paragraph 7.08 apply. 7.08 For matches played under the knockout system, if the two teams involved in a tie score the same number of goals over the two legs, the team which scores more away goals qualifies for the next stage. If this procedure does not produce a result, i.e. if the two teams score the same number of home and away goals, two 15-minute periods of extra time are played at the end of the second leg. If, during extra time, both teams score the same number of goals, away goals count double (i.e. the visiting team qualifies). If no goals are scored during extra time, kicks from the penalty mark (Article 16) determine which team qualifies for the next stage.
Article 8
B. Final tournament
8.01 The UEFA Executive Committee has entrusted the Polish Football Association (PZPN) and the Football Federation of Ukraine (FFU) with the joint organisation and staging of the final tournament.
Group formation
8.02 The UEFA administration divides the 16 teams that qualify for the final tournament into four groups (A, B, C, D) of four teams each. 8.03 By means of a draw. the four groups are formed as follows:
9. Annexes
121
Group A Group B Group C Group D A1 B1 C1 D1 A2 B2 C2 D2 A3 B3 C3 D3 A4 B4 C4 D4
Coefficients
8.04 The seeded teams are the host associations, with the reigning European champions if they qualify, and one or two teams with the best coefficients according to the UEFA national team coefficient ranking system (see Annex I, paragraph 1.2.2). The other finalists are assigned to the four groups by means of a draw, according to their coefficients. 8.05 If any of the teams in question have the same coefficient, the criteria listed in paragraph 7.02 apply. Results from the play-offs referred to in paragraph 7.07 are thereby not taken into consideration. Group match schedule 8.06 Each team plays each of the other teams in the same group according to a league system (three points for a win, one point for a draw, no points for a defeat). The group matches are played according to the following schedule. The last two matches in each group must both kick off at the same time. The first-named team is considered as the home team. Match day 1 Match day 2 Match day 3 Group A A1 vs A2 A1 vs A3 A4 vs A1 A3 vs A4 A2 vs A4 A2 vs A3 Group B B1 vs B2 B1 vs B3 B4 vs B1 B3 vs B4 B2 vs B4 B2 vs B3 Group C C1 vs C2 C1 vs C3 C4 vs C1 C3 vs C4 C2 vs C4 C2 vs C3 Group D D1 vs D2 D1 vs D3 D4 vs D1 D3 vs D4 D2 vs D4 D2 vs D3
Equality of points after the group matches
8.07 If two or more teams are equal on points on completion of the group matches, the following criteria are applied, in the order given; to determine the rankings: a) higher number of points obtained in the matches among the teams in question; b) superior goal difference in the matches among the teams in question (if more than two teams finish equal on points); c) higher number of goals scored in the matches among the teams in question (if more than two teams finish equal on points); d) superior goal difference in all the group matches; e) higher number of goals scored in all the group matches; f) position in the UEFA national team coefficient ranking system (see Annex I, paragraph 1.2.2); g) fair play conduct of the teams (final tournament); h) drawing of lots.
9. Annexes
122
8.08 If two teams which have the same number of points, the same number of goals scored and conceded play their last group match against each other and are still equal at the end of that match, the ranking of the two teams in question is determined by kicks from the penalty mark (Article 16), provided no other teams within the group have the same number of points on completion of all group matches. Should more than two teams have the same number of points, the criteria listed under paragraph 8.07 apply.
Quarter-finals
8.09 The winners and runners-up in each group play the quarter-finals over one match, as follows: Match 1 Winner Group A vs Runner-up Group B Match 2 Winner Group B vs Runner-up Group A Match 3 Winner Group C vs Runner-up Group D Match 4 Winner Group D vs Runner-up Group C
Semi-finals
8.10 The four winners of the quarter-finals play the semi-finals over one match, as follows: Winner Match 1 vs Winner Match 3 Winner Match 2 vs Winner Match 4
Final 8.11 The winners of the semi-finals play in the final. Same number of goals in a quarter-final, semi-final or the final 8.12 If the result stands as a draw at the end of normal playing time, extra time of two periods of 15 minutes is played. If the two teams are still equal after extra time, the winners are determined by kicks from the penalty mark (Article 16).
VIII Match Dates. Venues and Kick-off Times Article 11
11.01 The competition will be staged over the two seasons that follow the 2010 FIFA World Cup.
A. Qualifying competition Match dates
11.02 The following twelve dates are reserved for the group matches in the qualifying competition of the 2010-12 UEFA European Football Championship: 2010 a) 3/4 and 7 September 2010 b) 8/9 and 12 October 2010 2011 c) 25/26 and 29 March 2011 d) 3/4 and 7 June 2011 e) 2/3 and 6 September 2011 f) 7/8 and 11 October 2011 (dates for the last matches in all groups) 11.03 The following two dates are reserved for the play-off matches between the remaining eight runners-up: a) 11/12 November 2011 b) 15 November 2011
9. Annexes
123
11.04 Matches as referred to in paragraph 11.02 may be played on other dates if both associations concerned are in agreement and the principles governing the release of players for association teams as laid out in Annex 1, Article 1 of the FIFA Regulations for the Status and Transfer of Players are adhered to. The associations in a group have 30 days after the draw in which to agree on the order of their matches. The exact date of each match must be specified (e.g. Saturday 4 September 2010). If the associations in question are unable to reach agreement, the matches will take place according to a standard fixture list drawn up by the UEFA administration. This standard fixture list will take into consideration relevant factors for the match organisation, such as weather conditions. 11.05 For reasons of sporting fairness, the UEFA administration is entitled to order matches within a group to be played at the same time. Any subsequent changes of date are subject to the approval of the UEFA administration. The host association of a match must, in this case, also inform the other associations in the group.
.
Venues and kick-off times
11.06 The venues of the matches are fixed by the host associations and announced to their opponents and the UEFA administration at least 60 days in advance. When fixing a venue, the host association must take into account the length of the journey to be undertaken by the visiting association. Unless the visiting association agrees otherwise, the venue for a qualifying match must be no more than a two-hour bus drive from the nearest international airport. The kick-off times must also be announced to the UEFA administration at least 60 days before the match.
Arrival of the teams at the match venue
11.07 The associations must arrange for their teams to arrive at the match venue early enough in order to be able to hold their pre-match press conference before the media deadlines of both involved countries and in any case no later than 24 hours before kick-off.
B. Final tournament Match dates
11.08 The final tournament will take place from 8 June to 1 July 2012. Venues and kick-off times 11.09 The UEFA administration is responsible for drawing up the match schedule for the final tournament. Each team must have at least 48 rest hours between each match played. Arrival of the teams in host countries 11.10 Each association taking part in the final tournament must arrive at their team hotel in one of the host countries at least five days before their first match. Arrival of the teams at the match venue 11.11 The teams must arrive at their transfer hotel or be within a 60km radius of the stadium where their match is to be played no later than 24 hours before kick-off.
Training grounds
11.12 UEFA offers each association a number of pre-selected training grounds. Should an association choose a training ground other than those which have been pre-selected, then the association concerned bears all the costs incurred. 11.13 In any case, any and all training grounds used by associations are referred to as “official” from five days before the first match of the final tournament and the
provisions laid out in paragraph 28.14 apply.
9. Annexes
124
Annex D: S2
III Trophies and Medals Article 5 Trophy
The original trophy, which is used for the official presentation ceremony at the final, remains in UEFA's keeping at all times. A full-size replica trophy, the UEFA Champions League winners trophy, is awarded to the winning club. Any club which wins the competition three consecutive times or five times in total receives a special mark of recognition. Once a cycle of three successive wins or five in total has been completed, the club concerned starts a new cycle from zero. Replica trophies awarded to winners of the UEFA Champions League (past and current) must remain within the relevant club’s control at all times and must not leave its region or the country of its association without UEFA’s prior written consent. Clubs must not permit a replica trophy to be used in any context where a third party (including, without limitation, their sponsors and other commercial partners) is granted visibility or in any other way which could lead to an association between any third party and the trophy and/or the competition. Clubs must comply with any trophy use guidelines that may be issued by the UEFA administration from time to time.
Medals
Thirty gold medals are presented to the winning club, and 30 silver medals to the runner-up. Additional medals may not be produced.
IV Responsibilities Article 6
UEFA responsibilities
UEFA insures its own area of responsibility in accordance with the present regulations: third-party liability insurance spectator accident insurance (for the final only) group accident insurance for UEFA delegates legal expenses insurance (restricted to criminal matters).
Responsibilities of the associations and clubs
The clubs are responsible for the behaviour of their players, officials, members, supporters and any person carrying out a function at a match on their behalf. The home club (or the host association) is responsible for order and security before, during and after the match. The home club (or the host association) may be called to account for incidents of any kind and may be disciplined. The club considered the home club must stage the relevant matches at the ground in accordance with the instructions of UEFA (or of a third party acting on UEFA’s behalf) and in cooperation with the association concerned. However, the club is considered solely accountable for all of its obligations in this respect, unless the relevant body or bodies decide(s) otherwise. Irrespective of UEFA’s insurance coverage, each club and host association must conclude insurance coverage with reputable insurers at their own cost, in relation to any and all risks, according to the following principles: each club must conclude and maintain insurance coverage to fully cover all of its risks in connection with its participation in the competition; in addition, the home club or the host association must conclude and maintain insurance coverage for the risks in connection with staging and organising its home matches and which must include, without limitation, third-party liability insurance (for all third parties participating in matches or attending the relevant
9. Annexes
125
venue) providing for appropriate guaranteed sums for damages to persons, objects and property, as well as for pure economic losses corresponding to the specific circumstances of the club or association concerned; to the same extent as in paragraph b) above, the host association of the final match must conclude and maintain insurance coverage to fully cover all of its risks in connection with the staging and organisation of the final match; if the home club or the host association is not the owner of the stadium used, it is also responsible for providing adequate and fully comprehensive insurance cover, including third-party liability and property damage, taken out by the relevant stadium owner and/or tenant; the home club and the host association must ensure that UEFA is included in all insurance policies as defined in the present paragraph and must hold UEFA harmless from any and all claims for liability arising in relation to the staging and organising of the relevant matches; In any case, UEFA may ask anyone involved to provide, free of charge, written releases of liability and/or hold harmless notes, and/or confirmations and/or copies of the policies concerned in one of UEFA’s official languages. The clubs undertake that their team will arrive at the match venue by the evening before the match at the latest and to fulfil their media obligations the day before the match. Visiting clubs undertake not to play any other matches when travelling to and from away matches in this competition.
V Competition System Article 7
Number of rounds
As shown in Annex Ib. the competition consists of: the qualifying phase for the UEFA Champions League: first qualifying round second qualifying round third qualifying round play-offs the UEFA Champions League: group stage (six matchdays) round of 16 quarter-finals semi-finals final.
Qualifying phase
Qualifying-phase matches are played according to the cup (knockout) system, with each club playing each opponent twice, in home and away matches. The team which scores the greater aggregate of goals in the two matches qualifies for the next stage (second qualifying round, third qualifying round or play-offs, as applicable). Otherwise, the stipulations of Article 8 apply. The clubs defeated in the first and second qualifying rounds are eliminated from the competition. The clubs defeated in the third qualifying round are entitled to play in the play-offs of the UEFA Europa League in progress. Clubs from the same association cannot be drawn against each other.
Play-offs
Play-off matches are played according to the cup (knockout) system, with each club playing each opponent twice, in home and away matches. The team which scores the greater aggregate of goals in the two matches qualifies for the UEFA
9. Annexes
126
Champions League group stage. Otherwise, the stipulations of Article 8 apply. The clubs defeated in the play-offs are entitled to play in the group stage of the UEFA Europa League in progress. Clubs from the same association cannot be drawn against each other. 1st matchday: 2 v 3 4th matchday: 1 v 3 4 v 1 4 v 2 2nd matchday: 1 v 2 5th matchday: 3 v 2 3 v 4 1 v 4 3rd matchday: 3 v 1 6th matchday: 2 v 1 2 v 4 4 v 3 If two or more teams are equal on points on completion of the group matches, the following criteria are applied to determine the rankings (in descending order): higher number of points obtained in the group matches played among the teams in question; superior goal difference from the group matches played among the teams in question; higher number of goals scored in the group matches played among the teams in question; higher number of goals scored away from home in the group matches played among the teams in question; if, after having applied criteria a) to d), two teams still have an equal ranking, criteria a) to d) are reapplied to determine the final ranking of the two teams. If this procedure does not lead to a decision, criteria f) to h) apply; superior goal difference from all group matches played; higher number of goals scored from all group matches played; higher number of coefficient points accumulated by the club in question, as well as its association, over the previous five seasons (see paragraph 9.02). The eight group-winners and eight runners-up of the group stage qualify for the round of 16. The clubs that finish this stage in third position in their group move into the round of 32 of the current UEFA Europa League. The clubs that finish this stage in fourth position in their group are eliminated. The four best third-ranked teams are seeded for the UEFA Europa League round of 32. This ranking is determined in accordance with the following criteria (in descending order): higher number of points obtained in the group matches superior goal difference higher number of goals scored higher number of away goals scored higher number of wins higher number of away wins higher number of coefficient points accumulated by the club in question, as well as its association, over the previous five seasons (see paragraph 9.02).
Round of 16
The round of 16 pairings are determined by means of a draw. The round of 16 is played under the knockout system, on a home-and-away basis (two legs). The UEFA administration ensures that the following principles are respected. Clubs from the same association cannot be drawn against each other. The winners and runners-up of the same group cannot be drawn against each other. The group-winners cannot be drawn against each other. The runners-up cannot be drawn against each other. The runners-up must play the first leg at home.
9. Annexes
127
The team which scores the greater aggregate of goals in the two matches qualifies for the quarter-finals. Otherwise, the stipulations of Article 8 apply.
Quarter-finals
The eight winners of the round of 16 contest the quarter-finals. The quarter-final pairings are determined by means of a draw. The quarter-finals are played under the knockout system, on a home-and-away basis (two legs). The team which scores the greater aggregate of goals in the two matches qualifies for the semi-finals. Otherwise, the stipulations of Article 8 apply.
Semi-finals
The four winners of the quarter-finals contest the semi-finals. The semi-final pairings are determined by means of a draw. The semi-finals are played under the knockout system, on a home-and-away basis (two legs). The team which scores the greater aggregate of goals in the two matches qualifies for the final. Otherwise, the stipulations of Article 8 apply.
Final
The final is played as one single match at a neutral venue. If the result stands as a draw at the end of normal playing time, extra time of two periods of 15 minutes is played. If one of the teams scores more goals than the other during extra time, that team is declared the winner. If the two teams are still equal after extra time, the winner is determined by kicks from the penalty mark (Article 17). The provisions of Article 8 do not apply to the final.
VII Fixtures, Match Dates, Venues and Kick-off Times Article 12
Match dates
All matches are played according to the UEFA Match Calendar (see Annex Ic). These dates are final and binding on all concerned, subject to the provisions of paragraphs 12.04, 12.05 and 12.06. The following principles apply to this competition: With the exception of the final, which is played on a Saturday, UEFA Champions League matches are played on Tuesdays and Wednesdays. From the play-offs onwards, the UEFA administration decides which UEFA Champions League matches are to be played on Tuesdays and which on Wednesdays on the basis of the relevant draw. As a rule, each club plays the same number of matches on a Tuesday and on a Wednesday. Matches within the same group are played on the same day. Exceptions to this rule can be made by the UEFA administration.
Kick-off times
As a rule, the play-offs, group matches, round of 16 matches, quarter-finals, semi-finals and final kick off at 20.45 CET. Exceptions to this rule can be set by the UEFA administration. In principle, the fixtures within a group on the last matchday must be played simultaneously. The UEFA administration is authorised to fix the kick-off times.
Confirmation of fixtures for the qualifying phase
The venues, dates and kick-off times for matches in the three qualifying rounds must be confirmed and communicated to the UEFA administration in writing by the associations of the clubs concerned by the deadline set by the UEFA administration. The UEFA administration may alter or confirm dates and kick-off times according to the principles set by the Club Competitions Committee. Failure to respect this provision may result in disciplinary measures.
9. Annexes
128
Automatic reversals If more than one club from the same city, or within a radius of 50km (31 miles) of each other, are taking part in the UEFA Champions League competition and/or play in the same stadium, and if the association and the clubs concerned explicitly declare when entering the clubs that their matches cannot be played on the same day, the UEFA administration may alter or confirm dates and kick-off times according to the principles set by the Club Competitions Committee. If more than one club from the same city, or within a radius of 50km (31 miles) of each other, are taking part in any of the UEFA club competitions and/or play in the same stadium, and if the association and the clubs concerned explicitly declare when entering the clubs that their matches cannot be played on the same day, priority is given to UEFA Champions League matches and UEFA Europa League matches are reversed.
Venues as from the play-offs
In principle, from the play-offs, a club must play all its matches in the competition at one and the same ground. Matches may be played either at the ground of the home club or at another ground in the same or another city within the territory of its association, or, if so decided by the UEFA administration and/or the UEFA disciplinary bodies, in the territory of another UEFA member association for reasons of safety or as a result of a disciplinary measure. In principle, venues are approved only if direct international flights and/or charter flights are able to land within an acceptable distance of the venue, in the country of the club concerned. If the match is being played in another city or country, the venue is subject to the approval of the UEFA administration.
Alternative venues
If, at any time during the season, the UEFA administration deems that, for whatever reason, some venues may not be fit for staging a match, UEFA may consult the associations and clubs concerned and ask them to propose an alternative venue, in accordance with the UEFA requirements. Should such an association and club not be able to propose an acceptable alternative venue by the deadline set by the UEFA administration, UEFA may select an alternative, neutral venue and make all the necessary arrangements for the staging of the match together with the relevant association and local authorities. In both cases, the costs of staging the match are borne by the home club. The UEFA administration take a final decision on the match venue in due time.
Final
The final is organised by a local organising committee (LOC) on the basis of a contract between the host association and UEFA. The date and venue are chosen by the Executive Committee. In principle, the local organisation of the final is entrusted to a different association each year.
9. Annexes
129
Annex E: Table comparing S1 to C1
S1 C1 Rules applied
III Trophy, Plaques and Medals
III Trophy, Plaques and Medals
-
Article 3 Trophy
Article 3 Trophy
-
3.01 The original trophy,
which is used for the official presentation ceremony at the final and at other official
events approved by UEFA, remains in UEFA’s keeping and ownership at all times.
3.01 The original trophy is
used for the official trophy handover at the final and for other official events that are
approved by UEFA. The original trophy remains in the keeping and ownership of
UEFA at all times.
- General: Avoid long and
complex sentences - General: Be as explicit and precise as possible (‘official
presentation ceremony’ = ‘official trophy handover’) - Grammar: Avoid
possessive case (´s ) and possessive pronouns (UEFA’s) - Grammar: Try to align
prepositions that refer to the same verb (‘at’=’for’)
A full-size replica trophy, the
UEFA European Football Championship winner’s trophy, is awarded to the
winning association.
The winning association
receives a full-size replica trophy, the winner's trophy of the UEFA European Football
Championship.
- General: Avoid long and
complex sentences - Lexis: Avoid noun clusters, try to explain them:
UEFA European Football Championship winner’s trophy - Grammar: Transform
passive into active constructions (‘is awarded to’-‘ receives’)
3.02 Any association which wins the trophy three
consecutive times or five times in total receives a special mark of recognition.
3.02 An association which wins the trophy three
consecutive times or which wins it five times in total receives a special mark of
recognition.
- Lexis: Avoid the determiner any
- General: Be as explicit and precise as possible (insertion of ‘which wins it’)
Once a cycle of three successive wins or five in total has been completed, the
association concerned starts a new cycle from zero.
Once an association completes a cycle of three successive wins or a cycle of
five wins in total, the association starts a new cycle from zero.
- Grammar: Transform passive into active constructions (‘an association completes’, not
‘has been completed’) - -General: Be as explicit and precise as possible (repeat
‘wins’)
3.03 Replica trophies awarded to winners of the UEFA European Football
Championship (past and current) must remain within the relevant association’s
control at all times and must not leave the association’s country without UEFA’s prior
written consent.
3.03 The relevant associations must at all times remain in possession of the
replica t rophies that the past and the current winners of the UEFA European Football
Championship have received. The replicas must not leave the country of the relevant
association without the prior written consent of UEFA.
- Grammar: Transform passive into active constructions - General: Order the parts
of the sentences logically (begin with subject ‘the relevant associations’)
- General: Avoid long and complex sentences - General: Be as explicit
and precise as possible (‘possession’ instead of ‘control’) - Grammar: Avoid
possessive case (´s ) and possessive pronouns - Grammar: Use
determiners (the, a) wherever
9. Annexes
130
S1 C1 Rules applied
possible
Associations must not permit a replica trophy to be used in any context where a third
party (including, without limitation, associations’ sponsors and other
commercial partners) is granted visibility or in any other way which could lead to
an association between any third party and the replica trophy and/or the
competition.
The associations must not approve the use of a replica trophy in a context where a
third party is granted visibility. Third parties include, without limitation, sponsors of the
associations as well as other commercial partners. The associations must not
approve the use of a replica trophy in a way that could create a link between a third
party and the replica trophy and/or the competition.
- General: Avoid long and complex sentences - General: Be as explicit
and precise as possible (‘the use of’ instead of ‘to be used’;
‘approve’ instead of ‘permit’) - Grammar: Avoid possessive case (´s ) and
possessive pronouns - Lexis: Avoid the term association when it does not
designate a football association - Grammar: Use determiners (the, a) wherever
possible
Associations must comply with any trophy use guidelines that the UEFA
administration may issue from time to time.
The associations must comply with trophy use guidelines of any kind that the
UEFA administration can issue from time to time.
- Lexis: Avoid the determiner any - Grammar: Use
determiners (the, a) wherever possible - Lexis: Avoid the
ambiguous modal verb may
3.04 Associations may not, and may not permit any third party to, develop, create, use,
sell or distribute any promotional materials or merchandise bearing any
representation of the trophy or any replica thereof (including, without limitation,
trophy li ft images) or use any such representation in a manner that could lead to an
association between any third party and the t rophy, replica trophy and/or the competition.
3.04 Associations must not develop, create, use, sell, or distribute any promotional
materials or any merchandise that bear a representation of the trophy or a replica of the
trophy (including, without limitation, images that show the winners li fting the trophy).
Associations must not use such a representation in a manner that could lead to a
link between a third party and the trophy, the replica trophy and/or the competition.
Furthermore, they must not permit a third party to do any of the above.
- General: Avoid long and complex sentences - Lexis: Avoid the
ambiguous modal verb may - Grammar: Avoid the use of ing-words
- General: Be as explicit and precise as possible (‘trophy lift images’)
- Lexis: Avoid the term association when it does not designate a football association
- -Grammar: Use determiners (the, a) wherever possible
Commemorative plaque
Commemorative plaques - General: Be as explicit
and precise as possible (plural is more fitting because every association receives one, so it’s
more than one plaque)
3.05 Each association that competes in the final tournament receives a
commemorative plaque.
3.05 Each association that competes in the final tournament receives a
commemorative plaque.
-
Semi-finalist plaque
Semi-finalist plaque
-
3.06 Each defeated semi-
finalist receives a plaque.
3.06 Each defeated semi-
finalist receives a plaque.
-
Final plaque
Final plaque -
3.07 The finalists receive a 3.07 Each finalist receives a - General: Be as explicit
9. Annexes
131
S1 C1 Rules applied
plaque.
plaque. and precise as possible
Medals
Medals -
3.08 Forty gold medals are
presented to the winning team and 40 silver medals to the runner-up. 40 bronze
medals are presented to the defeated semi-finalists.
3.08 The winning team
receives 40 gold medals. The runner-up receives 40 silver medals . The defeated semi-
finalists each receive 40 bronze medals.
- Grammar: Transform
passive into active constructions - General: Be as explicit
and precise as possible
Additional medals may not be
produced.
Additional medals will not be
produced.
- Lexis: Avoid the
ambiguous modal verb may
IV Responsibilities
IV Responsibilities -
Article 4 Responsibilities of the
associations
Article 4 Responsibilities of the
associations
-
4.01 The associations are responsible for the behaviour
of their players, officials, members, supporters and anyone carrying out a
function at a match on their behalf.
4.01 An association is responsible for the behaviour
of its players, its officials, its members, its supporters and every person who carries out
a function at a match on behalf of the association.
- Grammar: Avoid the use of ing-words
- General: Be as explicit and precise as possible (precise ‘on their behalf’; repeat subject
‘the associations’; put ‘the associations’ into singular to emphasize that each single
association is responsible for the behaviour of its people; repeat ‘its’; ‘anyone’=’every person’)
- Grammar: Avoid possessive case (´s ) and possessive pronouns (‘their’=’of
the association’)
4.02 The association in the
territory of which a qualifying match or the final tournament is being staged is considered
the host association.
4.02 The national association
that hosts a qualifying match or the final tournament on its territory is considered the
host association for that match or tournament.
- Grammar: Transform
passive into active constructions - General: Be as explicit and precise as possible (insert
‘national association’; insert ‘ for that match or tournament’/lacking in S1)
- General: Avoid overly complex constructions and write clearly (‘in the territory of which’
= ‘on its territory’)
4.03 The host association is responsible for order and security before, during and
after the match. The host association may be called to account for incidents of any
kind and may be disciplined.
4.03 The host association is responsible for order and security before, during and
after the matches. The host association can be called to account and be disciplined for
incidents of any kind.
- Lexis: Avoid the ambiguous modal verb may - Lexis: Avoid the
determiner any - General: Avoid overly complex constructions and write
clearly (align ‘be called to account and be disciplined for…’)
4.04 Matches must, in
principle, be played in a stadium within the territory of
4.04 In principle, matches
must be played in a stadium on the territory of the
- General: Be as explicit
and precise as possible (insert ‘respective’)
9. Annexes
132
S1 C1 Rules applied
the host association. respective host association. - General: Try to avoid synonyms or different paraphrases expressing the
same content in order to ensure uniformity (‘within the territory’ = ‘on the territory’)
- General: Order the parts of the sentences logically (do not separate verbs chain ‘must be
played’, put the adjective ‘in principle’ at the beginning)
Exceptionally, matches may be played in the territory of
another UEFA member association, if so decided by the UEFA administration
and/or the disciplinary bodies, for reasons of safety or as a result of a disciplinary
measure.
Exceptionally, matches can be played on the territory of
another UEFA member association, if the UEFA administration and/or the
disciplinary bodies decide so (for reasons of safety or as a result of a disciplinary
measure).
- Lexis: Avoid the ambiguous modal verb may
- General: Try to avoid synonyms or different paraphrases expressing the
same content in order to ensure uniformity (‘in the territory’ = ‘on the territory’)
- Grammar: Transform passive into active constructions
Additional responsibilities for the final tournament
Additional responsibilities for the final tournament
-
4.05 The UEFA administration informs the associations participating in
the final tournament about any further guidelines, directives or decisions related
to the final tournament and provides them with all relevant documents in due
time.
4.05 The UEFA administration informs the associations that participate
in the final tournament about any further guidelines, directives, or decisions that
are related to the final tournament. The UEFA administration provides the
associations with all relevant documents in due time.
- Grammar: Avoid the use of ing-words - General: Avoid long and
complex sentences - Grammar: Minimize the use of personal pronouns (they,
it, etc.): ‘them’=’the associations’
VI Competition System
VI Competition System
-
Article 6
Competition stages
Article 6
Competition stages
-
6.01 The competition consists of a qualifying competition
and a final tournament.
6.01 The competition consists of a qualifying competition
and of a final tournament.
- Grammar: Repeat the preposition in conjoined
constructions where appropriate (‘of’)
Article 7 A. Qualifying competition
Group formation
Article 7 A. Qualifying competition
Group formation
-
7.01 The teams of the host associations for the final
tournament, Poland and Ukraine, qualify automatically for the
final tournament.
7.01 The teams of the host associations for the final
tournament, Poland and Ukraine, qualify automatically for the final tournament.
-
The remaining teams are drawn into 6 groups of 6 teams and 3
groups of 5 teams.
The remaining teams are drawn into 6 groups of 6 teams and 3 groups of 5
teams.
-
The groups are formed by the The UEFA administration - General: Avoid long and
9. Annexes
133
S1 C1 Rules applied
UEFA administration by means of a draw, with seeded teams, on
completion of the 2010 FIFA World Cup qualifying competition.
forms the groups by means of a draw, on completion of the qualifying competition of the
2010 FIFA World Cup. The draw is based on a seeding system.
complex sentences - General: Order the parts of the sentences logically
- General: Be as explicit and precise as possible (specify ‘with seeded teams’)
The reigning European
champions are always seeded.
The reigning European
champion is always seeded.
- General: Be as explicit
and precise as possible (there is only one European champion – singular)
The other associations are
classified on the basis of the UEFA national team coefficient
ranking system (see Annex I, paragraph 1.2.1).
The remaining associations
are classified on the basis of the UEFA national team coefficient ranking system
(see Annex I, paragraph 1.2.1).
- General: Be as explicit
and precise as possible (‘remaining’ instead of ‘other”)
The decisions of the UEFA
administration are final.
The decisions of the UEFA administration are final.
-
7.02 If any of the teams in question have the same
coefficient, the following criteria, relating only to the most
recent qualifying competition, are determinant, in the order given:
7.02 If two or more of these teams have the same
coefficient, the following criteria are determinant. Only the most recent qualifying
competition is taken into consideration. The criteria are applied in the order given :
- Lexis: Avoid the determiner any
- General: Be as explicit and precise as possible (‘two or
more of these teams’ instead of ‘any’) - General: Avoid long and
complex sentences
a) UEFA national team coefficient from the matches played;
a) UEFA national team coefficient resulting from the matches played;
- General: Be as explicit and precise as possible (add ‘resulting from’)
b) average goal difference;
b) average goal difference; -
c) average number of goals scored;
c) average number of goals scored;
-
d) average number of away goals scored;
d) average number of away goals scored;
-
e) fair play ranking;
e) fair play ranking; -
f) drawing of lots.
f) drawing of lots. -
Match system for the qualifying competition
Match system for the qualifying competition
-
7.03 The matches in the qualifying competition are played in groups according to
the league system, with each team playing all opponents in
its group in a series of home and away matches.
7.03 The matches in the qualifying competition are played in groups according to
the league system. Each team plays against every other team in its group in a
series of home and away matches.
- General: Avoid long and complex sentences - Grammar: Avoid the use
of ing-words - General: Be as explicit and precise as possible (‘play
against’) - General: Try to avoid synonyms or different
paraphrases expressing the same content in order to ensure
9. Annexes
134
S1 C1 Rules applied
uniformity (‘all opponents’=’every other team’)
Three points are awarded for a win, one point for a draw,
and no points for a defeat.
Three points are awarded for a win, one point for a draw,
and no points for a defeat.
-
Equality of points after the group matches
Equality of points after the group matches
-
7.04 If two or more teams are equal on points on completion of the group matches, the
following criteria are applied, in the order given, to determine the rankings:
7.04 If two or more teams are equal on points on completion of the group matches, the
following criteria are applied to determine the rankings. They are applied in the order
given:
- General: Avoid long and complex sentences
a) higher number of points obtained in the group
matches played among the teams in question;
a) higher number of points that the teams obtained in the
group matches played among the teams in question;
- Grammar: Do not omit relative pronouns (‘that’)
- General: Be as explicit and precise as possible (‘the teams in questions’)
b) superior goal difference
from the group matches played among the teams in question;
b) superior goal difference
from the group matches played among the teams in question;
-
c) higher number of goals scored in the group matches played among the teams
in question;
c) higher number of goals scored in the group matches played among the teams in
question;
-
d) higher number of goals scored away from home in
the group matches played among the teams in question;
d) higher number of away goals scored in the group
matches played among the teams in question;
- General: Try to avoid synonyms or different
paraphrases expressing the same content in order to ensure uniformity (‘away goals’)
e) if, after applying criteria a)
to d) to several teams, two or more teams still have an equal ranking, the criteria
a) to d) will be reapplied to determine the ranking of these teams.
e) If the criteria a) to d) have
been applied to several teams and two or more teams still have an equal ranking,
the criteria a) to d) are reapplied to determine the ranking of these teams.
- Grammar: avoid future
tense - Avoid overly complex constructions and write clearly
- Grammar: Avoid the use of ing-words
If this procedure does not
lead to a decision, criteria f) to j) apply;
If this procedure does not
lead to a decision, the criteria f) to j) are applied;
- Grammar: Use
determiners (the, a) wherever possible - Lexis: Avoid the active
form of the verb apply
f) superior goal difference in all group matches;
f) superior goal difference in all group matches;
-
g) higher number of goals
scored in all group matches;
g) higher number of goals
scored in all the group matches;
- General: Try to avoid
synonyms or different paraphrases expressing the same content in order to ensure
uniformity (‘in all the’, see 8.07 e) )
9. Annexes
135
S1 C1 Rules applied
h) higher number of goals scored away from home in all group matches;
h) higher number of away goals scored in all the group matches;
- General: Try to avoid synonyms or different paraphrases expressing the
same content in order to ensure uniformity (‘in all the’, see 8.07 e) )
i) fair play ranking in all group
matches;
i) fair play ranking in all the
group matches;
- General: Try to avoid
synonyms or different paraphrases expressing the same content in order to ensure
uniformity (‘in all the’, see 8.07 e) )
j) drawing of lots.
j) drawing of lots. -
Qualification for the final
tournament
Qualification for the final
tournament
-
7.05 The nine group winners and the best runner-up qualify
directly for the final tournament.
7.05 The nine group winners and the best runner-up qualify
directly for the final tournament.
-
7.06 To determine the best
runner-up, only results against the teams in first, third, fourth
and fi fth place are taken into account, with the following criteria being taken into
consideration in the order given:
7.06 Only the results against
the teams in the first, third, fourth and fi fth place are taken into account to
determine the best runner-up. For this purpose, the following criteria are applied
in the order given:
- General: Order the parts
of the sentences logically - General: Avoid long and complex sentences
- General: Try to avoid synonyms or different paraphrases expressing the
same content in order to ensure uniformity (‘are taken into consideration’ = ‘are applied’)
a) higher number of points
obtained in these matches;
higher number of points that
the teams obtained in these matches;
-
b) superior goal difference from these matches;
- -
c) higher number of goals scored in these matches;
- -
d) higher number of away
goals scored in these matches;
- -
e) position in the UEFA
national team coefficient ranking system (see Annex I, paragraph 1.2.2);
- -
f) fair play ranking in these matches;
- -
g) drawing of lots.
- -
7.07 The eight remaining runners-up contest play-off matches.
- -
9. Annexes
136
S1 C1 Rules applied
The four ties are determined by means of a draw.
- -
The four runners-up with the best position in the UEFA
national team coefficient ranking system (see Annex I, paragraph 1.2.2) are seeded
for the draw.
- -
If any of the teams in question have the same coefficient, the criteria listed in paragraph
7.02 apply.
If two or more of these teams have the same coefficient, the criteria of paragraph 7.02 are
applied.
- Lexis: Avoid the determiner any - General: Be as explicit
and precise as possible (‘two or more of these teams’ instead of
‘any’) - General: Avoid overly complex constructions and write
clearly (‘listed in’=’of’) - Lexis: Avoid the active
form of the verb apply
The play-offs are played according to the knockout system, with each team
playing each opponent twice, in home and away matches.
The play-offs are played according to the knockout system (in home and away
matches).
- Grammar: Avoid the use of ing-words) - General: Avoid long and
complex sentences - General: Be as explicit and precise as possible (with
each team playing each opponent twice, in home and away matches’ - unnecessary
and even a little confusing, almost as if they play two sets of home and away matches)
The seeded teams play the
return match at home.
The seeded teams play the
return match at home.
-
The teams which score the greater aggregate of goals in the two matches qualify for
the final tournament.
The teams which score the greater aggregate of goals qualify for the final
tournament.
- General: Be as explicit and precise as possible (‘greater
aggregate of’: aggregate means total, i.e. over both matches: ‘in the two matches’=unnecessary)
Otherwise, the provisions of paragraph 7.08 apply.
Otherwise, the provisions of paragraph 7.08 are applied.
- Lexis: Avoid the active form of the verb apply
7.08 For matches played under the knockout system, if
the two teams involved in a tie score the same number of goals over the two legs, the
team which scores more away goals qualifies for the next stage.
For matches that are played under the knockout system,
the following criterion is applied: i f both teams score the same number of
aggregate goals, the team which scores more away goals qualifies for the next
stage.
- General: Avoid long and complex sentences
- General: Try to avoid synonyms or different paraphrases expressing the
same content in order to ensure uniformity (‘over the two legs’=’aggregate goals’ (see
above)) - General: avoid redundancies (‘involved in a tie’)
If this procedure does not
produce a result, i.e. if the two teams score the same number of home and away
goals, two 15-minute periods of extra time are played at the end of the second leg.
If this procedure does not
lead to a decision, i.e. if both teams score the same number of home and away
goals, an extra time of two periods of 15 minutes is played at the end of the
- General: Try to avoid
synonyms or different paraphrases expressing the
same content in order to ensure uniformity (8.12: ‘an extra time of two periods of 15 minutes’; 7.04
e): ‘if this procedure does not
9. Annexes
137
S1 C1 Rules applied
return match. lead to a decision’) - General: Be as explicit
and precise as possible (‘second leg’=’return match’)
If, during extra time, both teams score the same
number of goals, away goals count double (i.e. the visiting team qualifies).
If, during extra time, both teams score the same
number of goals, the away goals count double (i.e. the visiting team qualifies).
-
If no goals are scored during
extra time, kicks from the penalty mark (Article 16) determine which team
qualifies for the next stage.
If no goals are scored during
extra time, the team that qualifies for the final tournament is determined by
kicks from the penalty mark (see Article 16).
- General: Try to avoid
synonyms or different paraphrases expressing the same content in order to ensure
uniformity (8.08: ‘determined by kicks from the penalty mark’)
Article 8
B. Final tournament
- -
8.01 The UEFA Executive Committee has entrusted the
Polish Football Association (PZPN) and the Football Federation of Ukraine (FFU)
with the joint organisation and staging of the final tournament.
- -
Group formation
-
8.02 The UEFA administration divides the 16
teams that qualify for the final tournament into four groups (A, B, C, D)
of four teams each.
8.02 16 teams qualify for the final tournament. The UEFA
administration divides these 16 teams into four groups of four teams each (groups A, B,
C, D).
- General: Avoid long and complex sentences
- Grammar: Avoid parenthesis (‘groups A, B, C, D’ into brackets at the end of the
sentence in order to not split up the sentence)
8.03 By means of a draw, the four groups are formed as
follows: Group A Group B Group C Group D
A1 B1 C1 D1 A2 B2 C2 D2 A3 B3 C3 D3
A4 B4 C4 D4
- -
Coefficients
- -
8.04 The seeded teams are
the host associations, with the reigning European champions if they qualify, and
one or two teams with the best coefficients according to the UEFA
national team coefficient ranking system (see Annex I, paragraph 1.2.2).
8.04 The following teams are
seeded: the host associations, the reigning European champion if he
qualifies, and one or two teams with the best coefficients according to the
UEFA national team coefficient ranking system (see Annex I, paragraph
- General: Avoid long and
complex sentences - General: Be as explicit and precise as possible (there is
only one European champion – singular)
9. Annexes
138
S1 C1 Rules applied
1.2.2).
The other finalists are assigned to the four groups by means of a
draw, according to their coefficients.
The other finalists are assigned to the four groups by means of a draw
(according to their coefficients).
- General: Avoid long and complex sentences - Grammar: Avoid phrasal
constituents that are added after a comma
8.05 If any of the teams in
question have the same coefficient, the criteria listed in
paragraph 7.02 apply.
8.05 If two or more of the
teams in question have the same coefficient, the criteria of paragraph 7.02 are
applied.
- Lexis: Avoid the
determiner any - Lexis: Avoid the active form of the verb apply
- General: Avoid overly complex constructions and write
clearly (‘listed in’=’of’)
Results from the play-offs referred to in paragraph 7.07
are thereby not taken into consideration.
The results from the play-offs (see paragraph 7.07) are
thereby not taken into consideration.
- Grammar: Use determiners (the, a) wherever
possible - General: Avoid overly
complex constructions and write clearly/General: Try to avoid synonyms or different
paraphrases expressing the same content in order to ensure uniformity (‘referred to in’=’see’)
Group match schedule
- -
8.06 Each team plays each of the other teams in the same group according to a league
system (three points for a win, one point for a draw, no points for a defeat).
8.06 Each team plays against every other team in its group according to a league system
(three points for a win, one point for a draw, no points for a defeat).
- General: Be as explicit and precise as possible: ‘play’: add preposition ‘against’; precise
‘each of the other teams’=’every other team of its group” - General: Try to avoid
synonyms or different paraphrases expressing the same content in order to ensure
uniformity (‘in the same group’=’in its group’ (see article 7.03)
The
group matches are played according to the following schedule.
- -
The last two
matches in each group must both kick off at the same time.
- -
The first-named team is considered as the home
team. Match day 1 Match day 2 Match day 3
Group A A1 vs A2 A1 vs A3 A4 vs A1 A3 vs A4 A2 vs A4 A2 vs A3
Group B B1 vs B2 B1 vs B3 B4 vs B1 B3 vs B4 B2 vs B4 B2 vs B3
Group C C1 vs C2 C1 vs C3 C4 vs C1
The first-named team is considered the home team.
-
- General: Try to avoid synonyms or different
paraphrases expressing the same content in order to ensure uniformity: ‘is considered as’=’is
considered’ (see article 4.02)
9. Annexes
139
S1 C1 Rules applied
C3 vs C4 C2 vs C4 C2 vs C3 Group D D1 vs D2 D1 vs D3 D4 vs D1
D3 vs D4 D2 vs D4 D2 vs D3
Equality of points after the group matches
- -
8.07 If two or more teams are equal on points on completion of the group matches, the
following criteria are applied, in the order given; to determine the rankings:
8.07 If two or more teams are equal on points on completion of the group matches, the
following criteria are applied to determine the rankings. They are applied in the order
given:
- General: Avoid long and complex sentences
a) higher number of points obtained in the matches among the teams in question;
a) higher number of points that the teams obtained in the matches played among the
teams in question;
- General: Try to avoid synonyms or different paraphrases expressing the
same content in order to ensure uniformity: ‘matches played among’ (see 7.04)
b) superior goal difference in
the matches among the teams in question (if more than two teams finish equal
on points);
b) superior goal difference
from the matches played among the teams in question (if more than two teams are
equal on points on completion of the group matches);
- General: Try to avoid
synonyms or different paraphrases expressing the same content in order to ensure
uniformity: ‘from the matches played among’ (see 7.04) - General: Be as explicit
and precise as possible (‘finish equal on points’)
c) higher number of goals scored in the matches among
the teams in question (if more than two teams finish equal on points);
c) higher number of goals scored in the matches played
among the teams in question (if more than two teams are equal on points on completion
of the group matches);
- General: Try to avoid synonyms or different
paraphrases expressing the same content in order to ensure uniformity: ‘the matches played
among’ (see 7.04) - General: Be as explicit and precise as possible (‘finish
equal on points’)
d) superior goal difference in all the group matches;
- -
e) higher number of goals
scored in all the group matches;
- -
f) position in the UEFA
national team coefficient ranking system (see Annex I, paragraph 1.2.2);
- -
g) fair play conduct of the teams (final tournament);
g) fair play conduct of the teams in question (final tournament);
- General: Try to avoid synonyms or different paraphrases expressing the
same content in order to ensure uniformity: ‘teams in question’ (see above)
h) drawing of lots.
- -
9. Annexes
140
S1 C1 Rules applied
8.08 If two teams which have the same number of points, the same number of goals
scored and conceded play their last group match against each other and are still
equal at the end of that match, the ranking of the two teams in question is
determined by kicks from the penalty mark (Article 16), provided no other teams
within the group have the same number of points on completion of all group
matches.
8.08 Two teams have the same number of points and the same goal difference.
These teams play their last group match against each other. If this match ends in a
draw, the ranking of the two teams in question is determined by kicks from the
penalty mark (see Article 16; provided that no other teams within the same group have
the same number of points on completion of the group matches).
- General: Avoid long and complex sentences - General: Be as explicit
and precise as possible/ Try to avoid synonyms or different paraphrases expressing the
same content in order to ensure uniformity (‘same number of goals scored and
conceded’=’goal difference’) - General: Be as explicit and precise as possible: ‘are still
equal at the end of that match’=’match ends in a draw’ - Grammar: Do not omit
relative pronouns (‘that’) - Grammar: Use determiners (the, a) wherever
possible - General: avoid redundancies (‘all’)
Should more than two teams
have the same number of points, the criteria listed under paragraph 8.07
apply.
Should more than two teams
have the same number of points, the criteria of paragraph 8.07 are applied.
- General: Be as explicit
and precise as possible: ‘listed in’=’of’ - Lexis: Avoid the active
form of the verb apply
Quarter-finals
- -
8.09 The winners and
runners-up in each group play the quarter-finals over one match,
as follows:
8.09 The winners and
runners-up in each group play the quarter-finals over one match. For this purpose,
the following scheme is applied:
- Grammar: Avoid phrasal
constituents that are added after a comma
Match 1 Winner Group A vs
Runner-up Group B Match 2 Winner Group B vs Runner-up Group A
Match 3 Winner Group C vs Runner-up Group D Match 4 Winner Group D vs
Runner-up Group C
- -
Semi-finals
Semi-final
8.10 The four winners of the
quarter-finals play the semi-finals over one match, as follows:
8.10 The four winners of the
quarter-finals play the semi-finals over one match. For this purpose, the following
scheme is applied:
- Grammar: Avoid phrasal
constituents that are added after a comma
Winner Match 1 vs Winner Match 3 Winner Match 2 vs Winner
Match 4
- -
Final
- -
9. Annexes
141
S1 C1 Rules applied
8.11 The winners of the semi-finals play in the final.
- -
Same number of goals in a
quarter-final, semi-final or the final
Same number of goals in a
quarter-final, in a semi-final or in the final.
- Grammar: Use
determiners (the, a) wherever possible - Grammar: Repeat the
preposition in conjoined constructions where appropriate (‘in)
8.12 If the result stands as a
draw at the end of normal playing time, extra time of two periods of 15 minutes is
played.
8.12 If the match ends in a
draw after normal playing time, an extra time of two periods of 15 minutes is
played.
- Grammar: Use
determiners (the, a) wherever possible - General: Be as explicit
and precise as possible/Try to avoid synonyms or different paraphrases expressing the
same content in order to ensure uniformity: ‘result stands as a draw’=’match ends in a draw’
(see 8.08)
If the two teams are still equal after extra time, the winners are determined by
kicks from the penalty mark (Article 16).
f the extra time ends in a draw, the winner is determined by kicks from the
penalty mark (see Article 16).
- General: Be as explicit and precise as possible (‘the two teams are still equal after extra
time’=’extra time ends in a draw’; ‘the winner’ in singular since there is only one winner per
match) - General: Try to avoid synonyms or different
paraphrases expressing the same content in order to ensure uniformity (insert ‘see’)
VIII Match Dates, Venues
and Kick-off Times Article 11
- -
11.01 The competition will be
staged over the two seasons that follow the 2010 FIFA World Cup.
11.01 The competition will
take place following the 2010 FIFA World Cup. It will be staged over two seasons.
- General: Avoid long and
complex sentences
A. Qualifying competition Match dates
- -
11.02 The following twelve
dates are reserved for the group matches in the qualifying
competition of the 2010-12 UEFA European Football Championship:
11.02 The group matches in
the qualifying competition of the 2010-12 UEFA European Football Championship are
scheduled for the following twelve dates:
- General: Be as explicit
and precise as possible (reserve-schedule: matches are scheduled)
2010 a) 3/4 and 7 September 2010 b) 8/9 and 12 October 2010
2011 c) 25/26 and 29 March 2011 d) 3/4 and 7 June 2011
- -
9. Annexes
142
S1 C1 Rules applied
e) 2/3 and 6 September 2011
f) 7/8 and 11 October 2011 (dates for the last matches in
all groups)
- -
11.03 The following two dates are reserved for the play -off
matches between the remaining eight runners -up: a) 11/12 November 2011
b) 15 November 2011
11.03 The play-off matches between the remaining eight
runners-up are scheduled for the following two dates:
- General: Be as explicit and precise as possible
(reserve-schedule: matches are scheduled) - Order the parts of the
sentences logically
11.04 Matches as referred to in paragraph 11.02 may be
played on other dates if both associations concerned are in agreement and the principles
governing the release of players for association teams as laid out in Annex 1,
Article 1 of the FIFA Regulations for the Status and Transfer of Players are
adhered to.
11.04 Under the following conditions, the group
matches mentioned in paragraph 11.02 can be played on other dates : if both
associations in question agree and if the principles governing the release of
players for association teams as laid out in Annex 1, Article 1 of the FIFA Regulations for
the Status and Transfer of Players are complied with.
- General: Avoid long and complex sentences
- General: Be as explicit and precise as possible (‘group matches’ instead of ‘matches’;
“are in agreement’=’agree’) - Lexis: Avoid the ambiguous modal verb may -
uniformity: “in question” instead of “concerned” - General: Try to avoid
synonyms or different paraphrases expressing the same content in order to ensure
uniformity (‘are adhered to’=’are complied with’, see 3.02)
The associations in a group have
30 days after the draw in which to agree on the order of their matches.
The associations in a group have 30 days after the draw
in which they have to agree on the order of their matches.
- General: Be as explicit and precise as possible (‘in
which to agree on’=’in which they have to agree’
The exact date of each match
must be specified (e.g. Saturday 4 September 2010).
- -
If the associations in question
are unable to reach agreement, the matches will take place
according to a standard fixture list drawn up by the UEFA administration.
If the associations in question
are unable to reach agreement, the matches will take place according to a
standard fixture list that is drawn up by the UEFA administration.
- Grammar: Do not omit
relative pronouns (‘that’)
This standard fixture list will
take into consideration relevant factors for the match
organisation, such as weather conditions.
This standard fixture list takes
into consideration factors that are relevant for the match organisation, such as climate.
- Grammar: avoid future
tense - General: Be as explicit and precise as possible:
‘relevant factors for..’ = ‘factors that are relevant for’; ‘weather conditions’ = ‘climate: not the
weather, but the climate should be taken into consideration)
11.05 For reasons of sporting fairness, the UEFA
administration is entitled to order matches within a group to be
11.05 For reasons of sporting fairness, the UEFA
administration can demand that matches within the same group are played at the same
- Grammar: turn passive into active constructions (‘is
entitled to order matches’=’can demand that matches’ - Grammar: Avoid future tense
9. Annexes
143
S1 C1 Rules applied
played at the same time. time. (‘to be played at’=’are played’) - General: Be as explicit and precise as possible (‘within
the same group’)
Any subsequent changes of date are subject to the approval of the UEFA
administration.
Subsequent changes of date are subject to the approval of the UEFA administration.
- Lexis: Avoid the determiner any
The host association of a match must, in this case, also inform the other
associations in the group.
-
Venues and kick-off times
- -
11.06 The venues of the
matches are fixed by the host associations and announced to
their opponents and the UEFA administration at least 60 days in advance.
11.06 The match venues are
fixed by the host associations. They have to be announced to the visiting
associations and to the UEFA administration at least 60 days in advance.
- General: Avoid long and
complex sentences - General: Make a uniform use of singular and plural (here
better singular because clearer)
When
fixing a venue, the host association must take into account the length of the
journey to be undertaken by the visiting association.
When a host association fixes
a venue, it must take into account the length of the journey of the visiting
association.
- Grammar: avoid the use
of ing-words - General: avoid redundancies (‘to be undertaken
by’) - General: Make a uniform use of singular and plural (here
better singular because clearer)
Unless the visiting association agrees otherwise, the venue for a qualifying match must be
no more than a twohour bus drive from the nearest international airport.
The venue for a qualifying match must be no more than a twohour bus drive from the
nearest international airport, unless the visiting association agrees otherwise.
- General: order the parts of the sentences logically
- General: Make a uniform use of singular and plural (here better singular because clearer)
The kick-off times must also
be announced to the UEFA administration at least 60 days before the match.
The kick-off times must be
announced to the UEFA administration at least 60 days before the match.
- Lexis: Avoid redundant
conjunctions such as “also”, “as well”, etc.
- General: Make a uniform use of singular and plural (here better singular because clearer)
Arrival of the teams at the
match venue
- -
11.07 The associations must arrange for their teams to
arrive at the match venue early enough in order to be able to
hold their pre-match press conference before the media deadlines of both
involved countries and in any case no later than 24 hours before kick-off.
11.07 The associations must arrange that their teams
arrive early enough at the match venue so that they can hold their pre-match press
conference before the media deadlines in both involved countries. In any case, the
teams must arrive at the match venue at least 24 hours before kick-off.
- General: Try to avoid
synonyms or different paraphrases expressing the same content in order to ensure
uniformity (‘pre-match press conference in both countries’ instead of ‘of’; ‘no later than 24h
before’=’at least 24h before’) - General: Avoid long and complex sentences
- General: Be as explicit and precise as possible (‘arrange for their teams to
9. Annexes
144
S1 C1 Rules applied
arrive’=’arrange that their teams arrive’; ‘in order to be able to’-‘so that they can’)
B. Final tournament
Match dates
- -
11.08 The final tournament will take place from 8 June to
1 July 2012.
- -
Venues and kick-off times
11.09 The UEFA
administration is responsible for drawing up the match schedule for the
final tournament.
11.09 The UEFA
administration is responsible for the drawing up of the match schedule for the final
tournament.
- Grammar: avoid the use
of ing-words
Each team must have at least 48 rest hours between each match
played.
The teams must have at least 48 rest hours between the matches.
- General: Be as explicit and precise as possible (‘each
team’=’the teams’; ‘between each match played’=’between the matches’)
Arrival of the teams in host
countries
- -
11.10 Each association taking part in the final tournament
must arrive at their team hotel in one of the host countries at least five days before their
first match.
11.10 The teams must arrive at their team hotel in one of
the host countries at least five days prior to their first match.
- General: Be as explicit and precise as possible (‘the
teams’ instead of ‘each association’) - General: avoid
redundancies (‘associations taking part in’=’the teams’)
Arrival of the teams at the match venue
- -
11.11 The teams must arrive
at their transfer hotel or be within a 60km radius of the stadium where their match is
to be played no later than 24 hours before kick-off.
11.11 The teams must arrive
at their transfer hotel at least 24 hours before kick-off; in any case, they must be within
a radius of 60km of their venue at least 24 hours before kick-off.
- General: Avoid long and
complex sentences - General: Try to avoid synonyms or different
paraphrases expressing the same content in order to ensure uniformity (no later
than…before’=’at least…before’) - Lexis: avoid noun clusters, try to explain them
(‘60km radius’=’radius of 60km’) - General: Be as explicit and precise as possible/ Try to
avoid synonyms or different paraphrases expressing the same content in order to ensure
uniformity (‘stadium where their match is to be played’ = ‘their venue’)
Training grounds
- -
11.12 UEFA offers each UEFA offers each association -
9. Annexes
145
S1 C1 Rules applied
association a number of pre-selected training grounds.
taking part in the final tournament a number of pre-selected training grounds.
Should
an association choose a training ground other than those which have been
preselected, then the association concerned bears all the costs
incurred.
Should an association choose
a training ground that is not part of this preselection, the association bears all the
costs that are incurred.
- General: Be as explicit
and precise as possible (‘other than those which have been peselected’=’ that is not part of
this preselection’) - General: avoid redundancies (‘concerned’;
‘then’) - Grammar: do not omit relative pronouns (‘the costs
incurred’ = ‘the costs that are incurred’)
11.13 In any case, any and all training grounds used by
associations are referred to as “official” from five days before
the first match of the final tournament and the provisions laid out in
paragraph 28.14 apply.
11.13 As of the fi fth day
before the first match of the final tournament, any and all training grounds that are used
by the associations are referred to as official training grounds. From that moment,
the provisions of paragraph 28.14 are applied.
- General: order the parts
of the sentences logically - Grammar: do not omit relative pronouns (‘that’)
- General: Be as explicit and precise as possible
(‘official’=’official training grounds’) - General: avoid
redundancies (‘laid out in=’of’) - Lexis: Avoid the active form of the verb apply
- General: Avoid long and complex sentences
9. Annexes
146
Annex F: MT-oriented writing rules for UEFA regulations1
I General
1. Avoid long and complex sentences, if possible:
The original trophy, which is used for the official presentation ceremony at the final and at other official events approved by UEFA, remains in UEFA’s keeping and ownership at all times. vs. The original trophy is used for the official trophy handover at the final
and for other official events that are approved by UEFA. It remains in the keeping and ownership of UEFA at all times.
2. Order the parts of the sentences logically:
Begin a sentence with the subject (subject-verb-object), when possible, in order to respect the topic-comment structure2: The following two dates are reserved for the play-off matches between the remaining eight runners-up: vs. The play-off matches between the remaining eight runners-up
are scheduled for the following two dates:
3. Every segment has to be independent on its own:
Make sure that every segment can stand alone syntactically, i.e. avoid splitting up sentences into parts that do not make sense without a preceding part:
This rule applies if
more than one club from the same country…
is seeded for… 4. Avoid overly complex constructions and write clearly:
The association in the territory of which a qualifying match or the final tournament is being staged vs. The national association that hosts a qualifying match or the final
tournament on its territory 5. Be as explicit and precise as possible:
For example, write group matches instead of matches when there is talk of group matches, or write the final tournament instead of the next stage when the final tournament is meant. 6. Simplify sentence structures and expressions whenever possible:
if both associations concerned are in agreement vs. if both associations in
question agree
7. Avoid redundancies3:
the length of the journey to be undertaken by the visiting association vs. the length of the journey of the visiting association
each association taking part in the final tournament vs. the teams
8. Avoid omissions, even if they are easy to presuppose for humans:
1 These writing rules are partly based on the controlled language rules elaborated by
(Bernth and Gdaniec 2001), (T. Mitamura 1999) and (T. et al. Mitamura 2003). 2 Topic = already familiar or mentioned piece of information, comment=newly introduced
piece of information 3 This rule seems to contradict rule no. 4 (“be explicit and precise”). The author (or post -
editor) of a text has to decide on his own about the rules he judges appropriate in every specific case.
9. Annexes
147
30 days after the draw in which to agree on the order of their matches vs. 30 days
after the draw in which they have to agree on the order of their matches
9. Try to avoid synonyms or different paraphrases expressing the same content in order to ensure uniformity: Try to use constantly the same expressions instead of varying with different paraphrases. Even if both versions of the following examples are correct, try to avoid variations and decide for one solution for all regulations.
in a stadium within the territory of the host association vs. in a stadium
on the territory of the respective host association goals scored away from home vs. away goals
the same number of goals over the two legs vs. the same number of
aggregate goals with the following criteria being taken into consideration vs. the
following criteria are applied higher number of goals scored in all group matches vs. higher number
of goals scored in all the group matches higher number of goals scored away from home in all group matches
vs. higher number of away goals scored in all the group matches fair play ranking in all group matches; vs. fair play ranking in all the
group matches; two 15-minute periods of extra time vs. an extra time of two 15-minute
periods
if this procedure does not produce a result vs. If this procedure does
not lead to a decision
kicks from the penalty mark […] determine which team qualifies for the
next stage vs the team that qualifies for the next stage is determined by kicks
from the penalty mark […]
Each team plays each of the other teams in the same group vs. Each
team plays against every other team in its group
The first-named team is considered as the home team vs. The first-
named team is considered the home team.
the matches among the teams in question vs. the matches played
among the teams in question
pre-match press conference before the media deadlines of both involved countries vs. pre-match press conference before the media deadlines in both involved countries
10. Make a uniform use of singular and plural: The venues of the matches are fixed by the host associations vs. The venue of a
match is fixed by the host association
II Lexis
11. Avoid noun clusters, try to explain them: full-size replica trophy vs. replica of the trophy that is as large as the original
12. Avoid the determiner any:
Only use any in negative sentences: associations must not develop, create, use, sell or distribute any promotional materials or any merchandise
9. Annexes
148
In all other cases, specify using any kind or substitute for another determiner, e.g. a or the: must comply with any trophy use guidelines vs. must comply with trophy use
guidelines of any kind any association which wins vs. an association which wins
Exception: typical idiomatic constructions like any further, of any kind, any kind of, etc.
13. Avoid the ambiguous modal verb may:
Clearly specify its sense and use “can”, “must”, “have to”, etc. instead. Exceptionally, matches may be played in the territory of another UEFA
member association vs. Exceptionally, matches can be played on the territory
of another UEFA member association Associations may not […] develop, create, use, sell or distribute […] vs.
Associations must not develop, create, use, sell or distribute […]
14. Avoid the term association when it does not designate a football association:
In this case, use the term link instead. 15. Avoid the active form of the verb apply
Prefer the passive construction to be applied: the criteria listed in paragraph 7.02 apply vs. the criteria of paragraph 7.02 are applied
16. Avoid redundant conjunctions such as ‘also’, ‘as well’, etc.
If necessary, prefer the use of conjunctions at the beginning of a sentence (moreover, furthermore, additionally, etc.) 17. Only use the combination ‘against each other’ when it means ‘against one another’.
In any other case, use ‘against every other’, e.g. against every other team in the group.
III Grammar
18. Transform passive into active constructions: is awarded to vs. receives
19. Avoid the use of ing-words, if possible:
use articles with ing-words when they are used as nouns responsible for drawing up the match schedule vs. responsible for the drawing up
of the match schedule avoid ing-words when they are used as verbs
the associations participating in the final tournament vs. the associations that
participate in the final tournament exception: typical idioms like including, without limitation or following,
according to
20. Use determiners (the, a) wherever possible If this procedure does not lead to a decision, criteria f) to j) apply vs. If this procedure
does not lead to a decision, the criteria f) to j) are applied;
Exception: Do not necessarily use determiners in titles and subtitles
9. Annexes
149
21. Avoid future tense: If future tense is inevitable, prefer the ‘will’-future tense; avoid the ‘be to’-future tense: the UEFA administration is entitled to order matches within a group to be played at the same time vs. the UEFA administration can demand that matches within the same
group are played at the same time. 22. Avoid parenthesis: Try not to separate sentence parts that logically belong together:
Higher number of coefficient points accumulated by the club in question , as well as its association, over the previous five seasons vs. Higher number of coefficient points
accumulated by the club in question and the association of the club over the previous five seasons Matches must, in principle, be played in a stadium within the territory of the host association. vs. In principle, matches must be played in a stadium on the territory of
the respective host association. 23. Do not coordinate verb phrases:
Try to repeat the verb in order to disambiguate the meaning: Forty gold medals are presented to the winning team and 40 silver medals to the runner-up vs. Forty gold medals are presented to the winning team and 40 silver
medals are presented to the runner-up. 24. Avoid possessive case (´s) and possessive pronouns, if possible:
Prefer repeating the subject, even when the term has already been used in the sentence.
remains in UEFA’s keeping and ownership at all times vs. remains in
the keeping and ownership of UEFA at all times on their behalf vs. on behalf of the association
25. Minimize the use of personal pronouns (they, it, etc.)
For example, prefer repeating the association instead of using the personal pronoun it, even when the term has already been used in the sentence. 26. Do not omit relative pronouns, even though this seems to be redundant for humans:
Write that, which, who, etc. explicitly 27. Repeat the preposition in conjoined constructions where appropriate, even though this seems to be redundant for humans: Same number of goals in a quarter-final, semi-final or the final vs. Same number of
goals in a quarter-final, in a semi-final or in the final. .
28. Try to align prepositions that refer to the same verb, if possible:
The original trophy, which is used for the official presentation ceremony at the final and at other official events vs. The original trophy is used for the official trophy handover at
the final and for other official events 29. Do not reduce relative clauses, always write the expanded form, even though this seems to be redundant for humans: standard fixture list drawn up by the UEFA administration vs. standard fixture list that
is drawn up by the UEFA administration
30. Avoid phrasal constituents that are added after a comma:
9. Annexes
150
The winners and runners-up in each group play the quarter-finals over one match, as follows: vs. The winners and runners-up in each group play the quarter-finals over one
match. For this purpose, the following scheme is applied:
9. Annexes
151
Annex G: C1
III Trophy. Plaques and Medals Article 3 Trophy
3.01 The original trophy is used for the official trophy handover at the final and for other official events that are approved by UEFA. The original trophy remains in the keeping and ownership of UEFA at all times. The winning association receives a full-size replica trophy, the winner's trophy of the UEFA European Football Championship. 3.02 An association which wins the trophy three consecutive times or which wins it five times in total receives a special mark of recognition. Once an association completes a cycle of three successive wins or a cycle of five wins in total, the association starts a new cycle from zero. 3.03 The relevant associations must at all times remain in possession of the replica trophies that the past and the current winners of the UEFA European Football Championship have received. The replicas must not leave the country of the relevant association without the prior written consent of UEFA. The associations must not approve the use of a replica trophy in a context where a third party is granted visibility. Third parties include, without limitation, sponsors of the associations as well as other commercial partners. The associations must not approve the use of a replica trophy in a way that could create a link between a third party and the replica trophy and/or the competition. The associations must comply with trophy use guidelines of any kind that the UEFA administration can issue from time to time. 3.04 Associations must not develop, create, use, sell or distribute any promotional materials or any merchandise that bear a representation of the trophy or a replica of the trophy (including, without limitation, images that show the winners lifting the trophy). Associations must not use such a representation in a manner that could lead to a link between a third party and the trophy, the replica trophy and/or the competition. Furthermore, they must not permit a third party to do any of the above.
Commemorative plaques
3.05 Each association that competes in the final tournament receives a commemorative plaque.
Semi-finalist plaque
3.06 Each defeated semi-finalist receives a plaque.
Final plaque
3.07 Each finalist receives a plaque.
Medals
3.08 The winning team receives 40 gold medals. The runner-up receives 40 silver medals. The defeated semi-finalists each receive 40 bronze medals. Additional medals will not be produced.
IV Responsibilities Article 4
Responsibilities of the associations
4.01 An association is responsible for the behaviour of its players, its officials, its members, its supporters and every person who carries out a function at a match on behalf of the association.
9. Annexes
152
4.02 The national association that hosts a qualifying match or the final tournament on its territory is considered the host association for that match or tournament. 4.03 The host association is responsible for order and security before, during and after the matches. The host association can be called to account and be disciplined for incidents of any kind. 4.04 In principle, matches must be played in a stadium on the territory of the respective host association. Exceptionally, matches can be played on the territory of another UEFA member association, if the UEFA administration and/or the disciplinary bodies decide so (for reasons of safety or as a result of a disciplinary measure). Additional responsibilities for the final tournament 4.05 The UEFA administration informs the associations that participate in the final tournament about any further guidelines, directives, or decisions that are related to the final tournament. The UEFA administration provides the associations with all relevant documents in due time.
VI Competition System Article 6
Competition stages
6.01 The competition consists of a qualifying competition and of a final tournament.
Article 7 A. Qualifying competition
Group formation
7.01 The teams of the host associations for the final tournament, Poland and Ukraine, qualify automatically for the final tournament. The remaining teams are drawn into 6 groups of 6 teams and 3 groups of 5 teams. The UEFA administration forms the groups by means of a draw, on completion of the qualifying competition of the 2010 FIFA World Cup. The draw is based on a seeding system. The reigning European champion is always seeded. The remaining associations are classified on the basis of the UEFA national team coefficient ranking system (see Annex I, paragraph 1.2.1). The decisions of the UEFA administration are final. 7.02 If two or more of these teams have the same coefficient, the following criteria are determinant. Only the most recent qualifying competition is taken into consideration. The criteria are applied in the order given: a) UEFA national team coefficient resulting from the matches played; b) average goal difference; c) average number of goals scored; d) average number of away goals scored; e) fair play ranking; f) drawing of lots.
Match system for the qualifying competition
7.03 The matches in the qualifying competition are played in groups according to the league system. Each team plays against every other team in its group in a series of home and away matches. Three points are awarded for a win, one point for a draw, and no points for a defeat. Equality of points after the group matches 7.04 If two or more teams are equal on points on completion of the group matches, the following criteria are applied to determine the rankings. They are applied in the order given: a) higher number of points that the teams obtained in the group matches played among the teams in question;
9. Annexes
153
b) superior goal difference from the group matches played among the teams in question; c) higher number of goals scored in the group matches played among the teams in question; d) higher number of away goals scored in the group matches played among the teams in question; e) If the criteria a) to d) have been applied to several teams and two or more teams still have an equal ranking, the criteria a) to d) are reapplied to determine the ranking of these teams. If this procedure does not lead to a decision, the criteria f) to j) are applied; f) superior goal difference in all group matches; g) higher number of goals scored in all the group matches; h) higher number of away goals scored in all the group matches; i) fair play ranking in all the group matches; j) drawing of lots.
Qualification for the final tournament
7.05 The nine group winners and the best runner-up qualify directly for the final tournament. 7.06 Only the results against the teams in first, third, fourth and fifth place are taken into account to determine the best runner-up. For this purpose, the following criteria are applied in the order given: a) higher number of points that the teams obtained in these matches; b) superior goal difference from these matches; c) higher number of goals scored in these matches; d) higher number of away goals scored in these matches; e) position in the UEFA national team coefficient ranking system (see Annex I, paragraph 1.2.2); f) fair play ranking in these matches; g) drawing of lots. 7.07 The eight remaining runners-up contest play-off matches. The four ties are determined by means of a draw. The four runners-up with the best position in the UEFA national team coefficient ranking system (see Annex I, paragraph 1.2.2) are seeded for the draw. If two or more of these teams have the same coefficient, the criteria of paragraph 7.02 are applied. The play-offs are played according to the knockout system (in home and away matches). The seeded teams play the return match at home. The teams which score the greater aggregate of goals qualify for the final tournament. Otherwise, the provisions of paragraph 7.08 are applied. 7.08 For matches that are played under the knockout system, the following criterion is applied: if both teams score the same number of aggregate goals, the team which scores more away goals qualifies for the next stage. If this procedure does not lead to a decision, i.e. if both teams score the same number of home and away goals, an extra time of two periods of 15 minutes is played at the end of the return match. If, during extra time, both teams score the same number of goals, the away goals count double (i.e. the visiting team qualifies). If no goals are scored during extra time, the team that qualifies for the final tournament is determined by kicks from the penalty mark (see Article 16).
Article 8 B. Final tournament
8.01 The UEFA Executive Committee has entrusted the Polish Football Association (PZPN) and the Football Federation of Ukraine (FFU) with the joint organisation and staging of the final tournament. Group formation
9. Annexes
154
8.02 Sixteen teams qualify for the final tournament. The UEFA administration divides these 16 teams into four groups of four teams each (groups A, B, C, D). 8.03 By means of a draw, the four groups are formed as follows: Group A Group B Group C Group D A1 B1 C1 D1 A2 B2 C2 D2 A3 B3 C3 D3 A4 B4 C4 D4
Coefficients
8.04 The following teams are seeded: the host associations, the reigning European champion if it qualifies, and one or two teams with the best coefficients according to the UEFA national team coefficient ranking system (see Annex I, paragraph 1.2.2).The other finalists are assigned to the four groups by means of a draw (according to their coefficients). 8.05 If two or more of the teams in question have the same coefficient, the criteria of paragraph 7.02 are applied. The results from the play-offs (see paragraph 7.07) are thereby not taken into consideration. Group match schedule 8.06 Each team plays against every other team in its group according to a league system (three points for a win, one point for a draw, no points for a defeat). The group matches are played according to the following schedule. The last two matches in each group must both kick off at the same time. The first-named team is considered the home team. Match day 1 Match day 2 Match day 3
Group A A1 vs A2 A1 vs A3 A4 vs A1 A3 vs A4 A2 vs A4 A2 vs A3
Group B B1 vs B2 B1 vs B3 B4 vs B1 B3 vs B4 B2 vs B4 B2 vs B3
Group C C1 vs C2 C1 vs C3 C4 vs C1 C3 vs C4 C2 vs C4 C2 vs C3
Group D D1 vs D2 D1 vs D3 D4 vs D1 D3 vs D4 D2 vs D4 D2 vs D3
Equality of points after the group matches
8.07 If two or more teams are equal on points on completion of the group matches, the following criteria are applied to determine the rankings. They are applied in the order given: a) higher number of points that the teams obtained in the matches played among the teams in question; b) superior goal difference from the matches played among the teams in question (if more than two teams are equal on points on completion of the group matches); b) superior goal difference in the matches among the teams in question (if more than two teams finish equal on points); c) higher number of goals scored in the matches played among the teams in question (if more than two teams are equal on points on completion of the group matches); d) superior goal difference in all the group matches; e) higher number of goals scored in all the group matches; f) position in the UEFA national team coefficient ranking system (see Annex I, paragraph 1.2.2);
9. Annexes
155
g) fair play conduct of the teams in question (final tournament); h) drawing of lots. 8.08 Two teams have the same number of points and the same goal difference. These teams play their last group match against each other. If this match ends in a draw, the ranking of the two teams in question is determined by kicks from the penalty mark (see Article 16; provided that no other teams within the same group have the same number of points on completion of the group matches). Should more than two teams have the same number of points, the criteria of paragraph 8.07 are applied.
Quarter-finals
8.09 The winners and runners-up in each group play the quarter-finals over one match. For this purpose, the following scheme is applied: Match 1 Winner Group A vs Runner-up Group B Match 2 Winner Group B vs Runner-up Group A Match 3 Winner Group C vs Runner-up Group D Match 4 Winner Group D vs Runner-up Group C
Semi-final
8.10 The four winners of the quarter-finals play the semi-finals over one match. For this purpose, the following scheme is applied: Winner Match 1 vs Winner Match 3 Winner Match 2 vs Winner Match 4
Final
8.11 The winners of the semi-finals play in the final.
Same number of goals in a quarter-final, in a semi-final or in the final
8.12 If the match ends in a draw after normal playing time, an extra time of two periods of 15 minutes is played. If the extra time ends in a draw, the winner is determined by kicks from the penalty mark (see Article 16).
VIII Match Dates. Venues and Kick-off Times
Article 11
11.01 The competition takefors place following the 2010 FIFA World Cup. It is staged over two seasons.
A. Qualifying competition Match dates
11.02 The group matches in the qualifying competition of the 2010-12 UEFA European Football Championship are scheduled for the following twelve dates: 2010 a) 3/4 and 7 September 2010 b) 8/9 and 12 October 2010 2011 c) 25/26 and 29 March 2011 d) 3/4 and 7 June 2011 e) 2/3 and 6 September 2011 f) 7/8 and 11 October 2011 (dates for the last matches in all groups)
9. Annexes
156
11.03 The play-off matches between the remaining eight runners-up are scheduled for the following two dates: a) 11/12 November 2011 b) 15 November 2011 11.04 Under the following conditions, the group matches mentioned in paragraph 11.02 can be played on other dates : if both associations in question agree and if the principles governing the release of players for association teams as laid out in Annex 1, Article 1 of the FIFA Regulations for the Status and Transfer of Players are complied with. The associations in a group have 30 days after the draw in which they have to agree on the order of their matches. The exact date of each match must be specified (e.g. Saturday 4 September 2010). If the associations in question are unable to reach agreement, the matches will take place according to a standard fixture list that is drawn up by the UEFA administration. This standard fixture list takes into consideration factors that are relevant for the match organisation, such as climate. 11.05 For reasons of sporting fairness, the UEFA administration can demand that matches within the same group are played at the same time. Subsequent changes of date are subject to the approval of the UEFA administration. The host association of a match must, in this case, also inform the other associations in the same group.
Venues and kick-off times
11.06 The match venues are fixed by the host associations. They have to be announced to the visiting associations and to the UEFA administration at least 60 days in advance. When a host association fixes a venue, it must take into account the length of the journey of the visiting association. The venue for a qualifying match must be no more than a two-hour bus drive from the nearest international airport, unless the visiting association agrees otherwise. The kick-off times must be announced to the UEFA administration at least 60 days before the match. Arrival of the teams at the match venue 11.07 The associations must arrange that their teams arrive early enough at the match venue so that they can hold their pre-match press conference before the media deadlines in both involved countries. In any case, the teams must arrive at the match venue at least 24 hours before kick-off.
B. Final tournament Match dates
11.08 The final tournament will take place from 8 June to 1 July 2012. Venues and kick-off times 11.09 The UEFA administration is responsible for the drawing up of the match schedule for the final tournament. The teams must have at least 48 rest hours between each match. Arrival of the teams in the host countries 11.10 The teams must arrive at their team hotel in one of the host countries at least five days prior to their first match. Arrival of the teams at the match venue 11.11 The teams must arrive at their transfer hotel at least 24 hours before kick-off or be within a radius of 60km of their venue at least 24 hours before kick-off.
Training grounds
11.12 UEFA offers each association taking part in the final tournament a number of pre-selected training grounds. Should an association choose a training ground that is not part of this preselection, the association bears all the costs that are incurred. 11.13 As of the fifth day before the first match of the final tournament, any and all
9. Annexes
157
training grounds that are used by the associations are referred to as official training grounds. From that moment, the provisions of paragraph 28.14 are applied.
9. Annexes
158
Annex H: Table comparing S2 to C2
Subject 1:
S2 (part 1)
C2 (part 2)
Rules applied
Rules not applied;
errors
III Trophies and Medals Article 5
Trophy
The original
trophy, which is used for the official
presentation ceremony at the final, remains in
UEFA's keeping at all times. A full-size replica
trophy, the UEFA Champions League winners
trophy, is awarded to the winning club.
Any club which
wins the competition three consecutive
times or five times in total receives a
special mark of recognition. Once a cycle of three
successive wins or five in total has been completed,
the club concerned starts a new cycle from
zero.
Replica t rophies awarded to winners of the
UEFA Champions League (past and
current) must remain within the relevant club’s
control at all times and must not leave its
III Trophies and Medals Article 5
Trophy
The original trophy,
which is used for the official presentation ceremony at the
final, remains in the keeping of UEFA at all times. The
winning club receives the winners trophy of the UEFA
Champions League winners trophy, a replica of the trophy
that is as large as the original.
A club which wins
the competition three consecutive times or which wins it five
times in total receives a special mark of recognition.
Once a cycle of three successive wins or five in total has been
completed, the club concerned starts a new cycle from zero.
The rReplica trophies that are awarded to the
winners of the UEFA Champions League (past and current)
receive must remain within the relevant control of the club at
all times. Furthermore, the replica trophies and
- Grammar: Use determiners (the,
a) wherever possible - Grammar: Avoid
possessive case (´s) and possessive
pronouns - Grammar: Transform
passive into active - Lexis: avoid
noun clusters
- Lexis: Avoid the
determiner any
- Grammar: Avoid
possessive case (´s) and possessive
pronouns (2x)
- Lexis: avoid noun
clusters (‘UEFA Champions League winners trophy’)
- Grammar: Do not omit
relative pronouns
- Grammar: Use determiners (the, a) wherever possible (2x)
- Grammar: Transform passive into active - General: Avoid long
and complex sentences - Grammar: Avoid possessive case (´s)
and possessive pronouns (‘its’)
9. Annexes
159
region or the
country of its association without UEFA’s
prior written consent.
Clubs must not permit a replica
trophy to be used in any context where a third
party (including, without limitation, their sponsors
and other commercial partners) is
granted visibility or in any other way which could
lead to an association between any
third party and the trophy and/or the competition.
Clubs must comply with any trophy use
guidelines that may be issued by the UEFA
administration from time to time.
Medals
a. Thirty gold medals are presented to the
winning club, and 30 silver medals to the runner-up.
Additional medals may not be produced.
must not leave its
the region or the country of its the relevant association
without the prior written consent of the UEFA.
Clubs must not permit a replica
trophy that is used in a context where a third party (including,
without limitation, the sponsors of the club and other
commercial partners) is granted visibil ity or in another way which
could lead to a link between a third party and the trophy
and/or the competition. Clubs must comply with
trophy use guidelines of any kind that are issued
by the UEFA administration from time to time.
Medals
t. The winning club
receives thirty gold medals and the runner-up receives
30 silver medals. Additional medals must not be
produced.
- General: Avoid overly complex
constructions and write clearly - Lexis: Avoid the
determiner any (2x) - Lexis: Avoid the
ambiguous modal verb may - Grammar: Avoid
possessive case (´s) and possessive
pronouns (‘their’) - Lexis: avoid ‘association’
when it does not refer to a football association
- Grammar: Transform
passive into active - Lexis: Avoid the
ambiguous modal verb may
9. Annexes
160
IV
Responsibilities
Article 6
UEFA responsibilities
b. UEFA insures its own area of responsibi lity in
accordance with the present regulations:
- third-party
liability
insurance
- spectator
accident
insurance (for
the final only)
- group
accident
insurance for
UEFA
delegates
- legal
expenses
insurance
(restricted to
criminal
matters).
Responsibilities of the
associations and clubs
c. The clubs are responsible for the behaviour of
their players, officials, members,
supporters and any person carrying out a
function at a match on their behalf.
d. The home club (or the host
association) is responsible for order and
security before, during and after
IV Responsibilities
Article 6
UEFA
responsibilities
u. UEFA insures its
own area of responsibi lity in accordance with the
present regulations:
- third-party liability
insurance
- spectator
accident
insurance (for the
final only)
- group accident
insurance for
UEFA delegates
- legal expenses
insurance
(restricted to
criminal matters).
Responsibilities of
the associations and clubs
v. The clubs are responsible for the behaviour of their
players, their officials, their members, their
supporters and each person that carries out a function at a
match on the behalf of the club.
w. The home club (or
the host association) is responsible for order and security
before, during and after the match. The home club (or the
- Lexis: Avoid the
determiner any - Grammar: Avoid possessive
case (´s) and possessive pronouns (‘their
behalf’) - Grammar: avoid the use of ing-
words
- General: avoid
omissions (‘their’)
9. Annexes
161
the match. The
home club (or the host association) may be called to
account for incidents of any kind and may be
disciplined.
e.
f. The club considered the
home club must stage the relevant matches
at the ground in accordance with the instructions of
UEFA (or of a third party acting on UEFA’s
behalf) and in cooperation with the association
concerned. However, the club is considered solely
accountable for all of its obligations in this
respect, unless the relevant body or bodies
decide(s) otherwise.
g. Irrespective of UEFA’s
insurance coverage, each club and host
association must conclude insurance
coverage with reputable insurers at their
own cost, in relation to any and all risks,
according to the following principles:
i. each club must
conclude and maintain
host association) can
be called to account for incidents of any kind and has tocan
be disciplined.
x. The club which is considered the home club must stage the
relevant matches at the ground in accordance with the
instructions of UEFA; the instructions can also be issued by (or
of a third party that acts on the behalf of UEFA). The relevant
club must furthermore stage the relevant matches
and in cooperation with the association concerned. However,
the club is considered solely accountable for all of
its obligations in this respect, unless the relevant body or
bodies decide(s) otherwise.
y. Irrespective of the insurance coverage
of the UEFA, each club and host association must
conclude an insurance coverage with reputable
insurers at their own cost, in relation to any and all risks. For
this purpose, the following principles are applied:
i. each club must conclude and
maintain insurance coverage to fully
- Lexis: Avoid the ambiguous modal verb may
- Grammar: avoid the use of ing-
words
-Grammar: Avoid possessive case
(´s) and possessive pronouns
- Grammar: Use determiners (the, a) wherever
possible - General: Avoid long and complex
sentences - Lexis: Avoid the active form of the
verb apply
-Semantic error: ‘can be disciplined’
- Grammar: Do not omit relative pronouns (‘which’)
- General: Avoid long and complex sentences
Grammar mistake (‘the UEFA’)
9. Annexes
162
insurance
coverage to ful ly cover all of its risks in
connection with its participation in the competition;
ii.
iii. in addition, the home club or the host association
must conclude and maintain insurance
coverage for the risks in connection with
staging and organising its home matches
and which must include, without limitation, third-
party liability insurance (for all third parties participating in
matches or attending the relevant venue)
providing for appropriate guaranteed sums
for damages to persons, objects and property, as
well as for pure economic losses corresponding to
the specific circumstances of the club or
association concerned;
iii. to the same
extent as in paragraph b) above, the host
association of the final match must conclude and
maintain insurance coverage to ful ly
cover all of its risks in connection with
the staging and
cover all of its risks
in connection with its participation in the competition;
ii. in addition, the home club or the host association must
conclude and maintain an insurance coverage
for the risks in connection with staging and
organising its home matches. This must include, without
limitation, third-party liability insurance (for all third parties that
participate in matches or attend the relevant venue).
This third-party liability insurance must providinge for
appropriate guaranteed sums for damages to persons,
objects and property. Furthermore, the third-party liability
insurance must provide , as well as for pure economic
losses which corresponding to the specific
circumstances of the club or association concerned;
iii.
iv. to the same extent as in paragraph b) above, the host
association of the final match must conclude and
maintain insurance coverage to fully cover all of its risks
in connection with the staging and organisation of the
final match;
z.
- Grammar: Use
determiners (the, a) wherever possible
- Grammar: avoid the use of ing-words
- General: Avoid long and complex sentences (3x) - Grammar: avoid the
use of ing-words (2x)
9. Annexes
163
organisation of
the final match;
iv.
v. if the home club or the host
association is not the owner of the stadium used, it
is also responsible for providing
adequate and fully comprehensive
insurance cover, including third-party liability and
property damage, taken out by the relevant stadium
owner and/or tenant;
h.
v. the home club and the host
association must ensure that UEFA is included
in all insurance policies as defined in the
present paragraph and must hold UEFA
harmless from any and all claims for liability
arising in relation to the staging and organising of
the relevant matches;
In any case, UEFA may ask
anyone involved to provide, free of charge, written
releases of liability and/or hold harmless
notes, and/or confirmations
i. if the home club or the host association is not the owner of
the stadium used, it is also responsible for the providing of
an adequate and fully comprehensive insurance cover. The
relevant stadium owner and/or tenant has to conclude and
maintain this insurance cover. The insurance cover has
to include, including third-party liability insurance and
property damage insurance;, taken out by the relevant
stadium owner and/or tenant;
ii.
iii. the home club and
the host association must ensure that UEFA is included in
all insurance policies as defined in the present paragraph.
Furthermore, and they must hold UEFA harmless from any
and all claims for liability arising that arise in relation to
the staging and organising of the relevant matches;
In any case, UEFA
can ask anyone involved to provide, free of charge,
written releases of liability and/or hold harmless notes,
and/or confirmations and/or copies of the policies concerned in
one of the official
- Grammar: Use determiners (the,
a) wherever possible (2x)
- Lexis: Avoid the ambiguous modal
verb may -Grammar: Avoid possessive case
(´s) and possessive pronouns
- General: Avoid long and complex sentences
- Grammar: avoid the use of ing-words - General: avoid
omissions -Grammar: transform passive into active
- General: Avoid long and complex sentences
- Grammar: avoid the use of ing-words
Grammar mistake (‘the UEFA’)
9. Annexes
164
and/or copies of
the policies concerned in one of UEFA’s official
languages.
i. The clubs
undertake that their team will arrive at the
match venue by the evening before the match
at the latest and to fulfil their media obligations
the day before the match.
j. Visiting clubs undertake not to
play any other matches when travelling to and
from away matches in this competition.
V Competition
System
Article 7
Number of rounds
k. As shown in Annex Ib, the
competition consists of:
i. a) the qualifying phase for the UEFA
Champions League:
- first qualifying
round
- second
qualifying round
- third qualifying
round
ii. b) play-offs
iii. c) the UEFA Champions League:
- group stage (six
languages of the
UEFA.
aa. The clubs undertake that their team arrives at the match
venue by the evening before the match at the latest .
The clubs undertake and to fulfil their media obligations
the day before the match.
bb. VThe visiting clubs
undertake not to play other matches when they travel to and
from away matches in this competition.
V Competition System
Article 7
Number of rounds
cc. As shown in Annex Ib, the competition consists of:
i. a) the qualifying phase for the UEFA
Champions League:
- first qualifying
round
- second qualifying
round
- third qualifying
round
ii. b) play-offs
iii. c) the UEFA
Champions League:
- group stage (six
matchdays)
- round of 16
- quarter-finals
- semi-finals
- final.
-Grammar: avoid future tense
- Lexis: Avoid the determiner any
- General: Avoid long and complex sentences
- Grammar: Use determiners (the, a) wherever possible
9. Annexes
165
matchdays)
- round of 16
- quarter-finals
- semi-finals
- final.
Qualifying phase
l. Qualifying-phase matches are
played according to the cup (knockout)
system, with each club playing each opponent
twice, in home and away matches. The
team which scores the greater
aggregate of goals in the two matches qualifies
for the next stage (second qualifying round,
third qualifying round or play-offs, as
applicable). Otherwise, the stipulations of
Article 8 apply. The clubs defeated in the
first and second qualifying rounds are eliminated
from the competition. The clubs defeated in
the third qualifying round are entitled to
play in the play-offs of the UEFA Europa League
in progress. Clubs from the same association
cannot be drawn against each other.
Qualifying phase
dd. Qualifying-phaseThe matches of the qualifying phase are
played according to the cup (knockout) system (or knockout
system),; with each club playings each opponent twice, in
home and away matches. The team which scores the
greater aggregate of goals in the two matches qualifies for
the next match stage (second qualifying round, third
qualifying round or play-offs, as applicable).
Otherwise, the stipulations of Article. 8 are applied.
The clubs that were defeated in the first and second
qualifying rounds are eliminated from the competition. The
clubs that were defeated in the third qualifying round are
entitled to play in the play-offs of the UEFA Europa
League in progress. Clubs from the same association cannot
be drawn against each other.
- Lexis: Avoid the
active form of the verb apply - Grammar: Do
not omit relative pronouns
- Semantic error: ‘the
next stage’ is not ‘the next match’, but the stages mentioned in
brackets - Lexis: avoid noun clusters
- New rule: Avoid parenthesis - General: avoid
omissions - Lexis: avoid the use of ing-words
-General: avoid long and complex sentences
9. Annexes
166
Play-offs
m. Play-off matches
are played according to the cup (knockout)
system, with each club playing each opponent
twice, in home and away matches. The
team which scores the greater
aggregate of goals in the two matches qualifies
for the UEFA Champions League group
stage. Otherwise, the stipulations of Article 8 apply.
The clubs defeated in the play-offs are
entitled to play in the group stage of the UEFA
Europa League in progress. Clubs from the
same association cannot be drawn against each
other.
Group stage
n. Once the play-offs have been completed, the
32 remaining clubs are drawn into eight groups
of four. Clubs from the same association
cannot be drawn into the same group.
o. Each club plays one home and
one away match against each other club in its
group. Three
Play-offs
ee. The Pplay-off
matches are played according to the cup (knockout) system:,
with each club playsing each opponent twice, in
home and away matches. The team which scores the
greater aggregate of goals in the two matches qualifies for
the UEFA Champions League group stage.
Otherwise, the stipulations of Article 8 are applied. The
clubs defeated in the play-offs are entitled to play in the group
stage of the UEFA Europa League in progress. Clubs from the same association
cannot be drawn against each other.
Group stage
ff. Once the play -offs
have been completed, the 32 remaining clubs are
drawn into eight groups of four teams. Clubs from the same
association cannot be drawn into the same group.
gg. Each club plays one home match and one
away match against each every other club in its group.
Three points are awarded for a win, and one point is
awarded for a draw, and none point is
- Lexis: Avoid the
active form of the verb apply
-Lexis: Avoid the active form of the
verb apply
-General: avoid
omissions - General: avoid long and complex sentences
- Lexis: avoid the use of ing-words
-avoid omissions (2x)
-new rule: Lexis: Only use the combination “against each other” when it means “against one another”. - grammar mistake: ‘none’ -style mistake ‘and’
9. Annexes
167
points are
awarded for a win, one point for a draw, and none
for a defeat. The following match sequence
applies:
p. [match
sequence]
q.
If two or more teams are equal
on points on completion of the group matches,
the following criteria are applied to
determine the rankings (in descending
order):
i. higher number of
points obtained in the group matches played among the teams
in question;
ii. superior goal
difference from the group matches played
among the teams in question;
iii. higher number of goals scored in the group
matches played among the teams in question;
iv. higher number of goals scored
away from home in the group matches played
among the teams in question;
v.
vi. if, after having
applied criteria a) to d), two teams still have an
equal ranking, criteria a) to d) are reapplied to
determine the
awarded for a
defeat. The following match sequence is applied:
[match sequence]
4 v 3
hh. If two or more teams are equal on points
on completion of the group matches, the following criteria are
applied to determine the rankings (in descending order):
i. higher number of points obtained in
the group matches played among the teams in question;
ii. superior goal difference from the
group matches played among the teams in question;
iii. higher number of goals scored in the
group matches played among the teams in question;
iv. higher number of away goals scored in
the group matches played among the teams in question;
v. if, after having
applied the criteria a) to d), have been applied and two
teams still have an equal ranking. In this case,, the criteria a)
to d) are reapplied to determine the final ranking of the two
teams. If this
- General: Try to avoid synonyms or different
paraphrases expressing the same content in
order to ensure uniformity
- General: avoid long
and complex sentences - Grammar: Use determiners (the, a)
wherever possible
9. Annexes
168
final ranking of
the two teams. If this procedure does not lead to
a decision, criteria f) to h) apply;
vii. superior goal difference from
all group matches played;
r.
i. higher number of goals scored from all group
matches played;
ii. higher number of coefficient points accumulated by
the club in question, as well as its
association, over the previous five seasons (see
paragraph 9.02).
s. The eight group-winners and eight runners-up of the
group stage qualify for the round of 16. The
clubs that finish this stage in third position in their
group move into the round of 32 of the current UEFA
Europa League. The clubs that finish this stage
in fourth position in their group are eliminated.
procedure does not
lead to a decision, the criteria f) to h) are applied;
vi.
vii. superior goal difference from all the group matches
played;
viii. higher number of
goals scored from all the group matches played;
ix.
x. higher number of coefficient points accumulated bythat
the club in question and the association of the club have
accumulated, as well as the association of the club, over the
previous five seasons (see paragraph 9.02).
xi.
ii. The eight group-winners and the eight runners-up of
the group stage qualify for the round of 16. The clubs that
finish this stage in the third position in their group move into
the round of 32 of the current UEFA Europa League. The
clubs that finish this stage in the fourth position in their
group are eliminated.
-Lexis: Avoid the
active form of the verb apply
- Grammar: Use determiners (the, a) wherever
possible (2x)
- Grammar: Use determiners (the,
a) wherever possible
- General: avoid long and complex sentences
-Grammar: transform passive into active -new rule: Grammar:
avoid parenthesis
- Grammar: Use determiners (the, a)
wherever possible
9. Annexes
169
Subject 2:
S2 (part 2) C2 (part 2)
Rules applied
Rules not
applied; errors
The four best third-ranked teams are seeded for the
UEFA Europa League round of 32.
This ranking is determined in
accordance with the following criteria (in descending order):
higher number of points obtained in
the group matches
superior goal difference
higher number of
goals scored
higher number of away goals scored
The four best third-ranked teams are seeded
for the UEFA Europa League round of 32. The
teams that play in the UEFA Europa League round of
32 are the best four that finished the group stage in
third position.
The following
criteria are applied (in descending
order) to determine these teamsthis
ranking:
A higher number of points that the team obtained in
the group matches
A difference in the number of goals that is
superiorsuperior goal difference
A higher number of goals that the
team scored
A higher number of goals that the team scored in
away matches
Grammar: Transform
passive into active
Grammar: Use
determiners (the, a) wherever possible
Grammar: Do not omit relative pronouns
General: avoid omissions
Grammar: Use determiners (the,
a) wherever possible
Grammar: Use determiners (the,
a) wherever possible General: avoid
omissions Grammar: Use
determiners (the, a) wherever possible
Avoid omissions General: Try to avoid synonyms
- Semantic error: not the same sense
- Semantic error:
not the same sense
- Error: term
9. Annexes
170
higher number of wins
higher number of away wins
higher number of coefficient points accumulated by the
club in question, as well as its association, over
the previous five seasons (see paragraph 9.02).
Round of 16 The round of 16 pairings are
determined by means of a draw.
The round of 16 is played under the knockout system,
on a home-and-away basis (two legs).
The UEFA administration
ensures that the following principles are respected.
A higher number of wins
A higher number of wins in away
matches
A higher number of coefficient points that the
club in question or thatand its association
accumulated over the previous five seasons (see
paragraph 9.02).
The Round of 16 A draw determines tThe
pairings for the round of 16 are determined by
means of a draw.
The round of 16 follows a knockout system
where there is one home match and one away
match (two legs). The UEFA
administration ensures the compliance with
or different
paraphrases expressing the same content in
order to ensure uniformity
Grammar: Use determiners (the, a) wherever
possible Grammar: Use
determiners (the, a) wherever possible
General: Try to avoid synonyms or different
paraphrases expressing the same content in
order to ensure uniformity
Grammar: Use determiners (the, a) wherever
possible - Grammar: Transform
passive into active
- Lexis: Avoid noun clusters - Grammar:
Transform passive into active
- Grammar: Transform passive into
active
Grammar: Use determiners (the,
a) wherever possible
- Semantic error: not the same
sense
- New rule: No determiners
necessary in titles and subtitles
- Error: wrong expression in English
9. Annexes
171
Clubs from the
same association cannot be drawn against each other.
The winners and runners-up of the
same group cannot be drawn against each other.
The group-winners
cannot be drawn against each other.
The runners-up
cannot be drawn against each other.
The runners-up must play the first
leg at home. The team which
scores the greater aggregate of goals in the two matches
qualifies for the quarter-finals.
Otherwise, the
stipulations of Article 8 apply.
Quarter-finals
The eight winners of
the round of 16 contest the quarter-finals. The quarter-
final pairings are determined by means of a draw.
the following
principles:
The clubs from the same association
cannot be drawn against each other.
The winners and
the runners-up of the same group cannot be drawn
against each other.
The winners in a groupgroup-winners cannot
be drawn against each other.
The runners-up cannot be drawn
against each other.
The runners-up must play the first leg at home.
The team which
scores the greater aggregate of goals in the
two matches qualifies for the quarter-finals.
Otherwise, the
stipulations of Article 8 are applied.
The Quarter-
finals
The eight winners of the round of 16 play the quarter-
finals. A draw determines theThe pairings
Grammar: Use
determiners (the, a) wherever possible
Lexis: Avoid the active form of the verb apply
General: Try to
avoid synonyms or different paraphrases
expressing the same content in order to ensure
uniformity
Error: term
New rule: see above
9. Annexes
172
The quarter-finals
are played under the knockout system, on a home-
and-away basis (two legs). The team which scores
the greater aggregate of goals in the two matches
qualifies for the semi-finals. Otherwise, the
stipulations of Article 8 apply.
Semi-finals The four winners of the quarter-finals
contest the semi-finals. The semi-final pairings are determined by
means of a draw. The semi-finals are played under the
knockout system, on a home-and-away basis (two
legs). The team which scores the greater aggregate of
goals in the two matches qualifies for the final.
Otherwise, the stipulations of Article 8 apply.
Final The final is played as one single match
at a neutral venue. If the result stands as a draw at the end
of normal playing time, extra time of two periods of 15
minutes is played. If one of the teams scores more goals
for the quarter-
finals are determined by means of a draw.
The quarter-finals follow a knockout system where
there is one home match and one away match (two
legs). The team which scores the greater
aggregate of goals in the two matches qualifies
for the semi-finals. Otherwise, the stipulations of
Article 8 are applied.
The Semi-finals The four winners of the quarter-
finals play the semi-finals. A draw determines tThe pairings of
the semi-finals are determined by means of a draw.
The semi-finals follow a knockout system where
there is one home match and one away match (two
legs). The team which scores the greater aggregate
of goals in the two matches qualifies for the final.
Otherwise, the stipulations of Article 8 are
applied.
The fFinal
The final is one single match. The venue of the final
is a neutral venue. If the result stands as a
draw at the end of the normal playing time,
there is an extra
- Grammar:
Transform passive into active
- General: avoid omissions - Lexis: avoid
noun clusters - Lexis: Avoid the active form of
the verb apply
General: Try to avoid synonyms or different
paraphrases expressing the same content in
order to ensure uniformity - Grammar:
Transform passive into active
- General: avoid omissions - Lexis: avoid
noun clusters - Lexis: Avoid the active form of
the verb apply
- General: avoid
long and complex sentences
- Grammar: Transform passive into
active (3x) Grammar: Use determiners (the,
a) wherever possible (3x)
Error: see above
New rule: see above
Error: see above
New rule: see above
9. Annexes
173
than the other
during extra time, that team is declared the winner.
If the two teams are still equal after extra time, the winner is
determined by kicks from the penalty mark (Article 17).
The provisions of Article 8 do not apply to the final.
VII Fixtures, Match Dates, Venues and
Kick-off Times
Article 12
Match dates All matches are
played according to the UEFA Match Calendar (see
Annex Ic). These dates are final and binding on all
concerned, subject to the provisions of paragraphs 12.04,
12.05 and 12.06. The following principles apply to
this competition:
With the exception
of the final, which is played on a Saturday, UEFA
Champions League matches are played on Tuesdays and
Wednesdays.
time of two
periods of 15 minutes. If one of the teams scores
more goals than the other team during the extra
time, that team wins. If the two teams are still
equal after the extra time, the kicks from the
penalty mark determine who winsthe winner
(Article 17). The provisions of Article 8 are not
applied to the final.
VII The fFixtures, the
Match Dates, the
Venues and the Kick-off Times
Article 12 The Match dates
The UEFA Match Calendar determines when
the matches take place (see the annex ic). The
dates of the UEFA Match Calendar are final
and bind everyone who isbinding on all
concerned. The dates are subject to the provisions
of the paragraphs 12.04, 12.05 and 12.06. The
following principles are applied to this
competition: The matches of
the UEFA Champions League matches
take place on Tuesdays and on Wednesdays.
The final is the
-General: avoid
omissions - Lexis: Avoid the active form of
the verb apply
- Grammar: Transform
passive into active (2x) -General: avoid
omissions - General: avoid long and
complex sentences
- General: avoid long and complex
sentences - Grammar: Repeat the
preposition in conjoined
Error: no determined
article before “kicks from the penalty mark”
Error/style: ‘who wins’ too colloquial
New rule: see
above
Typical law language –
cannot be changed
- Grammar: Use determiners (the, a) wherever
possible
- Lexis: avoid noun clusters
9. Annexes
174
From the play -offs
onwards, the UEFA administration decides which
UEFA Champions League matches are to be played on
Tuesdays and which on Wednesdays on the
basis of the relevant draw.
As a rule, each club plays the same number of matches
on a Tuesday and on a Wednesday. Matches within the
same group are played on the same day. Exceptions to
this rule can be made by the UEFA administration.
Kick-off times
As a rule, the play-
offs, group matches, round of 16 matches, quarter-
finals, semi-finals and final kick off at 20.45 CET.
Exceptions to this rule can be set by the UEFA
administration.
only exception
and takes place on a Saturday.
From the play-
offs onwards, the UEFA administration
makes a draw and decides which UEFA
Champions League matches take place on
Tuesdays and which UEFA Champions
League matches take place on Wednesdays.
The UEFA administration bases this
decision on the relevant draw.
As a rule, each club plays the same number of
matches on a Tuesday and on a Wednesday.
Matches within the same group take place on the
same day. The UEFA administration
can make grant exceptions to this rule.
The Kick-off
times
As a rule, the play-offs, the group matches,
the matches of the round of 16, the quarter-finals,
the semi-finals and the final start at 20.45 CET.
The UEFA administration can make grant
exceptions to this
constructions
where appropriate - General: avoid
omissions
Grammar: Use determiners (the,
a) wherever possible (5x)
- Grammar: Transform
passive into active (2x) - General: Avoid
overly complex constructions and write clearly
- Lexis: avoid noun clusters
- Error: ‘makes a
draw and decides’; better: split up sentence
Error: ‘grant exceptions’
New rule: see
above
Error: see above
9. Annexes
175
In principle, the
fixtures within a group on the last matchday must be
played simultaneously. The UEFA
administration is authorised to fix the kick-off times.
Confirmation of fixtures for the
qualifying phase
The venues, dates and kick-off times for matches in the
three qualifying rounds must be confirmed and
communicated to the UEFA administration in
writing by the associations of the clubs concerned by
the deadline set by the UEFA administration. The
UEFA administration may alter or confirm
dates and kick-off times according to the principles set by
the Club Competitions Committee. Failure
to respect this provision may result in disciplinary
measures.
rule.
In principle, the fixtures within a
group on the last day of a matchmatchday
must take place simultaneously. The UEFA
administration can fixes the kick-off times.
The
Cconfirmation of fixtures for the
qualifying phase
The UEFA administration sets a deadline in
which Tthe associations of the clubs that are
concerned by the deadlines must confirm and communicate in
writing to the UEFA administration the
venues, the dates and the kick-off times for the
matches in the three qualifying rounds. The
UEFA administration can alter or can
confirm the dates and the kick-off times in
compliance with the principles of the Club
Competitions Committee. The noncompliance
with the provision can result in disciplinary
measures.
- Grammar:
Transform passive into active
- Grammar: Transform passive into
active - Grammar: Use determiners (the,
a) wherever possible (4x) Lexis: Avoid the
ambiguous modal verb may (2x)
General: Try to avoid synonyms or different
paraphrases expressing the same content in
order to ensure uniformity - General: avoid
overly complex constructions and write clearly
(2x) - Grammar: avoid
parenthesis
Term: matchday
Word choice: ‘is authorised’ – not the same
sense/see German translation
New rule: see
above
Error: omission of sentence part
9. Annexes
176
Automatic
reversals
If more than one club from the same city, or within a
radius of 50km (31 miles) of each other, are taking part in
the UEFA Champions League competition and/or
play in the same stadium, and if the association and the
clubs concerned explicitly declare when entering the
clubs that their matches cannot be played on the same
day, the UEFA administration may alter or confirm
dates and kick-off times according to the principles set by the Club
Competitions Committee.
If more than one club from the same city, or within a
radius of 50km (31 miles) of each other, are taking part in
any of the UEFA club competitions and/or play in the
same stadium, and if the association and the clubs
concerned explicitly declare when entering the clubs
that their matches cannot be played on the same day,
The Automatic reversals
If
mMore than one
club from the same city, or within a radius of
50km (31 miles) of each other, takes part in the
UEFA Champions League competition
and/or plays in the same stadium.
tThe association and the clubs
concerned explicitly declare at the moment of
entering the clubs that their matches cannot take place
on the same day. In this case, then the UEFA administration
can alter or confirm the dates and the kick-off
times in compliance with the principles of
the Club Competitions Committee.
If mMore than one
club from the same city, or within a radius of
50km (31 miles) of each other, takes part in any
one of the UEFA club competitions and/or plays in
the same stadium. tThe association
and the clubs concerned explicitly declare
at the moment of entering the clubs
- General: avoid
long and complex sentences (3x)
- General: avoid overly complex constructions
and write clearly - Lexis: avoid the use of ing-words
(3x) - Grammar: transform
passive into active Lexis: Avoid the
ambiguous modal verb may Grammar: Use
determiners (the, a) wherever possible (2x)
General: avoid
long and complex sentences (3x)
- Lexis: avoid the use of ing-words (2x)
Grammar: Use determiners (the, a) wherever
possible (2x) - Grammar: transform
passive into active (2x)
New rule: see above
New rule: Every segment has to be independent
on its own
9. Annexes
177
priority is given to
UEFA Champions League matches and UEFA Europa
League matches are reversed.
Venues as from
the play-offs In principle, from the play-offs, a club
must play all its matches in the competition at one
and the same ground. Matches may be played
either at the ground of the home club or at another ground in
the same or another city within the territory of its
association, or, if so decided by the UEFA
administration and/or the UEFA disciplinary bodies,
in the territory of another UEFA member association
for reasons of safety or as a result of a disciplinary
measure. In principle, venues are approved only if
direct international flights and/or charter flights are
able to land within an acceptable distance of the
venue, in the country of the club concerned. If the
match is being played in another city or country, the
venue is subject to the approval of the UEFA
administration.
that their matches
cannot take place on the same day. In this case,
then the UEFA Champions League matches
have priority and the UEFA Europa League matches
are reversed. The Venues as
from the play-offs
In principle, from
the play -offs, a club must play all its matches in the
competition at one and the same ground. The
matches can take place either at the ground of the
home club or at another ground in the same or another city within
the territory of the association of the home club. If the
UEFA administration and/or the UEFA
disciplinary bodies decide so for reasons of
safety or as a result of a disciplinary
measure, the matches can take place also in
within the territory of another UEFA member
association, for reasons of safety or as a result of a
disciplinary measure. In principle, the
UEFA administration approves venues
only if direct international flights and/or
charter flights can
Grammar: Use determiners (the,
a) wherever possible
Lexis: Avoid the ambiguous modal verb may
- Grammar: transform
passive into active (5x)
- Avoid possessive case (´s) and
possessive pronouns
General: avoid long and complex
sentences - General: avoid
overly complex constructions and write clearly
New rule: see above
- General: order
the parts of the sentences logically
General: Try to avoid synonyms or different
paraphrases expressing the same content in
order to ensure uniformity
9. Annexes
178
Alternative venues If, at any time
during the season, the UEFA administration
deems that, for whatever reason, some venues may
not be fit for staging a match, UEFA may consult the
associations and clubs concerned and ask them to
propose an alternative venue, in accordance with the
UEFA requirements. Should such an association and club
not be able to propose an acceptable
alternative venue by the deadline set by the UEFA
administration, UEFA may select an alternative,
neutral venue and make all the necessary
arrangements for the staging of the match together with
the relevant association and local authorities. In both cases, the
costs of staging the match are borne by the home club. The
UEFA administration take a final decision on
land within an
acceptable distance of the venue, in the
country of the club concerned. If the match takes
place in another city or in another country, the
UEFA administration must approve the
venue.
The Alternative venues
If, at any time
during the season,The the UEFA
administration deems that, for whatever reason,
some venues cannot stage a match, at any time during the
season and for whatever reason. In this case, the
UEFA can consult the associations concerned and
the clubs concerned and ask them to
propose an alternative venue, in accordance
with the UEFA requirements. If the associations
concerned and the clubs concerned cannot
propose an alternative venue that is acceptable
by the deadline set by the UEFA administration,
the UEFA can select another neutral venue. In
this case the UEFA makes all the necessary
arrangements for
- General: avoid overly complex constructions
and write clearly (2x) Grammar: Do not
omit relative pronouns Lexis: Avoid the
ambiguous modal verb may (3x) General: avoid
long and complex sentences
Grammar: Use determiners (the, a) wherever
possible - Grammar: transform
passive into active (2x) - Lexis: avoid
noun clusters
- General: avoid omissions
New rule: see above
General: avoid
long and complex sentences
Error: see above (‘the UEFA’, 2x)
- General: avoid
omissions (3x)
9. Annexes
179
the match venue in
due time.
Final
The final is organised by a local organising
committee (LOC) on the basis of a contract between the host association
and UEFA. The date and venue are chosen by the
Executive Committee. In principle, the local
organisation of the final is entrusted to a different
association each year.
the staging of the
match together with the relevant association and
the local authorities. In both cases, it is
the home club that pays for the costs to stage the
match. The UEFA administration takes a final
decision on the venue of the match in due
time.
The Final On the basis of a contract between
the host association and the UEFA, a local
organising committee (LOC) organizes the final. The
Executive Committee chooses the date
and the venue. In principle, each year it is a
different association that takes care of the
local organisation of the final.
- Grammar: transform
passive into active (3x) Grammar: Use
determiners (the, a) wherever possible
New rule: see above
Error: see above (‘the UEFA’, 2x)
Error: style/too colloquial (‘it is a
different association that..’)
9. Annexes
180
Annex I: C2
III Trophies and Medals
Article 5
Trophy
The original trophy, which is used for the official presentation ceremony at the final, remains in the keeping of UEFA at all times. The winning club receives the UEFA
Champions League winners trophy, a replica of the trophy that is as large as the original.
A club which wins the competition three consecutive times or five times in total receives a special mark of recognition. Once a cycle of three successive wins or five in total has
been completed, the club concerned starts a new cycle from zero.
Replica trophies that are awarded to winners of the UEFA Champions League (past and current) must remain within the relevant control of the club at all times and must not leave its region or the country of its association without the prior written consent of UEFA. Clubs must not permit a replica trophy that is used in a context where a third party (including, without limitation, the sponsors of the club and other commercial partners) is granted visibility or in another way which could lead to an association between a third party and the trophy and/or the competition. Clubs must comply with trophy use
guidelines of any kind that are issued by the UEFA administration from time to time.
Medals
The winning club receives thirty gold medals and the runner-up receives 30 silver
medals. Additional medals must not be produced.
IV Responsibilities
Article 6
UEFA responsibilities
UEFA insures its own area of responsibility in accordance with the present regulations:
third-party liability insurance spectator accident insurance (for the final only) group accident insurance for UEFA delegates legal expenses insurance (restricted to criminal matters).
Responsibilities of the associations and clubs
The clubs are responsible for the behaviour of their players, officials, members,
supporters and each person that carries out a function at a match on behalf of the club.
The home club (or the host association) is responsible for order and security before, during and after the match. The home club (or the host association) can be called to
account for incidents of any kind and has to be disciplined.
The club considered the home club must stage the relevant matches at the ground in accordance with the instructions of UEFA (or of a third party that acts on behalf of UEFA) and in cooperation with the association concerned. However, the club is considered solely accountable for all of its obligations in this respect, unless the relevant body or
bodies decide otherwise.
Irrespective of the insurance coverage of UEFA, each club and host association must conclude an insurance coverage with reputable insurers at their own cost, in relation to
any and all risks. For this purpose, the following principles are applied:
i. each club must conclude and maintain insurance coverage to fully cover all of its risks in connection with its participation in the
competition;
9. Annexes
181
ii. in addition, the home club or the host association must conclude and maintain an insurance coverage for the risks in connection with staging and organising its home matches.This must include, without limitation, third-party liability insurance (for all third parties that participate in matches or attend the relevant venue) providing for appropriate guaranteed sums for damages to persons, objects and property, as well as for pure economic losses corresponding to
the specific circumstances of the club or association concerned;
iii. to the same extent as in paragraph b) above, the host association of the final match must conclude and maintain insurance coverage to fully cover all of its risks in connection with the staging and organisation of the final match;
iv. if the home club or the host association is not the owner of the stadium used, it is also responsible for the providing of an adequate and fully comprehensive insurance cover, including third-party liability and property damage, taken out by the relevant
stadium owner and/or tenant;
v. the home club and the host association must ensure that UEFA is included in all insurance policies as defined in the present paragraph and must hold UEFA harmless from any and all claims for liability arising in relation to the staging and organising of the
relevant matches;
In any case, UEFA can ask anyone involved to provide, free of charge, written releases of liability and/or hold harmless notes, and/or confirmations and/or copies of the policies concerned in one of the official languages of UEFA.
The clubs undertake that their team arrives at the match venue by the evening before the match at the latest and to fulfil the media obligations of the clubs the day before the
match.
Visiting clubs undertake not to play other matches when they travel to and from away
matches in this competition.
V Competition System
Article 7
Number of rounds
As shown in Annex Ib, the competition consists of:
i. the qualifying phase for the UEFA Champions League:
first qualifying round second qualifying round third qualifying round
ii. play-offs;
iii. UEFA Champions League:
group stage (six matchdays) round of 16 quarter-finals semi-finals final.
9. Annexes
182
Qualifying phase
Qualifying-phase matches are played according to the cup (knockout) system, with each club playing each opponent twice, in home and away matches. The team which scores the greater aggregate of goals in the two matches qualifies for the next match (second qualifying round, third qualifying round or play-offs, as applicable). Otherwise, the stipulations of Article 8 are applied. The clubs that were defeated in the first and second qualifying rounds are eliminated from the competition. The clubs that were defeated in the third qualifying round are entitled to play in the play-offs of the UEFA Europa League
in progress. Clubs from the same association cannot be drawn against each other.
Play-offs
Play-off matches are played according to the cup (knockout) system, with each club playing each opponent twice, in home and away matches. The team which scores the greater aggregate of goals in the two matches qualifies for the UEFA Champions League group stage. Otherwise, the stipulations of Article 8 are applied. The clubs defeated in the play-offs are entitled to play in the group stage of the UEFA Europa League in
progress. Clubs from the same association cannot be drawn against each other.
Group stage
Once the play-offs have been completed, the 32 remaining clubs are drawn into eight
groups of four. Clubs from the same association cannot be drawn into the same group.
Each club plays one home and one away match against each other club in its group.Three points are awarded for a win, and one point is awarded for a draw, and no
point is awarded for a defeat. The following match sequence is applied:
1st matchday: 2 v 3, 4 v 1.
2nd matchday: 1 v 2, 3 v 4.
3rd matchday: 3 v 1, 2 v 4. 4th matchday: 1 v 3, 4 v 2.
5th matchday: 3 v 2, 1 v 4.
6th matchday: 2 v 1, 4 v 3.
If two or more teams are equal on points on completion of the group matches, the
following criteria are applied to determine the rankings (in descending order):
a) higher number of points obtained in the group matches played among the
teams in question;
b) superior goal difference from the group matches played among the teams in
question;
c) higher number of goals scored in the group matches played among the
teams in question;
d) higher number of away goals scored in the group matches played among
the teams in question;
e) if, after having applied the criteria a) to d), two teams still have an equal ranking, the criteria a) to d) are reapplied to determine the final ranking of the two teams. If this procedure does not lead to a decision, the criteria f) to
h) are applied;
f) superior goal difference from all the group matches played;
g) higher number of goals scored from all the group matches played;
h) higher number of coefficient points accumulated by the club in question, as well as the association of the club, over the previous five seasons (see
paragraph 9.02).
9. Annexes
183
The eight group-winners and eight runners-up of the group stage qualify for the round of 16. The clubs that finish this stage in the third position in their group move into the round of 32 of the current UEFA Europa League. The clubs that finish this stage in the fourth
position in their group are eliminated.
The teams that play in the UEFA Europa League round of 32 are the best four that finished the group stage in third position. The following criteria are applied (in descending order) to determine these teams:
a) A higher number of points that the team obtained in the group matches b) A difference in the number of goals that is superior c) A higher number of goals that the team scored d) A higher number of goals that the team scored in away matches e) A higher number of wins f) A higher number of wins in away matches g) A higher number of coefficient points that the club in question or that its
association accumulated over the previous five seasons (see paragraph 9.02).
The Round of 16
A draw determines the pairings for the round of 16. The round of 16 follows a knockout system where there is one home match and one away match (two legs). The UEFA administration ensures the compliance with the following principles:
a) The clubs from the same association cannot be drawn against each other. b) The winners and the runners-up of the same group cannot be drawn against
each other. c) The winners in a group cannot be drawn against each other. d) The runners-up cannot be drawn against each other. e) The runners-up must play the first leg at home.
The team which scores the greater aggregate of goals in the two matches qualifies for the quarter-finals. Otherwise, the stipulations of Article 8 are applied.
The Quarter-finals
The eight winners of the round of 16 play the quarter-finals. A draw determines the pairings for the quarter-finals. The quarter-finals follow a knockout system where there is one home match and one away match (two legs).
The team which scores the greater aggregate of goals in the two matches qualifies for the semi-finals. Otherwise, the stipulations of Article 8 are applied.
Semi-finals
The four winners of the quarter-finals play the semi-finals. A draw determines the pairings of the semi-finals. The semi-finals follow a knockout system where there is one home match and one away match (two legs). The team which scores the greater aggregate of goals in the two matches qualifies for the
final. Otherwise, the stipulations of Article 8 are applied.
The final
The final is one single match. The venue of the final is a neutral venue. If the result stands as a draw at the end of the normal playing time, there is an extra time of two periods of 15 minutes. If one of the teams scores more goals than the other team during the extra time, that team wins. If the two teams are still equal after the extra time, the kicks from the penalty mark determine who wins (Article 17). The provisions of Article 8 are not applied to the final.
9. Annexes
184
VII Fixtures, Match Dates, Venues and Kick-off Times Article 12
Match dates
The UEFA Match Calendar determines when the matches take place (see the annex ic). The dates of the UEFA Match Calendar are final and bind everyone who is concerned. The dates are subject to the provisions of paragraphs 12.04, 12.05 and 12.06. The following principles are applied to this competition:
a) The UEFA Champions League matches take place on Tuesdays and on Wednesdays.The final is the only exception and takes place on a Saturday.
b) From the play-offs onwards, the UEFA administration makes a draw and decides which UEFA Champions League matches take place on Tuesdays and which UEFA Champions League matches take place on Wednesdays. As a rule, each club plays the same number of matches on a Tuesday and on a Wednesday. Matches within the same group take place on the same day. The UEFA administration can make exceptions to this rule.
Kick-off times
As a rule, the play-offs, the group matches, the matches of the round of 16, the quarter-finals, the semi-finals and the final start at 20.45 CET. The UEFA administration can make exceptions to this rule. In principle, the fixtures within a group on the last day of a match must take place simultaneously. The UEFA administration can fix the kick-off times.
Confirmation of fixtures for the qualifying phase
The associations of the clubs that are concerned by the deadlines must confirm and communicate in writing to the UEFA administration the venues, the dates and the kick-off times for the matches in the three qualifying rounds. The UEFA administration can alter or can confirm the dates and the kick-off times in compliance with the principles of the Club Competitions Committee. The noncompliance with the provision can result in disciplinary measures.
Automatic reversals
If
- more than one club from the same city, or within a radius of 50km (31 miles) of each other, takes part in the UEFA Champions League competition and/or plays in the same stadium
- the association and the clubs concerned explicitly declare at the moment of entering the clubs that their matches cannot take place on the same day
then the UEFA administration can alter or confirm the dates and the kick-off times in compliance with the principles of the Club Competitions Committee.
9. Annexes
185
If
- more than one club from the same city, or within a radius of 50km (31 miles) of each other, takes part in any one of the UEFA club competitions and/or plays in the same stadium
- the association and the clubs concerned explicitly declare at the moment of entering the clubs that their matches cannot take place on the same day
then the UEFA Champions League matches have prior ity and the UEFA Europa League matches are reversed.
Venues as from the play-offs In principle, from the play-offs, a club must play all its matches in the competition at one and the same ground. The matches can take place either at the ground of the home club or at another ground in the same or another city within the territory of the association of the home club. If the UEFA administration and/or the UEFA disciplinary bodies decide so, the matches can take place also in the territory of another UEFA member association, for reasons of safety or as a result of a disciplinary measure. In principle, the UEFA administration approves venues only if direct international flights and/or charter flights can land within an acceptable distance of the venue, in the country of the club concerned. If the match takes place in another city or country, the UEFA administration must approve the venue.
Alternative venues
If, at any time during the season, the UEFA administration deems that, for whatever reason, some venues cannot stage a match, UEFA can consult the associations and the clubs concerned and ask them to propose an alternative venue in accordance with the UEFA requirements. If the associations and the clubs cannot propose an alternative venue that is acceptable by the deadline set by the UEFA administration, UEFA can select another neutral venue. In this case UEFA makes all the necessary arrangements for the staging of the match together with the relevant association and the local authorities. In both cases, it is the home club that pays for the costs to stage the match. The UEFA administration takes a final decision on the venue of the match in due time.
Final
On the basis of a contract between the host association and UEFA, a local organising committee (LOC) organizes the final. The Executive Committee chooses the date and the venue. In principle, each year it is a different association that takes care of the local organisation of the final.
9. Annexes
186
Annex J: PE1
III Pokal, Plaketten und Medaillen Artikel 3
Pokal
3.01 Der Originalpokal wird für die offizielle Pokalübergabe beim Endspiel und für andere offizielle Veranstaltungen verwendet, die von der UEFA genehmigt werden. Der Originalpokal bleibt stets im Besitz der UEFA. Der Sieger erhält eine Nachbildung in Originalgrösse, die Siegertrophäe der UEFA-Fussball-Europameisterschaft. 3.02 Ein Verband, der den Pokal dreimal nacheinander gewinnt, oder der ihn insgesamt fünfmal gewinnt, erhält ein spezielles Zeichen der Anerkennung. Hat ein Verband den Wettbewerb dreimal nacheinander oder insgesamt fünfmal gewonnen, so fängt die Zählung für diesen Verband wieder bei Null an. 3.03 Die jeweiligen Verbände müssen stets im Besitz der Nachbildungen bleiben, die die früheren und aktuellen Gewinner der UEFA-Fussball-Europameisterschaft erhalten haben. Die Nachbildungen dürfen das Land des jeweiligen Verbands nicht ohne die vorherige schriftliche Genehmigung der UEFA verlassen. Die Verbände dürfen die Verwendung einer Nachbildung nicht in einem Zusammenhang genehmigen, in dem Dritte auftreten dürfen. Dritte beinhalten, ohne Einschränkung, Sponsoren der Verbände sowie andere kommerzielle Partner. Die Verbände dürfen nicht genehmigen, dass die Nachbildung auf eine Art verwendet wird, die eine Verbindung zwischen Dritten und der Nachbildung und/oder dem Wettbewerb herstellen könnte. Die Verbände müssen Richtlinien zur Verwendung des Pokals jeglicher Art einhalten, die die UEFA-Administration von Zeit zu Zeit herausgeben kann. 3.04 Verbände dürfen keine Werbematerialien oder Werbeartikel entwickeln, herstellen, verwenden, verkaufen oder verteilen, die eine Darstellung des Pokals oder der Nachbildung des Pokals enthalten (einschliesslich Bildern, die die Sieger zeigen, die den Pokal in den Händen halten). Verbände dürfen solch eine Darstellung nicht in einem Kontext verwenden, der zu einer Assoziation zwischen Dritten und dem Pokal, der Nachbildung und/oder dem Wettbewerb führen könnte. Ausserdem dürfen sie Dritten nicht erlauben, dies zu tun. Erinnerungsplaketten 3.05 Jeder Verband, der an der Endrunde teilnimmt, erhält eine Erinnerungsplakette. Plaketten für Halbfinalisten 3.06 Jeder unterlegene Halbfinalist erhält eine Plakette. Finalplakette 3.07 Jeder Finalist erhält eine Plakette.
Medaillen
3.08 Der Sieger erhält 40 Goldmedaillen. Der Zweitplatzierte erhält 40 Silbermedaillen. Die unterlegenen Halbfinalisten erhalten jeweils 40 Bronzemedaillen. Zusätzliche Medaillen werden nicht produziert.
IV Verantwortung Artikel 4
Verantwortung der Verbände
4.01 Ein Verband ist für das Verhalten seiner Spieler, seiner Offiziellen, seiner Mitglieder, seiner Anhänger und jeder Person verantwortlich, die im Auftrag des Verbands eine Funktion bei einem Spiel ausübt.
9. Annexes
187
4.02 Der Landesverband, der ein Qualifikationsspiel oder die Endrunde auf seinem Gebiet ausrichtet, gilt für dieses Spiel oder Turnier als Ausrichterverband. 4.03 Der Ausrichterverband ist für Ordnung und Sicherheit vor, während und nach den Spielen verantwortlich. Der Ausrichterverband kann für Zwischenfälle jeglicher Art zur Verantwortung gezogen und bestraft werden. 4.04 Prinzipiell müssen Spiele in einem Stadion auf dem Gebiet des jeweiligen Ausrichterverbands ausgetragen werden. In Ausnahmefällen können Spiele auf dem Gebiet eines anderen UEFA-Mitgliedsverbands ausgetragen werden, wenn die UEFA-Administration und/oder die Disziplinarinstanzen dies entscheiden (aus Sicherheitsgründen oder infolge einer Disziplinarmassnahme). Zusätzliche Verantwortung für die Endrunde 4.05 Die UEFA-Administration informiert die Verbände, die an der Endrunde teilnehmen, über etwaige zusätzliche Richtlinien, Weisungen oder Beschlüsse, die mit der Endrunde zusammenhängen. Die UEFA-Administration stellt den Verbänden die notwendigen Dokumente zu gegebener Zeit zur Verfügung.
VI Wettbewerbsmodus Artikel 6
Wettbewerbsphasen
6.01 Der Wettbewerb besteht aus einem Qualifikationswettbewerb und einer Endrunde.
Artikel 7
A. Qualifikationswettbewerb Gruppenbildung
7.01 Die Mannschaften der Ausrichterverbände für die Endrunde, Polen und die Ukraine, qualifizieren sich automatisch für die Endrunde. Die übrigen Mannschaften werden in sechs Gruppen von sechs Mannschaften und in drei Gruppen von fünf Mannschaften gelost. Die UEFA-Administration lost die Gruppen nach Abschluss des Qualifikationswettbewerbs der FIFA Fussball-Weltmeisterschaft 2010 aus. Die Auslosung basiert auf einem Setzsystem. Der amtierende Europameister ist immer gesetzt. Die übrigen Verbände werden auf der Grundlage der UEFA-Koeffizientenrangliste für Nationalmannschaften klassiert (vgl. Anhang I, Absatz 1.2.1). Die Entscheidungen der UEFA-Administration sind endgültig. 7.02 Wenn zwei oder mehr dieser Mannschaften denselben Koeffizienten haben, sind die folgenden Kriterien ausschlaggebend. Nur der letzte Qualifikationswettbewerb wird berücksichtigt. Sie werden in dieser Reihenfolge angewendet: a) UEFA-Koeffizient für Nationalmannschaften, resultierend aus den ausgetragenen Spielen; b) durchschnittliche Tordifferenz; c) durchschnittliche Anzahl erzielter Tore; d) durchschnittliche Anzahl erzielter Auswärtstore; e) Fairplay-Rangliste; f) Losentscheid.
Austragungsmodus für den Qualifikationswettbewerb
7.03 Die Spiele des Qualifikationswettbewerbs werden in Gruppen entsprechend dem Meisterschaftsmodus ausgetragen. Jede Mannschaft spielt in Hin- und Rückspiel gegen jede andere Mannschaft ihrer Gruppe. Drei Punkte werden für einen Sieg, ein Punkt für ein Unentschieden und null Punkte für eine Niederlage zugesprochen. Punktegleichheit nach den Gruppenspielen
9. Annexes
188
7.04 Wenn zwei oder mehr Mannschaften nach Abschluss der Gruppenspiele punktgleich liegen, werden die folgenden Kriterien angewendet, um die Platzierungen zu ermitteln. Die Kriterien werden in dieser Reihenfolge angewendet: a) höhere Punktzahl, die die Mannschaften in den Gruppenspielen der betreffenden Mannschaften erhielten; b) bessere Tordifferenz aus den Gruppenspielen der betreffenden Mannschaften; c) grössere Anzahl Tore aus den Gruppenspielen der betreffenden Mannschaften; d) grössere Anzahl Auswärtstore aus den Gruppenspielen der betreffenden Mannschaften; e) Wenn die Kriterien a) bis d) auf mehrere Mannschaften angewendet worden sind und immer noch zwei oder mehr Mannschaften denselben Platz belegen, werden die Kriterien a) bis d) erneut angewendet, um die Platzierung dieser Mannschaften zu ermitteln. Wenn dieses Verfahren zu keiner Entscheidung führt, werden die Kriterien f) bis j) angewendet; f) bessere Tordifferenz aus allen Gruppenspielen; g) grössere Anzahl erzielter Tore aus allen Gruppenspielen; h) grössere Anzahl erzielter Auswärtstore aus allen Gruppenspielen; i) Fairplay-Rangliste aus allen Gruppenspielen; j) Losentscheid.
Qualifikation für die Endrunde
7.05 Die neun Gruppensieger und der beste Zweitplatzierte qualifizieren sich direkt für die Endrunde. 7.06 Nur die Ergebnisse gegen die Mannschaften auf dem ersten, dritten, vierten und fünften Platz werden berücksichtigt, um den besten Zweitplatzierten zu ermitteln. Zu diesem Zweck werden die folgenden Kriterien in dieser Reihenfolge angewendet: a) höhere Punktzahl, die die Mannschaften in diesen Spielen erhielten; b) bessere Tordifferenz aus diesen Spielen; c) grössere Anzahl erzielter Tore aus diesen Spielen; d) grössere Anzahl erzielter Auswärtstore aus diesen Spielen;
e) Platzierung in der UEFA-Koeffizientenrangliste für Nationalmannschaften (vgl.-Anhang I, Absatz 1.2.2); f) Fairplay-Rangliste aus diesen Spielen; g) Losentscheid. 7.07 Die acht übrigen Zweitplatzierten bestreiten Entscheidungsspiele. Die vier Spielpaarungen werden durch Auslosung ermittelt. Die vier Zweitplatzierten mit der besten Platzierung in der UEFA-Koeffizientenrangliste für Nationalmannschaften (vgl.Anhang I, Absatz 1.2.2) sind für die Auslosung gesetzt. Wenn zwei oder mehr dieser Mannschaften denselben Koeffizienten haben, werden die Kriterien von Absatz 7.02 angewendet. Die Entscheidungsspiele werden nach dem K.-o.-System ausgetragen (in Hin- und Rückspiel). Die gesetzten Mannschaften bestreiten das Rückspiel zu Hause. Die Mannschaften, die insgesamt die meisten Tore erzielen, qualifizieren sich für die Endrunde. Andernfalls werden die Bestimmungen von Absatz 7.08 angewendet. 7.08 Für Spiele, die nach dem K.-o.-System ausgetragen werden, wird das folgende Kriterium angewendet: Wenn beide Mannschaften insgesamt gleich viele Tore erzielen, qualifiziert sich die Mannschaft, die mehr Auswärtstore erzielt, für die folgende Runde. Wenn dieses Verfahren zu keiner Entscheidung führt, d.h. wenn beide Mannschaften gleich viele Heim- und Auswärtstore erzielen, wird das Rückspiel um zweimal 15 Minuten verlängert. Wenn in der Verlängerung beide Mannschaften gleich viele Tore erzielen, zählen die Auswärtstore doppelt (d.h. die Gastmannschaft ist qualifiziert). Wenn keine Tore in der Verlängerung erzielt
9. Annexes
189
werden, wird die Mannschaft, die sich für die Endrunde qualifiziert, durch Elfmeterschiessen ermittelt (vgl.Artikel 16).
Artikel 8 B. Endrunde
8.01 Das UEFA-Exekutivkomitee hat den Polnischen Fussballverband (PZPN) und den Ukrainischen Fussballverband (FFU) mit der gemeinsamen Organisation und Ausrichtung der Endrunde betraut. Gruppenbildung 8.02 16 Mannschaften qualifizieren sich für die Endrunde. Die UEFA-Administration teilt diese 16 Mannschaften in vier Vierergruppen ein (Gruppen A, B, C und D). 8.03 Durch Auslosung werden die vier Gruppen wie folgt gebildet: Gruppe A Gruppe B Gruppe C Gruppe D A1 B1 C1 D1 A2 B2 C2 D2 A3 B3 C3 D3 A4 B4 C4 D4
Koeffizienten
8.04 Die folgenden Mannschaften sind gesetzt: die Ausrichterverbände, der amtierende Europameister, sofern qualifiziert, und ein oder zwei Mannschaften mit den besten Koeffizienten gemäss der UEFA-Koeffizientenrangliste für Nationalmannschaften (vgl. Anhang I, Absatz 1.2.2). Die anderen Endrundenteilnehmer werden den vier Gruppen durch Auslosung zugewiesen (entsprechend ihren Koeffizienten). 8.05 Wenn zwei oder mehr der betreffenden Mannschaften denselben Koeffizienten haben, werden die Kriterien von Absatz 7.02 angewendet. Die Ergebnisse der Entscheidungsspiele (vgl. Absatz 7.07) werden dabei nicht berücksichtigt. Gruppenspielplan 8.06 Jede Mannschaft spielt nach dem Meisterschaftsmodus gegen jede andere Mannschaft ihrer Gruppe (drei Punkte für einen Sieg, ein Punkt für ein Unentschieden, null Punkte für eine Niederlage). Die Gruppenspiele werden entsprechend dem folgenden Schema ausgetragen. Die letzten zwei Spiele in jeder Gruppe müssen beide zeitgleich beginnen. Die erstgenannte Mannschaft gilt als Heimmannschaft. 1. Spieltag 2. Spieltag 3. Spieltag Gruppe A A1 gegen A2 A1 gegen A3 A4 gegen A1 A3 gegen A4 A2 gegen A4 A2 gegen A3 Gruppe B B1 gegen B2 B1 gegen B3 B4 gegen B1 B3 gegen B4 B2 gegen B4 B2 gegen B3 Gruppe C C1 gegen C2 C1 gegen C3 C4 gegen C1 C3 gegen C4 C2 gegen C4 C2 gegen C3 Gruppe D D1 gegen D2 D1 gegen D3 D4 gegen D1 D3 gegen D4 D2 gegen D4 D2 gegen D3
Punktegleichheit nach den Gruppenspielen
8.07 Wenn zwei oder mehr Mannschaften nach Abschluss der Gruppenspiele punktgleich liegen, werden die folgenden Kriterien angewendet, um die Platzierungen zu ermitteln. Sie werden in dieser Reihenfolge angewendet: a) höhere Punktzahl aus den Spielen der betreffenden Mannschaften;
9. Annexes
190
b) bessere Tordifferenz aus den Spielen der betreffenden Mannschaften (wenn mehr als zwei Mannschaften nach Abschluss der Gruppenspiele punktgleich liegen); c) grössere Anzahl erzielter Tore aus den Spielen der betreffenden Mannschaften (wenn mehr als zwei Mannschaften nach Abschluss der Gruppenspiele punktgleich liegen); d) bessere Tordifferenz aus allen Gruppenspielen; e) grössere Anzahl erzielter Tore aus allen Gruppenspielen; f) Platzierung in der UEFA-Koeffizientenrangliste für Nationalmannschaften (vgl. Anhang I, Absatz 1.2.2); g) Fairplay-Verhalten der betreffenden Mannschaften (Endrunde); h) Losentscheid. 8.08 Zwei Mannschaften haben die gleiche Punktzahl und die gleiche Tordifferenz. Diese Mannschaften bestreiten ihr letztes Gruppenspiel gegeneinander. Wenn dieses Spiel unentschieden endet, wird die Platzierung der zwei betroffenen Mannschaften durch Elfmeterschiessen ermittelt (vgl. Artikel 16; vorausgesetzt, dass keine anderen Mannschaften derselben Gruppe nach Abschluss der Gruppenspiele die gleiche Punktzahl haben). Wenn mehr als zwei Mannschaften die gleiche Punktzahl haben, werden die Kriterien von Absatz 8.07 angewendet. Viertelfinale 8.09 Die Sieger und die Zweitplatzierten jeder Gruppe bestreiten die Viertelfinalspiele in einem Spiel. Zu diesem Zweck wird das folgende Schema angewendet: Spiel 1 Sieger Gruppe A gegen Zweitplatzierten Gruppe B Spiel 2 Sieger Gruppe B gegen Zweitplatzierten Gruppe A Spiel 3 Sieger Gruppe C gegen Zweitplatzierten Gruppe D Spiel 4 Sieger Gruppe D gegen Zweitplatzierten Gruppe C
Halbfinale
8.10 Die vier Sieger der Viertelfinalspiele bestreiten die Halbfinalspiele in jeweils einer Begegnung. Zu diesem Zweck wird das folgende Schema angewendet: Sieger Spiel 1 gegen Sieger Spiel 3 Sieger Spiel 2 gegen Sieger Spiel 4
Endspiel
8.11 Die Sieger der Halbfinale bestreiten das Endspiel. Gleiche Anzahl Tore in einem Viertel- oder Halbfinalspiel bzw. im Endspiel 8.12 Wenn das Spiel nach regulärer Spielzeit unentschieden endet, wird eine Verlängerung von zweimal 15 Minuten gespielt. Wenn die Verlängerung unentschieden endet, wird der Sieger durch Elfmeterschiessen ermittelt (vgl. Artikel 16).
VIII Spieldaten, Spielorte und Anstosszeiten
Artikel 11
11.01 Der Wettbewerb findet im Anschluss an die FIFA Fussball-Weltmeisterschaft 2010 statt. Er wird über zwei Spielzeiten ausgetragen. A. Qualifikationswettbewerb Spieldaten 11.02 Die Gruppenspiele des Qualifikationswettbewerbs der UEFA-Fussball-Europameisterschaft 2010-12 werden für die folgenden zwölf Daten angesetzt: 2010 a) 3./4. und 7. September 2010 b) 8./9. und 12. Oktober 2010
9. Annexes
191
2011 c) 25./26. und 29. März 2011 d) 3./4. und 7. Juni 2011 e) 2./3. und 6. September 2011 f) 7./8. und 11. Oktober 2011 (Daten für die letzten Spiele in allen Gruppen) 11.03 Die Entscheidungsspiele zwischen den übrigen acht Zweitplatzierten werden für die folgenden zwei Daten angesetzt: a) 11./12. November 2011 b) 15. November 2011 11.04 Unter den folgenden Bedingungen können die Gruppenspiele, die in Absatz 11.02 genannt werden, an anderen Daten ausgetragen werden: wenn sich beide betroffenen Verbände einigen und wenn die Grundsätze betreffend das Abstellen von Spielern für Auswahlmannschaften der Verbände gemäss Anhang 1, Artikel 1 des FIFA-Reglements bezüglich Status und Transfer von Spielern eingehalten werden. Die Verbände einer Gruppe haben sich innerhalb von 30 Tagen nach der Auslosung auf die Spielabfolge in ihrer Gruppe zu einigen. Das genaue Datum jedes Spiels muss angegeben werden (z.B. Samstag, 4. September 2010). Wenn die betroffenen Verbände keine Einigung erzielen können, werden die Spiele entsprechend einem Standardspielplan ausgetragen, der von der UEFA-Administration aufgestellt wird. Dieser Standardspielplan berücksichtigt Faktoren, die für die Spielorganisation relevant sind, wie z.B. klimatische Bedingungen. 11.05 Aus Gründen der sportlichen Fairness kann die UEFA-Administration verlangen, dass Spiele derselben Gruppe zeitgleich ausgetragen werden. Nachträgliche Datenänderungen setzen die Genehmigung der UEFA-Administration voraus. Der betreffende Ausrichterverband muss in diesem Fall auch die anderen Verbände in derselben Gruppe informieren.
Spielorte und Anstosszeiten 11.06 Die Spielorte werden von den Ausrichterverbänden festgesetzt. Sie müssen den Gastverbänden und der UEFA-Administration spätestens 60 Tage im Voraus bekannt gegeben werden. Wenn ein Ausrichterverband einen Spielort festsetzt, muss er die Dauer der Reise des Gastverbands berücksichtigen. Der Spielort für ein Qualifikationsspiel darf nicht weiter als 120 Bus-Fahrtminuten vom nächsten internationalen Flughafen entfernt sein, es sei denn, der Gastverband erklärt sich damit einverstanden. Die Anstosszeiten müssen der UEFA-Administration spätestens 60 Tage vor dem Spiel bekannt gegeben werden. Ankunft der Mannschaften am Spielort 11.07 Die Verbände müssen dafür sorgen, dass ihre Mannschaften früh genug am Spielort ankommen, damit sie ihre Medienkonferenz vor dem Spiel vor Redaktionsschluss in beiden beteiligten Ländern abhalten können. Auf jeden Fall müssen die Mannschaften spätestens 24 Stunden vor Spielbeginn am Spielort ankommen.
B. Endrunde Spieldaten
11.08 Die Endrunde wird vom 8. Juni bis zum 1. Juli 2012 ausgetragen. Spielorte und Anstosszeiten 11.09 Die UEFA-Administration ist für die Erstellung des Endrundenspielplans verantwortlich. Die Mannschaften müssen mindestens zwei Ruhetage (48 Stunden) zwischen den Partien haben. Ankunft der Mannschaften in den Ausrichterländern 11.10 Die Mannschaften müssen spätestens fünf Tage vor ihrem ersten Spiel in ihrem Mannschaftshotel in einem der Ausrichterländer ankommen.
9. Annexes
192
Ankunft der Mannschaften am Spielort 11.11 Die Mannschaften müssen spätestens 24 Stunden vor Spielbeginn in ihrem Transferhotel ankommen; auf jeden Fall müssen sie sich spätestens 24 Stunden vor Spielbeginn innerhalb eines Umkreises von 60 km um den jeweiligen Spielort aufhalten.
Trainingsplätze
11.12 Die UEFA bietet jedem Endrundenteilnehmer einige ausgesuchte Trainingsplätze an. Wenn ein Verband einen Trainingsplatz wählt, der nicht Teil dieser Vorauswahl ist, trägt der Verband alle Kosten, die entstehen. 11.13 Ab dem fünften Tag vor dem ersten Spiel der Endrunde gelten alle Trainingsplätze, die von den Verbänden verwendet werden, als offizielle Trainingsplätze. Ab diesem Zeitpunkt werden die Bestimmungen von Absatz 28.14 angewendet.
9. Annexes
193
Annex K: PE2
III Pokal, Plaketten und Medaillen Artikel 3
Pokal 3.01 Der Originalpokal wird für die offizielle Pokalübergabe beim Endspiel und für andere offizielle Veranstaltungen verwendet, die von der UEFA genehmigt werden. Der Originalpokal bleibt stets im Besitz der UEFA. Der Sieger erhält eine Nachbildung in Originalgrösse, die Siegertrophäe der UEFA-Fussball-Europameisterschaft. 3.02 Ein Verband, der den Pokal dreimal nacheinander gewinnt, oder der ihn insgesamt fünfmal gewinnt, erhält ein spezielles Zeichen der Anerkennung. Hat ein Verband den Wettbewerb dreimal nacheinander oder insgesamt fünfmal gewonnen, so fängt die Zählung für diesen Verband wieder bei Null an. 3.03 Nachbildungen, die die früheren und aktuellen Gewinner der UEFA-Fussball-Europameisterschaft erhalten haben, müssen stets unter der Kontrolle der jeweiligen Verbände bleiben. Die Nachbildungen dürfen das Land des jeweiligen Verbands nicht ohne die vorherige schriftliche Genehmigung der UEFA verlassen. Die Verbände dürfen keine Verwendung der Nachbildung in einem Zusammenhang genehmigen, in dem Dritte auftreten dürfen. Dritte beinhalten, ohne Einschränkung, Sponsoren der Verbände sowie andere kommerzielle Partner. Zudem dürfen die Verbände nicht genehmigen, dass die Nachbildung auf eine Art verwendet wird, die eine Verbindung zwischen Dritten und der Nachbildung und/oder dem Wettbewerb herstellen könnte. Die Verbände müssen jegliche Richtlinien zur Verwendung des Pokals einhalten, die die UEFA-Administration von Zeit zu Zeit herausgeben kann. 3.04 Die Verbände dürfen keine Werbematerialien oder Werbeartikel entwickeln, herstellen, verwenden, verkaufen oder verteilen, die eine Darstellung des Pokals oder der Nachbildung des Pokals enthalten (einschliesslich Bildern, die die Sieger mit dem Pokal zeigen). Verbände dürfen solche Darstellungen nicht in einem Kontext verwenden, der zu einer Assoziation zwischen Dritten und dem Pokal, der Nachbildung und/oder dem Wettbewerb führen könnte. Ausserdem dürfen sie Dritten nicht erlauben, dies zu tun.
Erinnerungsplaketten 3.05 Jeder Verband, der an der Endrunde teilnimmt, erhält eine Erinnerungsplakette.
Plaketten für Halbfinalisten
3.06 Jeder unterlegene Halbfinalist erhält eine Plakette.
Finalplakette 3.07 Jeder Finalist erhält eine Plakette.
Medaillen
3.08 Der Sieger erhält 40 Goldmedaillen. Der Zweitplatzierte erhält 40 Silbermedaillen. Die unterlegenen Halbfinalisten erhalten jeweils 40 Bronzemedaillen. Zusätzliche Medaillen werden nicht produziert.
IV Verantwortung
Artikel 4 Verantwortung der Verbände
4.01 Ein Verband ist für das Verhalten seiner Spieler, seiner Offiziellen, seiner Mitglieder, seiner Anhänger und aller Personen verantwortlich, die im Auftrag des Verbands eine Funktion bei einem Spiel ausüben.
9. Annexes
194
4.02 Der Landesverband, der ein Qualifikationsspiel oder die Endrunde auf seinem Gebiet ausrichtet, gilt für dieses Spiel oder Turnier als Ausrichterverband. 4.03 Der Ausrichterverband ist für Ordnung und Sicherheit vor, während und nach den Spielen verantwortlich. Der Ausrichterverband kann für Zwischenfälle jeglicher Art zur Verantwortung gezogen und bestraft werden. 4.04 Prinzipiell müssen Spiele in einem Stadion auf dem Gebiet des jeweiligen Ausrichterverbands ausgetragen werden. In Ausnahmefällen können Spiele auf Entscheidung der UEFA-Administration und/oder der Disziplinarinstanzen auf dem Gebiet eines anderen UEFA-Mitgliedsverbands ausgetragen werden (aus Sicherheitsgründen oder infolge einer Disziplinarmassnahme).
Zusätzliche Verantwortung für die Endrunde
4.05 Die UEFA-Administration informiert die Verbände, die an der Endrunde teilnehmen, über etwaige zusätzliche Richtlinien, Weisungen oder Beschlüsse, die mit der Endrunde zusammenhängen; sie stellt den Verbänden die notwendigen Dokumente zu gegebener Zeit zur Verfügung.
VI Wettbewerbsmodus Artikel 6
Wettbewerbsphasen
6.01 Der Wettbewerb besteht aus einem Qualifikationswettbewerb und einer Endrunde.
Artikel 7 A. Qualifikationswettbewerb
Gruppenbildung
7.01 Die Mannschaften der Endrunden-Ausrichter, Polen und die Ukraine, sind automatisch für die Endrunde qualifiziert. Die übrigen Mannschaften werden in sechs Gruppen à sechs Mannschaften und in drei Gruppen à fünf Mannschaften gelost. Die UEFA-Administration lost die Gruppen nach Abschluss des Qualifikationswettbewerbs der FIFA Fussball-Weltmeisterschaft 2010 aus. Die Auslosung basiert auf einem Setzsystem. Der amtierende Europameister ist immer gesetzt. Die übrigen Verbände werden auf der Grundlage der UEFA-Koeffizientenrangliste für Nationalmannschaften klassiert (vgl. Anhang I, Absatz 1.2.1). Die Entscheidungen der UEFA-Administration sind endgültig. 7.02 Wenn zwei oder mehr Mannschaften denselben Koeffizienten haben, sind die folgenden Kriterien ausschlaggebend. Nur der letzte Qualifikationswettbewerb wird berücksichtigt. Die Kriterien werden in dieser Reihenfolge angewendet: a) UEFA-Koeffizient für Nationalmannschaften, resultierend aus den ausgetragenen Spielen; b) durchschnittliche Tordifferenz; c) durchschnittliche Anzahl erzielter Tore; d) durchschnittliche Anzahl erzielter Auswärtstore; e) Fairplay-Rangliste; f) Losentscheid.
Austragungsmodus für den Qualifikationswettbewerb
7.03 Die Spiele des Qualifikationswettbewerbs werden in Gruppen entsprechend dem Meisterschaftsmodus ausgetragen. Jede Mannschaft spielt in Hin- und Rückspiel gegen jede andere Mannschaft ihrer Gruppe. Die Mannschaften erhalten drei Punkte für einen Sieg, einen Punkt für ein Unentschieden und null Punkte für eine Niederlage.
9. Annexes
195
Punktegleichheit nach den Gruppenspielen 7.04 Wenn zwei oder mehr Mannschaften nach Abschluss der Gruppenspiele punktgleich liegen, werden die folgenden Kriterien angewendet, um die Platzierungen zu ermitteln. Sie werden in dieser Reihenfolge angewendet: a) höhere Punktzahl, die die Mannschaften in den Gruppenspielen der betreffenden Mannschaften erhielten; b) bessere Tordifferenz aus den Gruppenspielen der betreffenden Mannschaften; c) grössere Anzahl Tore aus den Gruppenspielen der betreffenden Mannschaften; d) grössere Anzahl Auswärtstore aus den Gruppenspielen der betreffenden Mannschaften; e) Wenn die Kriterien a) bis d) auf mehrere Mannschaften angewendet worden sind und immer noch zwei oder mehr Mannschaften denselben Platz belegen, werden die Kriterien a) bis d) erneut angewendet, um die Platzierung dieser Mannschaften zu ermitteln. Wenn dieses Verfahren zu keiner Entscheidung führt, werden die Kriterien f) bis j) angewendet; f) bessere Tordifferenz aus allen Gruppenspielen; g) grössere Anzahl erzielter Tore aus allen Gruppenspielen; h) grössere Anzahl erzielter Auswärtstore aus allen Gruppenspielen; i) Fairplay-Rangliste aus allen Gruppenspielen; j) Losentscheid.
Qualifikation für die Endrunde 7.05 Die neun Gruppensieger und der beste Zweitplatzierte qualifizieren sich direkt für die Endrunde. 7.06 Nur die Ergebnisse gegen die Mannschaften auf dem ersten, dritten, vierten und fünften Platz werden berücksichtigt, um den besten Zweitplatzierten zu ermitteln. Zu diesem Zweck werden die folgenden Kriterien in dieser Reihenfolge angewendet: a) höhere Punktzahl, die die Mannschaften in diesen Spielen erhielten; b) bessere Tordifferenz aus diesen Spielen; c) grössere Anzahl erzielter Tore aus diesen Spielen; d) grössere Anzahl erzielter Auswärtstore aus diesen Spielen;
e) Platzierung in der UEFA-Koeffizientenrangliste für Nationalmannschaften (vgl. Anhang I, Absatz 1.2.2); f) Fairplay-Rangliste aus diesen Spielen; g) Losentscheid. 7.07 Die acht übrigen Zweitplatzierten bestreiten Entscheidungsspiele. Die vier Spielpaarungen werden durch Auslosung ermittelt. Die vier Zweitplatzierten mit der besten Platzierung in der UEFA-Koeffizientenrangliste für Nationalmannschaften (vgl. Anhang I, Absatz 1.2.2) sind für die Auslosung gesetzt. Wenn zwei oder mehr dieser Mannschaften denselben Koeffizienten haben, werden die Kriterien von Absatz 7.02 angewendet. Die Entscheidungsspiele werden nach dem K.-o.-System ausgetragen (in Hin- und Rückspiel). Die gesetzten Mannschaften bestreiten das Rückspiel zu Hause. Die Mannschaften, die insgesamt die meisten Tore erzielen, qualifizieren sich für die Endrunde. Andernfalls werden die Bestimmungen von Absatz 7.08 angewendet. 7.08 Für Spiele, die nach dem K.-o.-System ausgetragen werden, wird das folgende Kriterium angewendet: Wenn beide Mannschaften insgesamt gleich viele Tore erzielen, qualifiziert sich die Mannschaft, die mehr Auswärtstore erzielt, für die folgende Runde. Wenn dieses Verfahren zu keiner Entscheidung führt, d.h. wenn beide Mannschaften gleich viele Heim- und Auswärtstore erzielen, wird das Rückspiel um zweimal 15 Minuten verlängert. Wenn in der Verlängerung beide Mannschaften gleich viele Tore erzielen, zählen die Auswärtstore doppelt (d.h. die Gastmannschaft ist qualifiziert). Wenn keine Tore in der Verlängerung erzielt
9. Annexes
196
werden, wird die Mannschaft, die sich für die Endrunde qualifiziert, durch Elfmeterschiessen ermittelt (vgl. Artikel 16).
Artikel 8 B. Endrunde
8.01 Das UEFA-Exekutivkomitee hat den Polnischen Fussballverband (PZPN) und den Ukrainischen Fussballverband (FFU) mit der gemeinsamen Organisation und Ausrichtung der Endrunde betraut.
Gruppenbildung
8.02 16 Mannschaften qualifizieren sich für die Endrunde. Die UEFA-Administration teilt diese 16 Mannschaften in vier Vierergruppen ein (Gruppen A, B, C und D). 8.03 Durch Auslosung werden die vier Gruppen wie folgt gebildet: Gruppe A Gruppe B Gruppe C Gruppe D A1 B1 C1 D1 A2 B2 C2 D2 A3 B3 C3 D3 A4 B4 C4 D4
Koeffizienten
8.04 Die folgenden Mannschaften sind gesetzt: die Ausrichterverbände, der amtierende Europameister, sofern qualifiziert, und ein oder zwei Mannschaften mit den besten Koeffizienten gemäss der UEFA-Koeffizientenrangliste für Nationalmannschaften (vgl. Anhang I, Absatz 1.2.2). Die anderen Endrundenteilnehmer werden den vier Gruppen entsprechend ihren Koeffizienten durch Auslosung zugewiesen). 8.05 Wenn zwei oder mehr der betreffenden Mannschaften denselben Koeffizienten haben, werden die Kriterien von Absatz 7.02 angewendet. Die Ergebnisse der Entscheidungsspiele (vgl. Absatz 7.07) werden dabei nicht berücksichtigt.
Gruppenspielplan
8.06 Jede Mannschaft spielt gegen jede andere Mannschaft ihrer Gruppe nach dem Meisterschaftsmodus (drei Punkte für einen Sieg, ein Punkt für ein Unentschieden, null Punkte für eine Niederlage). Die Gruppenspiele werden entsprechend dem folgenden Schema ausgetragen. Die letzten zwei Spiele in jeder Gruppe müssen beide zeitgleich beginnen. Die erstgenannte Mannschaft gilt als Heimmannschaft. 1. Spieltag 2. Spieltag 3. Spieltag Gruppe A A1 gegen A2 A1 gegen A3 A4 gegen A1 A3 gegen A4 A2 gegen A4 A2 gegen A3 Gruppe B B1 gegen B2 B1 gegen B3 B4 gegen B1 B3 gegen B4 B2 gegen B4 B2 gegen B3 Gruppe C C1 gegen C2 C1 gegen C3 C4 gegen C1 C3 gegen C4 C2 gegen C4 C2 gegen C3 Gruppe D D1 gegen D2 D1 gegen D3 D4 gegen D1 D3 gegen D4 D2 gegen D4 D2 gegen D3
9. Annexes
197
Punktegleichheit nach den Gruppenspielen 8.07 Wenn zwei oder mehr Mannschaften nach Abschluss der Gruppenspiele punktgleich liegen, werden die folgenden Kriterien angewendet, um die Platzierungen zu ermitteln. Sie werden in dieser Reihenfolge angewendet: a) höhere Punktzahl aus den direkten Begegnungen; b) bessere Tordifferenz aus den direkten Begegnungen (wenn mehr als zwei Mannschaften nach Abschluss der Gruppenspiele punktgleich liegen); c) grössere Anzahl erzielter Tore aus den direkten Begegnungen (bei mehr als zwei punktgleichen Mannschaften nach Abschluss der Gruppenspiele); d) bessere Tordifferenz aus allen Gruppenspielen; e) grössere Anzahl erzielter Tore aus allen Gruppenspielen; f) Platzierung in der UEFA-Koeffizientenrangliste für Nationalmannschaften (vgl. Anhang I, Absatz 1.2.2); g) Fairplay-Verhalten der betreffenden Mannschaften (Endrunde); h) Losentscheid. 8.08 Zwei Mannschaften haben die gleiche Punktzahl und die gleiche Tordifferenz. Diese Mannschaften bestreiten ihr letztes Gruppenspiel gegeneinander. Wenn dieses Spiel unentschieden endet, wird die Platzierung der zwei betroffenen Mannschaften durch Elfmeterschiessen ermittelt (vgl. Artikel 16; vorausgesetzt, dass keine anderen Mannschaften derselben Gruppe nach Abschluss der Gruppenspiele die gleiche Punktzahl haben). Wenn mehr als zwei Mannschaften die gleiche Punktzahl haben, werden die Kriterien von Absatz 8.07 angewendet.
Viertelfinale
8.09 Die Sieger und Zweitplatzierten jeder Gruppe bestreiten die Viertelfinalspiele in einem Spiel. Zu diesem Zweck wird das folgende Schema angewendet: Spiel 1 Sieger Gruppe A gegen Zweitplatzierten Gruppe B Spiel 2 Sieger Gruppe B gegen Zweitplatzierten Gruppe A Spiel 3 Sieger Gruppe C gegen Zweitplatzierten Gruppe D Spiel 4 Sieger Gruppe D gegen Zweitplatzierten Gruppe C
Halbfinale
8.10 Die vier Sieger der Viertelfinalspiele bestreiten die Halbfinalspiele in jeweils einer Begegnung. Zu diesem Zweck wird das folgende Schema angewendet: Sieger Spiel 1 gegen Sieger Spiel 3 Sieger Spiel 2 gegen Sieger Spiel 4
Endspiel 8.11 Die Sieger der Halbfinale bestreiten das Endspiel. Gleiche Anzahl Tore in einem Viertel- oder Halbfinalspiel bzw. im Endspiel 8.12 Wenn das Spiel nach Ablauf der regulären Spielzeit unentschieden endet, wird eine Verlängerung von zweimal 15 Minuten gespielt. Wenn auch die Verlängerung unentschieden endet, wird der Sieger durch Elfmeterschiessen ermittelt (vgl. Artikel 16).
VIII Spieldaten, Spielorte und Anstosszeiten Artikel 11
11.01 Der Wettbewerb findet im Anschluss an die FIFA Fussball-Weltmeisterschaft 2010 statt. Er wird über zwei Spielzeiten ausgetragen.
A. Qualifikationswettbewerb
Spieldaten
11.02 Die Gruppenspiele des Qualifikationswettbewerbs der UEFA-Fussball-Europameisterschaft 2010-12 werden für die folgenden zwölf Daten angesetzt: 2010
9. Annexes
198
a) 3./4. und 7. September 2010 b) 8./9. und 12. Oktober 2010
2011
a) 25./26. und 29. März 2011 b) 3./4. und 7. Juni 2011 c) 2./3. und 6. September 2011
d) 7./8. und 11. Oktober 2011 (Daten für die letzten Spiele in allen Gruppen)
11.03 Die Entscheidungsspiele zwischen den übrigen acht Zweitplatzierten werden für die folgenden zwei Daten angesetzt: a) 11./12. November 2011 b) 15. November 2011 11.04 Unter den folgenden Bedingungen können die Gruppenspiele, die in Absatz 11.02 genannt werden, an anderen Daten ausgetragen werden: wenn sich beide betroffenen Verbände einigen und wenn die Grundsätze betreffend das Abstellen von Spielern für Auswahlmannschaften der Verbände gemäss Anhang 1, Artikel 1 des FIFA-Reglements bezüglich Status und Transfer von Spielern eingehalten werden. Die Verbände einer Gruppe haben sich innerhalb von 30 Tagen nach der Auslosung auf die Spielabfolge in ihrer Gruppe zu einigen. Das genaue Datum jedes Spiels muss angegeben werden (z.B. Samstag, 4. September 2010). Wenn die betroffenen Verbände keine Einigung erzielen können, werden die Spiele entsprechend einem Standardspielplan ausgetragen, der von der UEFA-Administration aufgestellt wird. In diesem Standardspielplan werden Faktoren berücksichtigt, die für die Spielorganisation relevant sind, wie z.B. klimatische Bedingungen. 11.05 Aus Gründen der sportlichen Fairness kann die UEFA-Administration verlangen, dass Spiele derselben Gruppe zeitgleich ausgetragen werden. Nachträgliche Datenänderungen setzen die Genehmigung der UEFA-Administration voraus. Der betreffende Ausrichterverband muss in diesem Fall auch die anderen Verbände in derselben Gruppe informieren.
Spielorte und Anstosszeiten
11.06 Die Spielorte werden von den Ausrichterverbänden festgesetzt. Sie müssen den Gastverbänden und der UEFA-Administration spätestens 60 Tage im Voraus bekannt gegeben werden. Wenn ein Ausrichterverband einen Spielort festsetzt, muss er die Dauer der Reise des Gastverbands berücksichtigen. Der Spielort für ein Qualifikationsspiel darf nicht weiter als 120 Bus-Fahrtminuten vom nächsten internationalen Flughafen entfernt sein, es sei denn, der Gastverband erklärt sich damit einverstanden. Die Anstosszeiten müssen der UEFA-Administration spätestens 60 Tage vor dem Spiel bekannt gegeben werden.
Ankunft der Mannschaften am Spielort
11.07 Die Verbände müssen dafür sorgen, dass ihre Mannschaften früh genug am Spielort ankommen, damit die Medienkonferenz vor dem Spiel vor Redaktionsschluss in beiden beteiligten Ländern abgehalten werden kann. Auf jeden Fall müssen die Mannschaften spätestens 24 Stunden vor Spielbeginn am Spielort eintreffen.
9. Annexes
199
B. Endrunde Spieldaten
11.08 Die Endrunde wird vom 8. Juni bis zum 1. Juli 2012 ausgetragen.
Spielorte und Anstosszeiten
11.09 Die UEFA-Administration ist für die Erstellung des Endrundenspielplans verantwortlich. Die Mannschaften haben Anspruch auf mindestens zwei Ruhetage (48 Stunden) zwischen den Spielen.
Ankunft der Mannschaften in den Ausrichterländern
11.10 Die Mannschaften müssen sich spätestens fünf Tage vor ihrem ersten Spiel in ihrem Mannschaftshotel in einem der Ausrichterländer einfinden.
Ankunft der Mannschaften am Spielort
11.11 Die Mannschaften müssen spätestens 24 Stunden vor Spielbeginn in ihrem Transferhotel eintreffen oder sich spätestens 24 Stunden vor Spielbeginn innerhalb eines Umkreises von 60 km um ihren Spielort aufhalten.
Trainingsplätze 11.12 Die UEFA stellt jedem Endrundenteilnehmer einige ausgesuchte Trainingsplätze zur Verfügung. Wenn ein Verband einen Trainingsplatz auswählt, der nicht Teil dieser Vorauswahl ist, trägt er alle dadurch entstehenden Kosten. 11.13 Ab dem fünften Tag vor dem ersten Spiel der Endrunde gelten alle Trainingsplätze, die von den Verbänden verwendet werden, als offizielle Trainingsplätze. Ab diesem Zeitpunkt werden die Bestimmungen von Absatz 28.14 angewendet.
9. Annexes
200
Annex L: PE3 III Pokale und Medaillen
Artikel 5
Pokal
Der Originalpokal, der für die offizielle Pokalübergabe beim Endspiel verwendet wird, bleibt stets im Besitz der UEFA. Der Sieger erhält eine Nachbildung des Pokals, die so groß wie das Original ist, die Siegertrophäe der UEFA Champions League. Ein Verein, der den Wettbewerb dreimal nacheinander oder insgesamt fünfmal gewinnt, erhält ein spezielles Zeichen der Anerkennung. Hat ein Verein den Wettbewerb dreimal nacheinander oder insgesamt fünfmal gewonnen, so fängt die Zählung für diesen Verein wieder bei Null an. Nachbildungen, die den früheren und aktuellen Gewinnern der UEFA Champions League überreicht werden, müssen stets unter der Kontrolle des betreffenden Vereins bleiben und dürfen die Region bzw. das Verbandsgebiet des Vereins nicht ohne die vorherige schriftliche Genehmigung der UEFA verlassen. Die Vereine dürfen nicht erlauben, dass die Nachbildung in einem Zusammenhang verwendet wird, in dem Dritte (einschliesslich Sponsoren des Vereins und andere kommerzielle Partner) auftreten dürfen oder der zu einer Assoziation zwischen Dritten und dem Pokal und/oder dem Wettbewerb führen könnte. Die Vereine müssen Richtlinien zur Verwendung des Pokals jeglicher Art einhalten, die von der UEFA-Administration von Zeit zu Zeit herausgegeben werden.
Medaillen
Der Sieger erhält dreißig Goldmedaillen, der Zweitplatzierte erhält 30
Silbermedaillen. Zusätzliche Medaillen dürfen nicht hergestellt werden.
IV Verantwortung
Artikel 6
Verantwortung der UEFA
Die UEFA schliesst für ihren sich aus vorliegendem Reglement ergebenden Zuständigkeitsbereich folgende Versicherungen ab: Haftpflichtversicherung Zuschauerunfallversicherung (für nur das Endspiel) Gruppenunfallversicherung für UEFA-Delegierte Rechtsschutzversicherung (beschränkt auf strafrechtliche Fälle).
Verantwortung der Verbände und Vereine
Die Vereine sind für das Verhalten ihrer Spieler, Offiziellen, Mitglieder, Anhänger und aller Personen verantwortlich, die im Auftrag des Vereins eine Funktion bei einem Spiel ausüben. Der Heimverein bzw. der Ausrichterverband ist für Ordnung und Sicherheit vor, während und nach dem Spiel verantwortlich. Der Heimverein bzw. der Ausrichterverband kann für Zwischenfälle jeglicher Art zur Verantwortung gezogen und bestraft werden. Der Verein, der als Heimverein gilt, muss die jeweiligen Spiele in Übereinstimmung mit den Anweisungen der UEFA (oder Dritten, die im Auftrag der UEFA agieren) und in Zusammenarbeit mit dem betreffenden Verband austragen. Jedoch gilt der Verein als allein verantwortlich für die Erfüllung all seiner Verpflichtungen in dieser Hinsicht, es sei denn das jeweilige Organ oder die jeweiligen Organe entscheidet (entscheiden) anders.
9. Annexes
201
Unabhängig von der Versicherungsdeckung der UEFA hat jeder Verein bzw. Ausrichterverband auf eigene Kosten bei einer angesehenen Versicherungsgesellschaft Versicherungen für sämtliche Risiken abzuschliessen.
Zu diesem Zweck werden die folgenden Grundsätze angewendet:
a) jeder Verein muss für Versicherungsdeckung sorgen, die alle Risiken in Zusammenhang mit seiner Teilnahme am Wettbewerb abdeckt;
b) zusätzlich muss der Heimverein bzw. der Ausrichterverband Versicherungen gegen sämtliche Risiken abschliessen, die in Zusammenhang mit der Ausrichtung und der Organisation seiner Heimspiele stehen. Diese Versicherungen müssen insbesondere eine Haftpflichtversicherung beinhalten (für alle Dritten, die an den Spielen teilnehmen oder den jeweiligen Spielort besuchen), die angemessene Garantiesummen für Personen- und Sachschäden sowie für reine Vermögensschäden, die den spezifischen Umständen des Vereins bzw. Verbands entsprechen, beinhaltet;
c) entsprechend Buchstabe b) muss der Ausrichterverband des Endspiels Versicherungen abschliessen, die alle seine Risiken in Zusammenhang mit der Ausrichtung und der Organisation des Endspiels abdecken;
d) wenn der Heimverein bzw. der Ausrichterverband nicht Eigentümer des verwendeten Stadions ist, ist er zusätzlich dafür verantwortlich, sicherzustellen, dass der Stadioneigentümer und/oder -betreiber einen angemessenen und umfassenden Versicherungsschutz einschliesslich Haftpflicht- und Gebäudeversicherung gewährleistet.
e) der Heimverein und der Ausrichterverband müssen garantieren, dass die UEFA in allen oben genannten Versicherungsverträgen beinhaltet wird und müssen gewährleisten, dass die UEFA von jeglicher Haftung befreit ist, die durch die Ausrichtung und die Organisation der jeweiligen Spiele
entsteht;
In jedem Falle kann die UEFA von allen Beteiligten verlangen, ihr kostenlos eine schriftliche Haftungsfreizeichnung und/oder Bestätigungen und/oder Kopien der betreffenden Policen in einer der offiziellen Sprachen der UEFA vorzulegen.
Die Vereine müssen dafür sorgen, dass ihre Mannschaften spätestens am Abend vor dem Spiel am Spielort eintreffen und dass sie ihre Verpflichtungen gegenüber den Medien am Tag vor dem Spiel erfüllen können. Auf der Reise zum und vom
Auswärtsspiel ist es dem Gastverein nicht gestattet, andere Spiele zu bestreiten.
V Wettbewerbsmodus
Artikel 7
Anzahl Runden
a) der Qualifikationsphase für die UEFA Champions League:
erste Qualifikationsrunde zweite Qualifikationsrunde dritte Qualifikationsrunde
b) Entscheidungsspielen;
c) der UEFA Champions League:
Gruppenphase (sechs Spieltage) Achtelfinale Viertelfinalspiele Halbfinale Endspiel.
9. Annexes
202
Qualifikationsphase
Die Spiele der Qualifikationsphase werden entsprechend dem Pokalsystem (K.-o.-System) ausgetragen, d.h. die Vereine treten in Hin- und Rückspiel zweimal gegeneinander an. Die Mannschaft, die insgesamt die meisten Tore in beiden Spielen erzielt, qualifiziert sich für die folgende Runde (je nach Fall für die zweite Qualifikationsrunde, die dritte Qualifikationsrunde oder die Entscheidungsspiele). Andernfalls werden die Bestimmungen von Artikel 8 angewendet. Die in der ersten und zweiten Qualifikationsrunde unterlegenen Vereine scheiden aus dem Wettbewerb aus. Die in der dritten Qualifikationsrunde unterlegenen Vereine sind berechtigt, an den Entscheidungsspielen der laufenden UEFA Europa League teilzunehmen. Vereine aus dem gleichen Verband können einander nicht zugelost werden.
Entscheidungsspiele
Die Entscheidungsspiele werden entsprechend dem Pokalsystem (K.-o.-System) gespielt, d.h. die Vereine treten in Hin- und Rückspiel zweimal gegeneinander an. Die Mannschaft, die insgesamt die meisten Tore in beiden Spielen erzielt, qualifiziert sich für die Gruppenphase der UEFA Champions League. Andernfalls werden die Bestimmungen von Artikel 8 angewendet. Die in den Entscheidungsspielen unterlegenen Vereine sind berechtigt, an der Gruppenphase der laufenden UEFA Europa League teilzunehmen. Vereine aus
dem gleichen Verband können einander nicht zugelost werden.
Gruppenphase
Sobald die Entscheidungsspiele abgeschlossen worden sind, werden die 32 übrigen Vereine in acht Vierergruppen gelost. Vereine aus dem gleichen Verband können nicht in dieselbe Gruppe kommen. Jeder Verein spielt ein Heim- und ein Auswärtsspiel gegen jeden anderen Verein seiner Gruppe. Drei Punkte werden für einen Sieg zugesprochen, ein Punkt wird für ein Unentschieden zugesprochen, und kein Punkt wird für eine Niederlage zugesprochen. Die Spielreihenfolge ist wie folgt:
1. Spieltag: 2 gegen 3, 4 gegen 1. 2. Spieltag: 1 gegen 2, 3 gegen 4.
3. Spieltag: 3 gegen 1, 2 gegen 4. 4. Spieltag: 1 gegen 3, 4 gegen 2. 5. Spieltag: 3 gegen 2, 1 gegen 4 6. Spieltag: 2 gegen 1, 4 gegen 3. Wenn zwei oder mehr Mannschaften nach Abschluss der Gruppenspiele punktgleich liegen, werden die folgenden Kriterien angewendet, um die
Platzierungen zu ermitteln (in der aufgeführten Reihenfolge):
a) höhere Punktzahl aus den Gruppenspielen der betreffenden Mannschaften;
b) bessere Tordifferenz aus den Gruppenspielen der betreffenden Mannschaften;
c) grössere Anzahl erzielter Tore in den Gruppenspielen der betreffenden Mannschaften;
d) grössere Anzahl Auswärtstore in den Gruppenspielen der betreffenden Mannschaften;
e) wenn nach Anwendung der Kriterien a) bis d) noch immer zwei Mannschaften denselben Platz belegen, werden die Kriterien a) bis d) erneut angewendet, um die endgültige Platzierung der zwei Mannschaften zu ermitteln. Wenn dieses Verfahren zu keiner Entscheidung führt, werden die Kriterien f) bis h) angewendet;
f) bessere Tordifferenz aus allen Gruppenspielen;
9. Annexes
203
g) grössere Anzahl erzielter Tore in allen Gruppenspielen; h) grössere Anzahl Koeffizientenpunkte, die durch den betreffenden Verein
sowie den Verband des Vereins in den vorhergehenden fünf Spielzeiten
erreicht wurde (vgl. Absatz 9.02).
Die acht Gruppensieger und die acht Gruppenzweiten der Gruppenphase qualifizieren sich für das Achtelfinale. Nach Abschluss der Gruppenphase wechseln die drittplatzierten Vereine jeder Gruppe in das Sechzehntelfinale der laufenden UEFA Europa League über. Die Viertplatzierten jeder Gruppe scheiden aus. Die vier besten drittplatzierten Mannschaften sind im Sechzehntelfinale der UEFA Europa League gesetzt. Die Setzliste wird nach den folgenden Kriterien in der aufgeführten Reihenfolge ermittelt:
a) höhere Punktzahl der Mannschaften aus den Gruppenspielen b) bessere Tordifferenz c) Grössere Anzahl erzielter Tore d) Grössere Anzahl Auswärtstore e) grössere Anzahl Siege f) grössere Anzahl Auswärtssiege g) Grössere Anzahl Koeffizientenpunkte, die der betreffende Verein und sein
Verband in den vorhergehenden fünf Spielzeiten ansammelten (vgl.
Absatz 9.02).
Achtelfinale
Die Spielpaarungen für das Achtelfinale werden ausgelost. Das Achtelfinale wird nach dem K.-o.-System gespielt, d.h. es gibt ein Heimspiel und ein Auswärtsspiel (zwei Spiele). Die UEFA-Administration stellt die Befolgung der folgenden Grundsätze sicher:
a) Vereine aus dem gleichen Verband können einander nicht zugelost werden.
b) Die Sieger und die Zweitplatzierten der gleichen Gruppe können einander nicht zugelost werden.
c) Die Gruppensieger können einander nicht zugelost werden. d) Die Zweitplatzierten können einander nicht zugelost werden. e) Die Zweitplatzierten müssen das Hinspiel zu Hause bestreiten.
Die Mannschaft, die insgesamt die meisten Tore in beiden Spielen erzielt, qualifiziert sich für das Viertelfinale. Andernfalls werden die Bestimmungen von Artikel 8 angewendet.
Viertelfinale Die acht Sieger des Achtelfinales bestreiten das Viertelfinale. Die Spielpaarungen für das Viertelfinale werden ausgelost. Das Viertelfinale wird nach dem K.-o.-System ausgetragen, d.h. es gibt ein Heimspiel und ein Auswärtsspiel (zwei Spiele). Die Mannschaft, die insgesamt die meisten Tore in beiden Spielen erzielt, qualifiziert sich für das Halbfinale. Andernfalls werden die Bestimmungen von Artikel 8 angewendet.
Halbfinale
Das vier Sieger des Viertelfinales bestreiten das Halbfinale. Die Halbfinalbegegnungen werden ausgelost. Das Halbfinale wird nach dem K.-o.-System ausgetragen, d.h. es gibt ein Heimspiel und ein Auswärtsspiel (zwei Spiele). Die Mannschaft, die insgesamt die meisten Tore in beiden Spielen erzielt, qualifiziert sich für das Endspiel. Andernfalls werden die Bestimmungen von Artikel
8 angewendet.
9. Annexes
204
Endspiel
Das Endspiel ist ein einzelnes Spiel. Das Endspiel wird an einem neutralen Spielort ausgetragen. Wenn das Ergebnis am Ende der regulären Spielzeit unentschieden steht, gibt es eine Verlängerung von zweimal 15 Minuten. Wenn eine der Mannschaften in der Verlängerung mehr Tore als die andere Mannschaft erzielt, gewinnt diese Mannschaft. Wenn das Spiel nach der Verlängerung noch immer unentschieden steht, wird der Sieger durch Elfmeterschiessen ermittelt (Artikel 17). Die Bestimmungen von Artikel 8 gelten nicht für das Endspiel.
VII Spielplan, Spieldaten, Spielorte und Anstosszeiten Artikel 12 Spieldaten
Alle Spiele finden gemäss dem UEFA-Spielkalender statt (vgl. Anhang IC). Die Spieldaten des UEFA-Spielkalenders sind endgültig und für alle beteiligten Parteien verbindlich. Die Spieldaten gelten vorbehaltlich der Bestimmungen der Absätze 12.04, 12.05 und 12.06. Die folgenden Grundsätze werden auf diesen Wettbewerb angewendet:
iv. Die Spiele der UEFA Champions League finden dienstags und mittwochs statt. Das Endspiel ist die einzige Ausnahme und findet an einem Samstag statt.
i. Ab den Entscheidungsspielen beschliesst die UEFA-Administration auf der Grundlage der jeweiligen Auslosung, welche Spiele der UEFA Champions League dienstags stattfinden und welche Spiele der UEFA Champions League mittwochs stattfinden. In der Regel trägt jeder Verein gleich viele Spiele dienstags und mittwochs aus. Spiele der gleichen Gruppe finden am gleichen Tag statt. Die UEFA-Administration kann Ausnahmen bewilligen.
Anstosszeiten
In der Regel beginnen die Entscheidungsspiele, die Gruppenspiele, die Spiele des Achtelfinales, des Viertelfinales, des Halbfinales sowie das Endspiel um 20.45 Uhr (MEZ). Die UEFA-Administration kann Ausnahmen bewilligen. Prinzipiell müssen die Spiele des letzten Spieltages innerhalb einer Gruppe gleichzeitig stattfinden. Die UEFA-Administration kann die Anstosszeiten
festsetzen.
Bestätigung der Spielpaarungen für die Qualifikationsphase
Die Verbände der betreffenden Vereine müssen die Spielorte, die Spieldaten und die Anstosszeiten für alle Spiele in den drei Qualifikationsrunden genehmigen und der UEFA-Administration die Genehmigung innerhalb der durch Letztere festgesetzten Frist schriftlich mitteilen. Die UEFA-Administration kann die Spieldaten und die Anstosszeiten gemäss den von der Kommission für Klubwettbewerbe festgelegten Grundsätzen ändern oder bestätigen. Der Verstoss gegen diese Bestimmung kann Disziplinarmassnahmen ergeben.
9. Annexes
205
Automatische Umstellungen
Wenn
- mehr als ein Verein aus der gleichen Stadt oder aus zwei weniger als 50 km auseinander liegenden Städten an der UEFA Champions League teilnimmt und/oder im selben Stadion spielt und - der betreffende Verband und die betreffenden Vereine bei der Wettbewerbsanmeldung ausdrücklich darauf hinweisen, dass ihre Spiele nicht am gleichen Tag stattfinden können dann kann die UEFA-Administration die Spieldaten und die Anstosszeiten gemäss den von der Kommission für Klubwettbewerbe festgelegten Grundsätzen ändern oder bestätigen. Wenn
- mehr als ein Verein aus der gleichen Stadt oder aus zwei weniger als 50 km auseinander liegenden Städten an den UEFA-Klubwettbewerben teilnimmt und/oder im selben Stadion spielt und
- der betreffende Verband und die betreffenden Vereine bei der Wettbewerbsanmeldung ausdrücklich darauf hinweisen, dass ihre Spiele nicht am gleichen Tag stattfinden können
dann haben die Spiele der UEFA Champions League Priorität und die Spiele der UEFA Europa League werden umgestellt.
Spielorte ab den Entscheidungsspielen
Ab den Entscheidungsspielen muss grundsätzlich jeder Verein alle seine Heimspiele im Wettbewerb im selben Stadion bestreiten. Die Spiele finden entweder im Stadion des Heimvereins oder in einem anderen Stadion derselben oder einer anderen Stadt im betreffenden Verbandsgebiet statt. Wenn die UEFA-Administration und/oder die UEFA-Disziplinarinstanzen so entscheiden, können die Spiele aus Sicherheitsgründen oder infolge einer Disziplinarmassnahme auf dem Gebiet eines anderen UEFA-Mitgliedsverbands stattfinden. Prinzipiell genehmigt die UEFA-Administration Spielorte nur, wenn internationale Direktflüge und/oder Charterflüge im Land des betroffen Vereins in zumutbarer Entfernung landen können. Wenn das Spiel in einer anderen Stadt oder in einem anderen Land stattfindet, muss die UEFA-Administration den Spielort genehmigen.
Ausweichstadien
Wenn die UEFA-Administration zu einem beliebigen Zeitpunkt der Saison der Ansicht ist, dass ein Spielort aus irgendeinem Grund für die Durchführung eines Spiels ungeeignet ist, kann die UEFA die Verbände und die betreffenden Vereine konsultieren und sie darum bitten, ein Ausweichstadion in Übereinstimmung mit den UEFA-Anforderungen vorzuschlagen. Wenn die Verbände und die Vereine innerhalb der von der UEFA-Administration gesetzten Frist kein geeignetes Ausweichstadion vorschlagen können, kann die UEFA einen anderen neutralen Spielort bestimmen. In diesem Fall trifft die UEFA zusammen mit dem jeweiligen Verband und den lokalen Behörden alle notwendigen Vorkehrungen für die Ausrichtung des Spiels. In beiden Fällen trägt der Heimverein die Kosten, das Spiel auszutragen. Die UEFA-Administration entscheidet endgültig und zu gegebener Zeit über den Spielort.
9. Annexes
206
Endspiel Auf der Grundlage eines Vertrags zwischen dem Ausrichterverband und der UEFA organisiert ein lokales Organisationskomitee (LOK) das Endspiel. Das Exekutivkomitee wählt das Datum und den Spielort. Prinzipiell kümmert sich jedes Jahr ein anderer Verband um die lokale Organisation des Endspiels.
9. Annexes
207
Annex M: Questionnaire 1: UEFA SLAN – English translation
Evaluation of controlled source text
Participants: English translators of the UEFA Language Services unit Material: Controlled version of an excerpt of the “UEFA Regulations of the UEFA European Football Championship 2010-12”
1. Brief introduction to Controlled Language (CL)
Whereas human translators can often turn badly written texts into well
written translations, MT systems cannot and will always turn bad input into bad
output (Arnold et al. 1994: 25). This is where controlled languages come into play:
they “address this problem by defining guidelines for and restrictions on the
language which is used to author texts” (Nyberg et al. 2003: 245). These
restrictions are explicitly defined by specifying constraints on lexicon, grammar and
style (idem) and aim at “reduc[ing] or eliminat[ing] the use of ambiguous and
complex sentence structures” (Nyberg et al. 2003: 247). Typical CL rules would be
“keep sentences short”, “prefer active to passive voice” or “always use
determiners”. When it comes to MT, texts should ideally be written “with MT in
mind from the beginning” (Bernth and Gdaniec 2001: 176) since rewriting complex
sentences according to controlled language rules is a very time-consuming and
therefore costly process.
There is no single CL for each language, but many individual approaches to
CL which are rarely accessible to the public. The lack of a ‘core rule set’ makes it
particularly difficult for organisations looking to introduce a CL “without reinventing
the wheel” (O’Brien 2003: 105).
A particular important issue in MT-related CL is the acknowledgement that
the input text quality can in no case be surpassed by output text quality. In other
words, a translation will at best have the same stylistic quality as the source text. If
the controlled source text is written in short and concise sentences with
redundancy (resulting from limited use of pronouns, avoiding coordination, etc.),
then the target text will have the same simple and sometimes cumbersome style.
This is very important to bear in mind since “re-introducing a ‘non-controlled style’”
(Nyberg et al. 2003: 274) by carrying out excessive post-editing would undo the
advantages of working with a CL. This topic as well as translators’ often excessive
expectations will be discussed in the next section. In any case, it is evident that
overly restrictive CL rules may introduce usability and productivity problems. As
Mitamura aptly states: it is“essential to find a middle ground which is productive
9. Annexes
208
and acceptable for authors and which promotes high-quality translation” (Mitamura
1999: 52).
2. Instructions
Keeping in mind the above mentioned particularities of controlled language.
please have a look at the annexed excerpt of the Regulations of the UEFA
European Football Championship 2010-12 which has been manually controlled
following precise CL rules (examples: Avoid long and complex sentences; Avoid
overly complex constructions and write clearly; Transform passive constructions
into active constructions; use determiners (the. a) wherever possible; see MT-
oriented writing rules for UEFA regulations). You do not have to look at the whole
excerpt in detail. but read the whole text in order to give a reliable evaluation of its
overall quality. Please do not in parallel look at the official. non-controlled version
of these regulations. Please mark sentences/passages that catch your eye (for
any of the reasons mentioned below) whilst reading. After you have read the text.
please answer the following questionnaire.
Remark: Please disregard formatting.
3. Questionnaire
1.) Which of the following statements concerning the overall text quality would
you agree with?
The text is satisfactory and would not need to be modified. The text is grammatically correct and proper terminology is used. There are grammar mistakes in the text.
x The text may be stylistically imperfect. but it fulfils its main objective. which
means that it transfers accurately all information. The stylistic imperfections are too serious to be accepted. The text is a lot easier to read than the non-controlled. official version. This way of writing would improve the comprehensibility and accessibility of UEFA regulations. The text is too pedestrian and artificial. it does not correspond to a ‘natural’
language. The sentences are too short. chopped and repetitive. The original version is
preferable to the controlled one. x The sentences are short. chopped and repetitive. but a lot more intelligible
than the original long and complicated sentences and therefore preferable to the latter.
Remark: “If I could rephrase the statement. I’d finish with ‘at least equal to the latter’ –
as I think I say in my comment. it’s hard to say one is better than the other. as
stylistically they’re different and neither is perfect. but because the controlled version
is so intelligible. it has the potential to be preferable from a comprehension point of
view. in which case the stylistic shortfalls would be justified in my opinion.”
9. Annexes
209
2.) By giving page and line numbers, please indicate the sentences that you
think would have to be edited. Describe very briefly for each
sentence/passage why you think changes would have to be carried out
(examples: stylistically inacceptable; too pedestrian; too short sentences,
etc.). If you have suggestions for improvement, please write them down.
Page and line Comments Suggested improvements
Example: p.1, ll. 7-8 Stylistically too
pedestrian, better: streamline the text
…
1. p.1. l. 20 Misplaced comma Associations must not develop. create. use. sell or distribute…
2. p.2. l. 18 Too pedestrian The competition consists of a qualifying
competition and a final tournament
3. p.4. l. 24 Misplaced article (“the”)
Only the results against the teams in first. third. fourth and fi fth place…
4. p.4. l.18 Stylistically poor to start sentence with
figure
Sixteen teams…
5. p.4. l. 27 Wrong pronoun the reigning European champion if it qualifies
6. p.6. l. 21 Mix of tenses (I know this is the same as the original. but to
improve on the original and maximise simplicity.
it may be better to harmonise the tenses (see para
7.01))
The competition takes place following the 2010 FIFA World Cup. It is staged over two seasons.
7. p.6. l. 26 Numbers over ten written as figures (except at start of
sentence) – purely stylistic principle. but widely applied
… for the following 12 dates
8. p.7. l. 10 Space before colon
to be deleted
9. p.7. l. 26 Hyphen missing two-hour bus drive
10. p.7. l. 37 Use active voice? responsible for drawing up the match schedule
11. p.7. l. 38 Misplaced article (“the”)
Either “between matches” or “between each match”
12. p.8. l. 1 Inconsistent use of
definite article compared with l. 4 below
Arrival of the teams in the host countries
13. p.8. ll. 5-6 Illogical clause
pairing
Use original ideally. Failing that:
The teams must arrive at their transfer hotel at least 24 hours before kick-off or be within a radius of 60km of the match
venue at least 24 hours before kick-off. OR By 24 hours before kick-off the teams
9. Annexes
210
must be at their transfer hotels or within
a radius of 60km of the match venue.
Feel free to extend the table
3.) To conclude, do you think this text (if necessary including minor
corrections) meets professional requirements?
x Yes No
4.) Assuming Machine Translation does considerably cut translation costs,
would you accept this writing style in your position as English translator and
editor?
x Yes No
5.) Any comments?
For question 1. I could almost have ticked all three of the first statements – there are
almost no grammatical mistakes. so the text is more or less satisfactory and would
not need to be modified much for the purpose at hand. The correct terminology is
used. Stylistically. even though some sentences are pedestrian and artificial. for
regulations the main concern is to be comprehensive and comprehensible. In this
respect. the ideal middle ground would be somewhere in between the controlled and
non-controlled versions: there could be improvements to both. but the principles
applied for the controlled version are sensible principles for regulations drafting
regardless of translation (short. simple sentences in the active voice).
4. Reference list
Primary literature
Controlled version of an excerpt of the UEFA Regulations of the
UEFA European Football Championship 2010-12
Secondary literature
Arnold. Doug. Lorna Balkan. Siety Meijer. R. Lee Humphreys. and Louisa Sadler. 1994.
Machine Translation: An Introductory Guide. Manchester ; Oxford ; Cambridge: NCC
Blackwell.
Nyberg. Eric. Teruko Mitamura. and Willem -Olaf Huijsen. 2003. “Controlled Language for
Authoring and Translation.” In Computers and Translation: A Translator’s Guide. ed. Harold
Somers. 246–282. Benjamins Translation Library 35. Amsterdam ; Philadelphia: J.
Benjamins.
Bernth. Arendse. and Claudia Gdaniec. 2001. “MTranslatability.” Machine Translation 16
(3): pp.175–218.
Mitamura. Teruko. 1999. “Controlled Language for Multilingual Machine Translation.”
Proceedings of MT Summit VII: pp.46–52.
O’Brien. Sharon. 2003. “Controlling Controlled English: An Analysis of Several Controlled
Language Rule Sets.” Proceedings of EAMT/CLAW 2003: pp.105–114.
9. Annexes
211
Annex N: Questionnaire 2: UEFA SLAN – German translation
Qualitative evaluation
Participants: German translators from UEFA SLAN
Material: Translated sentences from an excerpt of the UEFA Regulations of the UEFA European Football Championship 2010-12
Please have a look at the following sentences. They are all translated from an
excerpt of the Regulations of the UEFA European Football Championship 2010-12
using the Machine Translation software Systran. These translated sentences are
taken from different post-editing1 versions (‘raw output’, ‘minimal post-editing’ and
‘perfected post-editing’), which means that the translation quality may strongly vary
from sentence to sentence.
You can either answer the questionnaire in English or in German.
etc. 85 sentences per questionnaire
1 Following Allen’s definition, post-editing (PE) consists of editing, modifying and/or
correcting “pre-translated text that has been processed by an MT system from a source language into (a) target language(s)” (2003: 297). Post-edition can thus be regarded as the
logical counterpart of the revision of a human translation (Koby 2001: 4).
n° Sentence Qualitative evaluation (good, acceptable,
unacceptable, useless)
Any Comments?
Example: Der Ausrichterverband kann zur Verantwortung gezogen werden und bestraft werden für Zwischenfälle
jeglicher Art.
Good Acceptable Unacceptable
Useless
Word order Grammar Repitition
1. Dieser Standardspielplan berücksichtigt Faktoren, die für die Spielorganisation relevant sind, wie
klimatische Bedingungen.
Good Acceptable Unacceptable
Useless
2. Der Originalpokal, der für die offizielle Darstellungszeremonie beim Endspiel und an anderen offiziellen
Veranstaltungen verwendet wird, die von der UEFA genehmigt werden, bleibt im UEFA-Halten und Besitz
stets.
Good Acceptable Unacceptable
Useless
9. Annexes
212
Reference list
Primary literature
Translated sentences of an excerpt of the Regulations of the UEFA
European Football Championship 2010-12
Secondary literature
Allen, Jeffrey. 2003. “Post-editing.” In Computers and Translation: A
Translator’s Guide, ed. Harold Somers. Benjamins Translation Library 35.
Amsterdam ; Philadelphia: J. Benjamins.
Koby, Geoffrey S. 2001. “Editor’s Introduction - Post-Editing of Machine
Translation Output: Who, What, Why, and How (Much).” In Repairing Texts:
Empirical Investigations of Machine Translation Post-Editing Processes, ed.
Geoffrey S. Koby. Translation Studies 5. Kent Ohio: The Kent State Univ.
Press.
9. Annexes
213
Annex O: Questionnaire 3: Translators with different professional
backgrounds
Qualitative evaluation
Participants: Professional translators and MA students in translation degrees Material: Post-edited MT output of the controlled version of an excerpt of the UEFA Regulations of the UEFA European Football Championship 2010-12 1. Brief introduction to Controlled Language (CL)
Whereas human translators can often turn badly written texts into well
written translations, MT systems cannot and will always turn bad input into bad
output (Arnold et al. 1994: 25). This is where controlled languages come into play:
they “address this problem by defining guidelines for and restrictions on the
language which is used to author texts” (Nyberg et al. 2003: 245). These
restrictions are explicitly defined by specifying constraints on lexicon, grammar and
style (idem) and aim at “reduc[ing] or eliminat[ing] the use of ambiguous and
complex sentence structures” (Nyberg et al. 2003: 247). Typical CL rules would be
“keep sentences short”, “prefer active to passive voice” or “always use
determiners”. When it comes to MT, texts should ideally be written “with MT in
mind from the beginning” (Bernth and Gdaniec 2001: 176) since rewriting complex
sentences according to controlled language rules is a very time-consuming and
therefore costly process.
There is no single CL for each language, but many individual approaches to
CL which are rarely accessible to the public. The lack of a ‘core rule set’ makes it
particularly difficult for organisations looking to introduce a CL “without reinventing
the wheel” (O’Brien 2003: 105).
A particular important issue in MT-related CL is the acknowledgement that
the input text quality can in no case be surpassed by output text quality. In other
words, a translation will at best have the same stylistic quality as the source text. If
the controlled source text is written in short and concise sentences with
redundancy (resulting from limited use of pronouns, avoiding coordination, etc.),
then the target text will have the same simple and sometimes cumbersome style.
This is very important to bear in mind since “re-introducing a ‘non-controlled style’”
(Nyberg et al. 2003: 274) by carrying out excessive post-editing would undo the
advantages of working with a CL. This topic as well as translators’ often excessive
expectations will be discussed in the next section. In any case, it is evident that
overly restrictive CL rules may introduce usability and productivity problems. As
Mitamura aptly states: it is“essential to find a middle ground which is productive
9. Annexes
214
and acceptable for authors and which promotes high-quality translation” (Mitamura
1999: 52).
2. Instructions
The source text has been manually controlled following precise CL rules
(examples: Avoid long and complex sentences; Avoid overly complex
constructions and write clearly; Transform passive constructions into active
constructions; use determiners (the. a) wherever possible). The annexed
translation of an excerpt of the Regulations of the UEFA European Football
Championship 2010-12 is not the raw MT output. but the post-edited version of the
raw output. For the post-edition (PE). we have adhered to the following minimal PE
guidelines:
Retain as much of the raw machine translation output as possible. Resist the temptation to delete and rewrite too much.
Rectify only what is grammatically wrong and what is lexically essential for the understanding of the target text (words/phrases that are nonsensical or wrong)
Remember that all the words are probably present in the MT output (possibly in the wrong order).
Make sure that all information is accurately transferred. There is no need to change words for the sake of elegant variation and
originality. even if the style of the translation is repetitive or pedestrian: information accuracy prevails over stylistic considerations.
Don’t spend too much time over a problem (‘When in doubt. don’t’).
Before having a look at the annexed translation. please read the following
questionnaire. Keeping in mind the above mentioned particularities of controlled
language and the PE guidelines I adhered to, please have then a look at the
annexed translation. You do not have to go over the whole excerpt in detail, but
read the whole text in order to give a reliable evaluation of its overall quality. Whilst
reading, please mark sentences/passages that catch your eye for the reasons
described in the questionnaire below (page 3). After you have read the text, please
answer the following questionnaire. You can either answer it in English or in
German.
Remarks:
1.) At UEFA, German texts are generally written according to Germany’s
language standards; however, “ss” is used instead of “ß” (Swiss Standard
German).
2.) Please disregard formatting.
9. Annexes
215
2. Questionnaire
1.) Which of the following 4 statements concerning the translation quality would you
agree with?1
You would not need the source text (or the official German translation) to
understand the target text. The MT output is satisfactory and would not need to be
modified. It is grammatically correct and proper terminology is used. It may be
stylistically imperfect, but it fulfils the main objective, which means that it transfers
accurately all information.
You would not need the source text (or the official German translation) to
understand the target text. Even though the MT output contains minor grammatical
mistakes (word order, word formation, morphology, etc), you would not need to refer
to the source text in order to correct these mistakes.
Your understanding would be improved by consulting the source text due to
significant errors in the MT output (textual and syntactical coherence, textual
pragmatics, word formation, morphology). You would have to re-read the source text
a few times to correct these errors in the MT output.
You cannot understand the translation without consulting the source text. The
translation contains serious errors in any of the categories mentioned above (i.e.
style, word order, word formation, morphology, textual and syntactical coherence,
textual pragmatics, etc.). You could only produce a translation by dismissing most of
the MT output and/or re-translating from scratch.
2.) By giving page and line numbers, please indicate the sentences that you think
would need further post-editing (using the table on the next page). Describe very
briefly for each sentence/passage why you think further changes would have to be
carried out (examples: word order has to be changed; stylistically inacceptable; too
pedestrian; too complicated; too short sentences; too long sentences, etc.). If you
have suggestions for improvement, please write them down.
1 These evaluation criteria are based on Roturier (2004: 8)
9. Annexes
216
Page and line Comments Suggested improvements
Example:
p.1, ll. 7-8
Stylistically too
pedestrian, better: streamline the text
Ein Verband, der den Pokal dreimal
nacheinander oder insgesamt fünfmal gewinnt, erhält ein spezielles Zeichen der Anerkennung.
1.
2.
etc.
Feel free to extend the table
3.) To conclude, do you think this translation (if necessary including further minor
corrections) meets professional needs?
Yes No
4.) Assuming Machine Translation does considerably cut translation costs, would you
accept this translation if you were head of language services in a company or
organisation whatsoever?
Yes No
5.) Any comments?
Reference list
Primary literature
Post-edited MT output of the controlled version of an excerpt of the UEFA Regulations
of the UEFA European Football Championship 2010-12
Secondary literature
Arnold, Doug, Lorna Balkan, Siety Meijer, R. Lee Humphreys, and Louisa Sadler. 1994.
Machine Translation: An Introductory Guide. Manchester ; Oxford ; Cambridge: NCC
Blackwell.
Nyberg, Eric, Teruko Mitamura, and Willem-Olaf Huijsen. 2003. “Controlled Language
for Authoring and Translation.” In Computers and Translation: A Translator’s Guide, ed.
Harold Somers, 246–282. Benjamins Translation Library 35. Amsterdam ; Philadelphia:
J. Benjamins.
Bernth, Arendse, and Claudia Gdaniec. 2001. “MTranslatability.” Machine Translation
16 (3): pp.175–218.
Mitamura, Teruko. 1999. “Controlled Language for Multilingual Machine Translation.”
Proceedings of MT Summit VII: pp.46–52.
O’Brien, Sharon. 2003. “Controlling Controlled English: An Analysis of Several
Controlled Language Rule Sets.” Proceedings of EAMT/CLAW 2003: pp.105–114.
Roturier, Johann. 2004. “Assessing the Set of Controlled Language Rules: Can They
Improve the Performance of Commercial Machine Translation Systems?” Proceedings
of the 26th ASLIB Conference, Translating and the Computer: 14pp.
9. Annexes
217
Annex P: Results of Questionnaire 3 (Translators with different professional
backgrounds), question 2 (please indicate the sentences that you think
would need further post-editing)
Table I: Errors encountered
N° PE1 C1 Criticism
1 In Ausnahmefällen können
Spiele auf dem Gebiet eines anderen UEFA-Mitgliedsverbands
ausgetragen werden, wenn die UEFA-Administration und/oder die
Disziplinarinstanzen dies entscheiden (aus Sicherheitsgründen oder
infolge eine Disziplinarmassnahme).
In Ausnahmefällen kann
auf Entscheidung der UEFA-Administration und/oder der
Disziplinarinstanzen aus Sicherheitsgründen oder infolge einer
Disziplinarmassnahme auf ein Stadion eines anderen UEFA-Mitgliedsverbandes
ausgewichen werden.
Grammar mistake
(Genitiv)
2 Das UEFA-Exekutivkomitee
hat den Polnischen Fussballverband (PZPN) und den Unkrainischen
Fussballverband (FFU) mit der gemeinsamen Organisation und
Ausrichtung der Endrunde betraut.
Das UEFA-
Exekutivkomitee hat den Polnischen Fussballverband (PZPN)
und den Ukrainischen Fussballverband (FFU) mit der gemeinsamen
Organisation und Ausrichtung der Endrunde betraut.
Spelling mistake
3 Der Nationalverband, der ein Qualifikationsspiel oder
die Endrunde auf seinem Gebiet ausrichtet, gilt für dieses Spiel oder Turnier als
Ausrichterverband.
Der Landesverband, auf dessen Gebiet ein
Qualifikationsspiel bzw. die Endrunde ausgerichtet wird, gilt als
Ausrichterverband.
Terminology
4 Finalplakette
3.07 Jeder Finalist erhält eine Plakette.
Finalplaketten
3.07 Die Finalisten erhalten
je eine Plakette.
Consistency, use
plural
5 Die Verbände dürfen nicht
genehmigen, dass die Nachbildung auf eine Art verwendet wird, die eine
Verbindung zwischen Dritten und der Nachbildung und/oder dem Wettbewerb
herstellen könnte.
[…] oder der zu einer
Assoziation zwischen einem Dritten und der Nachbildung und/oder dem
Wettbewerb führen könnte.
Terminology/Consisten
cy ‘Assoziation‘ vs. ‘Verbindung’
9. Annexes
218
Table II: Adjustments necessary due to SL text structure (CL)
n° PE1 C1 HTS1 Critique
1 In Ausnahmefällen
können Spiele auf dem Gebiet eines anderen UEFA-
Mitgliedsverbands ausgetragen werden, wenn die
UEFA-Administration und/oder die
Disziplinarinstanzen dies entscheiden (aus
Sicherheitsgründen oder infolge eine Disziplinarmassnah
me).
Exceptionally,
matches can be played on the territory of another
UEFA member association, if the UEFA
administration and/or the disciplinary bodies
decide so (for reasons of safety or as a result of a
disciplinary measure).
In Ausnahmefällen
kann auf Entscheidung der UEFA-Administration
und/oder der Disziplinarinstanzen aus
Sicherheitsgründen oder infolge einer Disziplinarmassnah
me auf ein Stadion eines anderen UEFA-
Mitgliedsverbandes ausgewichen werden.
Grammatical
error (also see table above)
Unusual syntax, better: change
sentence structure, no brackets
2 Wenn zwei oder mehr dieser
Mannschaften denselben Koeffizienten
haben, sind die folgenden Kriterien ausschlaggebend.
Nur der letzte Qualifikationswettbewerb wird
berücksichtigt. Die Kriterien werden in dieser Reihenfolge
angewendet:
If two or more of these teams have
the same coefficient, the following criteria
are determinant. Only the most recent qualifying
competition is taken into consideration. The criteria are
applied in the order given:
Haben mehrere Mannschaften
denselben Koeffizienten, sind folgende Kriterien,
die sich nur auf den letzten Qualifikationswettbe
werb beziehen, in dieser Reihenfolge ausschlaggebend:
Syntax, separation of
sentence parts into independent
sentences not convenient (same remark
for other similar passages)
3 Wenn ein Verband einen Trainingsplatz wählt, der nicht Teil
dieser Vorauswahl ist, trägt der Verband alle
Kosten, die entstehen.
Should an association choose a training ground
that is not part of this preselection, the association
bears all the costs that are incurred.
Wählt ein Verband einen Trainingsplatz, der nicht zu den
ausgesuchten Plätzen gehört, übernimmt er alle
dadurch entstehenden Kosten.
Unnecessary relative clause, construction
too pedestrian not idiomatic
4 Ein Verband ist für das Verhalten seiner Spieler,
seiner Offiziellen, seiner Mitglieder, seiner Anhänger
und jeder Person verantwortlich, die im Auftrag des
Verbands eine Funktion bei einem Spiel ausübt.
An association is responsible for the behaviour of i ts
players, its officials, its members, its supporters and
every person who carries out a function at a match
on behalf of the association.
Die Verbände tragen die Verantwortung für das Verhalten
ihrer Spieler, Offiziellen, Mitglieder, Anhänger
und aller Personen, die in ihrem Auftrag bei einem Spiel eine
Funktion ausüben.
Repetition of pronouns: pedestrian
5 Verbände dürfen Associations must Die Verbände dürfen Redundancy,
9. Annexes
219
n° PE1 C1 HTS1 Critique
keine
Werbematerialien oder Werbeartikel entwickeln,
herstellen, verwenden, verkaufen oder
verteilen, die eine Darstellung des Pokals oder eine
Nachbildung des Pokals enthalten […].
not develop, create,
use, sell or distribute any promotional
materials or any merchandise that bear a
representation of the trophy or a replica of the trophy
(including, without limitation, images that show the
winners li fting the trophy).
keine
Werbematerialien oder -artikel entwickeln,
herstellen, verwenden, verkaufen oder
verteilen, die eine Darstellung des Pokals oder einer
Nachbildung desselben enthalten […].
here better
ellipsis
6 Unter den folgenden
Bedingungen können die Gruppenspiele, die
in Absatz 11.02 genannt werden, an anderen Daten
ausgetragen werden: wenn beide betroffenen Verbände sich
einigen und wenn die Grundsätze betreffend das
Abstellen von Spielern für Auswahlmannschaft
en der Verbände gemäss Anhang 1, Artikel 1 des FIFA-
Reglements bezüglich Status und Transfer von
Spielern eingehalten werden.
Under the following conditions, the
group matches mentioned in paragraph 11.02
can be played on other dates : if both associations in
question agree and if the principles governing the release of players
for association teams as laid out in Annex 1, Article 1
of the FIFA Regulations for the Status and Transfer
of Players are complied with.
Die unter Absatz 11.02 genannten
Gruppenspiele können an anderen Daten ausgetragen
werden, falls sich die beiden betroffenen Verbände einig sind
und die Grundsätze betreffend das Abstellen von Spielern für
Auswahlmannschaften der Verbände gemäss Anhang 1,
Artikel 1 des FIFA-Reglements bezüglich Status und
Transfer von Spielern eingehalten werden.
If-clauses unusual after
colon
7 Zwei Mannschaften haben die gleiche
Punktzahl und die gleiche Tordifferenz. Diese
Mannschaften bestreiten ihr letztes Gruppenspiel
gegeneinander. Wenn dieses Spiel unentschieden
endet, wird das Ordnen der zwei betroffenen
Mannschaften durch Elfmeterschiessen
ermittelt (vgl.-Artikel
Two teams have the same number
of points and the same goal difference. These
teams play their last group match against each other.
If this match ends in a draw, the ranking of the two
teams in question is determined by kicks from the penalty
mark (see Article 16; provided that no other teams within
the same group
Treffen zwei Mannschaften im
letzten Gruppenspiel aufeinander, die dieselbe Anzahl
Punkte, die gleiche Tordifferenz und dieselbe Anzahl Tore
aufweisen, und endet das betreffende Spiel unentschieden, wird
die Platzierung der beiden Mannschaften durch
Elfmeterschiessen (vgl. Artikel 16) ermittelt,
vorausgesetzt, dass
coherence/style: Begin
sentences with ‘wenn’, splitting up of
sentence into several parts not
appropriate
9. Annexes
220
n° PE1 C1 HTS1 Critique
16; vorausgesetzt
dass keine anderen Mannschaften derselben Gruppe
die gleiche Punktzahl nach Abschluss der
Gruppenspiele haben). Wenn mehr als zwei
Mannschaften die gleiche Punktzahl haben, werden die
Kriterien von Absatz 8.07 angewendet.
have the same
number of points on completion of the group matches).
Should more than two teams have the same number of
points, the criteria of paragraph 8.07 are applied.
keine anderen
Mannschaften derselben Gruppe nach Abschluss der
Gruppenspiele dieselbe Anzahl Punkte haben.
8 Die Verbände müssen Richtlinien
zur Verwendung des Pokals jeglicher Art einhalten, die
die UEFA-Administration von Zeit zu Zeit
herausgeben kann.
The associations must comply with
trophy use guidelines of any kind that the UEFA
administration can issue from time to time.
Die Verbände sind verpflichtet, jegliche
von der UEFA von Zeit zu Zeit herausgegebenen
Richtlinien zur Verwendung des Pokals einzuhalten.
Use of modal verbs
9 Die UEFA-Administration informiert die
Verbände, die an der Endrunde teilnehmen über
etwaige zusätzliche Richtlinien, Weisungen oder
Entscheidungen, die mit der Endrunde
zusammenhängen.
The UEFA administration informs the
associations that participate in the final tournament
about any further guidelines, directives, or
decisions that are related to the final tournament.
Die UEFA-Administration informiert die an der
Endrunde teilnehmenden Verbände über
etwaige zusätzliche Richtlinien, Weisungen oder
Beschlüsse im Zusammenhang mit der Endrunde […].
Pedestrian, streamline
7 Nur die Ergebnisse gegen die Mannschaften auf
dem ersten, dritten, vierten und fünften Platz werden
berücksichtigt, um den besten Zweitplatzierten zu
ermitteln.
Only the results against the teams in first, third, fourth and fifth place are taken into account to determine the best runner-up.
Für die Ermittlung des besten Zweitplatzierten
werden nur die Ergebnisse gegen die erst-, dritt-, viert-
und fünftplatzierten Mannschaften berücksichtigt.
Order of information, topic-comment
9. Annexes
221
Table III: Adjustments necessary due to the translation approach of the
system
n° PE1 C1 HTS1 Critique
1 Drei Punkte werden für einen Sieg, ein Punkt
für ein Unentschieden und null Punkte
für eine Niederlage zugesprochen.
Three points are awarded for a win, one point for
a draw, and no points for a defeat.
Ein Sieg ergibt drei Punkte, ein Unentschieden
einen Punkt und eine Niederlage null Punkte.
- -No grammatical uniformity of enumeration
(‘Punkte’ in plural, ‘Punkt’ in singular) - order of information,
topic-comment
2 Wenn zwei oder
mehr Mannschaften nach Abschluss
der Gruppenspiele punktgleich
liegen, werden die folgenden Kriterien
angewendet, um die Platzierungen zu ermitteln.
If two or more
teams are equal on points on completion of the
group matches, the following criteria are
applied to determine the rankings.
Wenn zwei oder
mehr Mannschaften nach Abschluss der
Gruppenspiele die gleiche Anzahl Punkte aufweisen,
wird die Platzierung nach folgenden Kriterien in dieser
Reihenfolge ermittelt:
Proposition: nominal
style (i.e. ‘werden zur Ermittlung‘)
3 (einschliesslich Bildern, die die Sieger zeigen,
die den Pokal in den Händen halten)
(including, without limitation, images that show the
winners li fting the trophy)
(einschliesslich Bildern von Pokalübergaben,
auf denen der Pokal zu sehen ist)
proposition: auf denen zu sehen ist
4 Zusätzliche
Medaillen werden nicht produziert.
Additional medals
will not be produced.
Die Herstellung
zusätzlicher Medaillen ist nicht erlaubt.
Proposition: nominal
style, passive
5 Die Verbände
müssen Richtlinien zur Verwendung des
Pokals jeglicher Art einhalten, die die UEFA-
Administration von Zeit zu Zeit herausgeben
kann.
The associations
must comply with trophy use guidelines of any
kind that the UEFA administration
can issue from time to time.
Die Verbände sind
verpflichtet, jegliche von der UEFA von Zeit zu
Zeit herausgegebenen Richtlinien zur
Verwendung des Pokals einzuhalten.
One subject: Direct
word-for-word translation, stylistically very
pedestrian; other subject: sense, antecedent (jeglicher
art could refer to use, which is not the case here)
6 Dritte beinhalten, ohne Einschränkung,
Sponsoren der Verbände sowie andere
kommerzielle Partner.
Third parties include, without limitation,
sponsors of the associations as well as other
commercial partners.
[…] in dem Dritte (einschliesslich Sponsoren und
anderer kommerzieller Partner) auftreten
dürfen.
Direct translation:. (without limitation);
7 Auf jeden Fall müssen die
Mannschaften spätestens 24
In any case, the teams must arrive
at the match venue at least 24
[…] damit die Medienkonferenz
vor dem Spiel vor Redaktionsschluss
Too pedestrian, literal translation of
‘in any case’
9. Annexes
222
n° PE1 C1 HTS1 Critique
Stunden vor
Spielbeginn am Spielort ankommen.
hours before kick-
off.
in den beiden
beteiligten Ländern abgehalten werden kann, spätestens
aber 24 Stunden vor Spielbeginn.
8 Der Originalpokal wird für die
offizielle Pokalübergabe beim Endspiel
und für andere offizielle Veranstaltungen
verwendet, die von der UEFA genehmigt
werden.
The original trophy is used for
the official trophy handover at the final and for other
official events that are approved by UEFA.
Der Originalpokal, der für die offizielle
Pokalübergabe beim Endspiel und für andere offizielle,
von der UEFA genehmigte Veranstaltungen
verwendet wird […].
Use of tenses Anteriority would be
more appropriate (wurden); official translation: und für
andere offizielle, von der UEFA genehmigte
Veranstaltungen
Table IV: Critique of sequences equally or similarly occurring in the official
translation
n° PE1 Critique
1 Die übrigen Verbände werden auf der Grundlage der UEFA-Koeffizientenrangliste für Nationalmannschaften klassiert (vgl. Anhang I,
Absatz 1.2.1).
Helvetism
2 […]: wenn sich beide betroffenen Verbände einigen und wenn die Grundsätze betreffend das Abstellen von Spielern für
Auswahlmannschaften der Verbände gemäss Anhang 1, Artikel 1 des FIFA -Reglements bezüglich Status und Transfer von Spielern
eingehalten werden.
Pedestrian construction, prepositions ‘bezüglich‘, ‘betreffend das Abstellen‘
(proposition: hinsichtlich)
3 Die Nachbildungen dürfen das Land des jeweiligen Verbands nicht ohne die vorherige schriftliche Genehmigung der UEFA verlassen.
Personification, too active; proposition: außerhalb der Landesgrenzen gelangen or the
like
4 Die Verbände dürfen die Verwendung einer Nachbildung nicht in einem Zusammenhang genehmigen, in dem Dritte auftreten dürfen.
incomprehensible
5 Die Verbände dürfen nicht genehmigen,
dass die Nachbildung auf eine Art verwendet wird, die eine Verbindung zwischen Dritten und der Nachbildung und/oder dem Wettbewerb
herstellen könnte.
Too complicated, proposition:
‘müssen sicherstellen, dass‘
6 Ausserdem dürfen sie Dritten nicht erlauben, dies zu tun.
proposition: nominal style
7 Der Ausrichterverband ist für Ordnung und
Sicherheit vor, während und nach den Spielen verantwortlich.
Misleading prepositions,
proposition: ‘trägt die Verantwortung für‘,‘muss dafür sorgen‘
8 Enumerations like:
grössere Anzahl Tore aus den Gruppenspielen der betreffenden Mannschaften
Preposition: ‘Anzahl von/an Toren‘
9 […] und die Grundsätze betreffend das Remark: ”too complicated, I don’t
9. Annexes
223
n° PE1 Critique
Abstellen von Spielern für
Auswahlmannschaften der Verbände gemäss Anhang 1, Artikel 1 des FIFA-Reglements bezüglich Status und Transfer von Spielern
eingehalten werden.“
understand this sentence”
10 Die Verbände dürfen die Verwendung einer Nachbildung nicht in einem Zusammenhang genehmigen, in dem Dritte auftreten dürfen.
Use of modal verbs Official translation: the same modal verb is used
11 Der Sieger erhält eine Nachbildung in
Originalgrösse, die Siegertrophäe der UEFA-Fussball-Europameisterschaft.
Criticised by two subjects:
1) stylistically inacceptable, word order hast o be changed 2) grammatically incorrect
proposition: ”Der Sieger erhält eine Nachbildung der Siegert rophäe der
UEFA-Fussball-Europameisterschaft in Originalgrösse.“
12 Die UEFA-Administration lost die Gruppen
nach Abschluss des Qualifikationswettbewerbs der FIFA Fussball -Weltmeisterschaft 2010 aus.
Stylistically inacceptable/ too
pedestrian; simplify and clarify