26
WHAT WE TALK ABOUT WHEN WE TALK ABOUT SYNONYMS (AND WHAT IT CAN TELL US ABOUT THESAURUSES) M. Lynne Murphy: University of Sussex, United Kingdom ([email protected]) Abstract This article uses corpus evidence to examine uses of the word synonym in two ways. First, it examines whether uses of synonym match common dictionary definitions of the word. This turns up both senses of synonym that are missing from general dictionaries and broadenings from the basic ‘sameness of meaning’ sense represented in most dic- tionaries. After reviewing user studies that discuss synonym searches, the article turns to the study of a web-derived corpus of text related to searching for synonyms. The corpus gives insight into the types of expressions that people seek synonyms for, the reasons they search for them, and how well thesauruses meet those needs. These are considered with reference to seven electronic thesauruses. The data indicate types of expression for which thesaurus treatment could be improved, including multiword, closed-class, and ‘vulgar’ expressions. Suggestions are made for future directions in electronic thesaurus design and usability research. 1. Introduction I have argued elsewhere (see Murphy 2003) that synonymy can be considered a metalexical relation; that is, what we, as speakers, know about synonyms is higher-level knowledge about our language rather than knowledge of the lan- guage that we use in language production. This means that synonym relations are linguistic knowledge that people reason about, to such an extent that many English speakers are comfortable using metalinguistic terminology like syno- nym and taking part in conversations (with themselves or others) about which words ‘have the same meaning as’ others or which word is ‘the best synonym’ for another. We can exploit these conversations in order to gain insight into what potential thesaurus users mean when they say synonym, and we can con- sider whether these insights can inform lexicographers. This article thus inves- tigates the meaning and use of synonym in existing corpora and a purpose-built International Journal of Lexicography, Vol. 26 No. 3, pp. 279–304 doi:10.1093/ijl/ect023 Advance access publication 2 August 2013 279 # 2013 Oxford University Press. All rights reserved. For permissions, please email: [email protected] at University of California, San Francisco on May 10, 2014 http://ijl.oxfordjournals.org/ Downloaded from

What We Talk About When We Talk About Synonyms: (and What it Can Tell Us About Thesauruses)

  • Upload
    m-l

  • View
    217

  • Download
    2

Embed Size (px)

Citation preview

Page 1: What We Talk About When We Talk About Synonyms: (and What it Can Tell Us About Thesauruses)

WHAT WE TALK ABOUT WHEN WETALK ABOUT SYNONYMS(AND WHAT IT CAN TELL USABOUT THESAURUSES)

M. Lynne Murphy: University of Sussex, United Kingdom ([email protected])

Abstract

This article uses corpus evidence to examine uses of the word synonym in two ways.

First, it examines whether uses of synonym match common dictionary definitions of the

word. This turns up both senses of synonym that are missing from general dictionaries

and broadenings from the basic ‘sameness of meaning’ sense represented in most dic-

tionaries. After reviewing user studies that discuss synonym searches, the article turns to

the study of a web-derived corpus of text related to searching for synonyms. The corpus

gives insight into the types of expressions that people seek synonyms for, the reasons

they search for them, and how well thesauruses meet those needs. These are considered

with reference to seven electronic thesauruses. The data indicate types of expression for

which thesaurus treatment could be improved, including multiword, closed-class, and

‘vulgar’ expressions. Suggestions are made for future directions in electronic thesaurus

design and usability research.

1. Introduction

I have argued elsewhere (see Murphy 2003) that synonymy can be considered a

metalexical relation; that is, what we, as speakers, know about synonyms is

higher-level knowledge about our language rather than knowledge of the lan-

guage that we use in language production. This means that synonym relations

are linguistic knowledge that people reason about, to such an extent that many

English speakers are comfortable using metalinguistic terminology like syno-

nym and taking part in conversations (with themselves or others) about which

words ‘have the same meaning as’ others or which word is ‘the best synonym’

for another. We can exploit these conversations in order to gain insight into

what potential thesaurus users mean when they say synonym, and we can con-

sider whether these insights can inform lexicographers. This article thus inves-

tigates the meaning and use of synonym in existing corpora and a purpose-built

International Journal of Lexicography, Vol. 26 No. 3, pp. 279–304doi:10.1093/ijl/ect023 Advance access publication 2 August 2013 279

# 2013 Oxford University Press. All rights reserved. For permissions,please email: [email protected]

at University of C

alifornia, San Francisco on May 10, 2014

http://ijl.oxfordjournals.org/D

ownloaded from

Page 2: What We Talk About When We Talk About Synonyms: (and What it Can Tell Us About Thesauruses)

web-derived corpus in order to investigate what non-lexicographers mean when

they say synonym and what problems they report in searching for synonyms.

In order to reconsider the senses of synonym, in section 2 I examine the

word’s use in three English corpora and compare the senses for synonym

found there to those in dictionary entries for the term. I then turn to the

question of how people interact with dictionaries and thesauruses in order to

find and use synonyms. To do this, I first briefly review literature that touches

on dictionary (or thesaurus) users and synonymy in section 3. In section 4, I

describe a web-derived corpus of discourse concerning synonym searches. The

corpus consists of extracts of personal writing that contain occurrences of

phrases such as need a synonym, find a synonym for, and better synonym. In

section 5 I identify the writers’ reasons for seeking synonyms and analyse some

of the types of expressions that pose problems for synonym seekers. In doing

so, I consider how well seven electronic thesauruses serve the synonym seekers’

needs.

Before continuing this discussion, a brief comment on the terminology for

lexicographical resources: Thesaurus is often treated as synonymous with dic-

tionary of synonyms by both publishers and dictionary users, although

(meta)lexicographers sometimes distinguish between orthographically orga-

nized dictionaries of synonyms and semantically organized thesauruses. With

electronic access to these resources, the distinction between alphabetic or se-

mantic organization matters less, since the user typically accesses entries by

typing a word into a search box. For ease of discussion, I use thesaurus here to

refer to any resource whose main purpose is the presentation of related words,

rather than the presentation of definitions, pronunciation, usage information,

etc. Of course, many dictionaries try to combine the semasiological and ono-

masiological jobs, presenting lists of related words after a definition. These are

not the main focus here, but some of the findings may be relevant to those types

of resources as well.

2. The senses of synonym

How do lexicographers treat the word synonym, and does that differ from how

synonym is used outside linguistics and lexicography? This section discusses

English dictionary treatments of the word synonym in comparison to corpus-

derived examples of the word’s use in order to determine whether there is

consonance among the meaning(s) of synonym in common use, the definitions

of synonym in dictionaries, and, in §5.3, the presentation of synonyms in the-

sauruses. The corpora used were the BYU-British National Corpus (BNC;

Davies 2004–), the Corpus of Contemporary American English (COCA;

Davies 2008–), and the EnTenTen08 corpus (via Sketch Engine; Kilgarriff

et al. 2004). For BNC and COCA, I was able to examine all instances of

synonym (152 and 276, respectively). From the ten billion words of the

280 M. Lynne Murphy

at University of C

alifornia, San Francisco on May 10, 2014

http://ijl.oxfordjournals.org/D

ownloaded from

Page 3: What We Talk About When We Talk About Synonyms: (and What it Can Tell Us About Thesauruses)

web-derived EnTenTen corpus, I took a random sample of 200 instances of

synonym(s).

Dictionaries intended for native speakers generally represent the same three

senses for synonym in the same order. The definitions from the American

Heritage Dictionary (AHD5) entry are presented in (1) – (3), with examples

(in italics) drawn from the corpora. The same senses are listed in the same order

in Collins English Dictionary (CED), Merriam-Webster On-line Dictionary

(MWOD), Oxford Dictionary of English (ODE), and Random House

Dictionary (RHD; via dictionary.com).

(1) A word having the same or nearly the same meaning as another word or

other words in a language.

It is also disturbing that he has used ‘overwhelming pain’ as a synonym

for paradoxical pain. (BNC: British Medical Journal, 1980)

(2) A word or expression that serves as a figurative or symbolic substitute for

another.

Islam has turned into a synonym for terrorism . . . (EnTenTen08: http://

zope.gush-shalom.org/home/en/channels/avnery/1207434781)

(3) Biology One of two or more scientific names that have been applied to

the same species or other taxonomic group.

Genus: NOMAPHILA Species: Nomaphila stricta Synonym:

Hygrophila stricta (BNC: W. V. de Thabrew, Popular tropical aquar-

ium plants. Cheltenham: Thornhill Press, 1981)

The corpora show, unsurprisingly, that most of uses of synonym fit into one of

these three sense categories, although it is sometimes difficult to distinguish

between the first two senses in text.

In addition to these, the corpora show additional senses of synonym in com-

puter jargon. One of the senses is defined within one of the BNC examples,

shown in (4).

(4) Computing two records allocated to one address

Whenever records are added to an existing file, as both home and

synonym records from the original load have already been stored

(BNC: O. Hanson, Design of computer data files, London: Pitman

Publishing, 1989)

This sense occurs repeatedly in the BNC data, largely due to repeated instances

from the same book. More recent data from EnTenTen08 shows a more gen-

eral use in computing, which involves ‘loosening’ the condition in sense (1) that

synonymy involves related words or linguistic expressions:

(5) Computing A symbol having the same value as another symbol or other

symbols within a symbolic system.

What we talk about when we talk about synonyms 281

at University of C

alifornia, San Francisco on May 10, 2014

http://ijl.oxfordjournals.org/D

ownloaded from

Page 4: What We Talk About When We Talk About Synonyms: (and What it Can Tell Us About Thesauruses)

You can also use & as a synonym of \0.

[EnTenTen08: http://www.regular-expressions.info/tcl.html]

The condition of sense (1) that the two words belong to ‘a language’ (which

is reasonably interpreted as ‘a single language’) is also loosened on some uses of

synonym, as in (6).

(6) An expression that is semantically equivalent to another in another lan-

guage; a translational equivalent.

The term ‘dosha’ has no equivalent in the English language, but any

discussion of Ayurveda must include the three doshas [. . .]. But, just

as there is no good synonym for ‘dosha,’ there is also no quick and easy

definition. (COCA: Vegetarian Times 300: 54, 2002)

Language-blind use of synonym is not unheard-of in linguistics and (especially)

the philosophy of language. In fact, none of the ten definitions of synonymy

from semantics texts surveyed in Murphy 2003 (p. 141) mentions sameness of

language as a criterion for synonymy, and only three include the condition of

substitutability or interchangeability in a sentential context, which entails

sameness of language. Nevertheless, sameness of language is usually implied.

The EnTenTen08 corpus sample included a protest against this usage from the

Language Hat blog:

I can’t go along with Joel, who doesn’t ‘have any problem with considering

terms in different languages to be synonyms’—that seems to me to stretch

the sense of synonym beyond the bounds of usefulness.

(http://www.languagehat.com/archives/002641.php, EnTenTen08)

Finally, while one might not go so far as to label these as a separate sense

from (6), the corpora include examples that seem to be best paraphrased as

‘A polysemous word in another language’, as in (7) and (8).

(7) To ingest, incorporarse, a synonym for loving and eating, must seem fairly

odd, a capricious use of language (COCA: A. J. Ponte, ‘Meaning to Eat’,

Massachusetts Review 42.2, 2001)

(8) The word ghanika, or food, is a synonym for ‘yam.’ (COCA:

M. Lepowsky, ‘The way of the ancestors’, Ethology 30.3, 1991)

In these examples, the English-speaking writer points out a single non-English

word that lexicalizes concepts represented by distinct words in English. While

synonym here could be interpreted as ‘translational equivalent’ (e.g. ghani-

ka= yam), the writer seems to be highlighting the fact that the non-English

word lexicalizes concepts represented by unrelated words in English. Within

English, it would be odd to look for a synonym for loving and eating, as in (7),

282 M. Lynne Murphy

at University of C

alifornia, San Francisco on May 10, 2014

http://ijl.oxfordjournals.org/D

ownloaded from

Page 5: What We Talk About When We Talk About Synonyms: (and What it Can Tell Us About Thesauruses)

since they do not comprise a meaning. In these cases, use of synonym seems to

arise from a desire to talk about polysemy as well as equivalence.

Dictionary definitions like (1) are vague about the semantic requirements of

synonymy, in that ‘having the same or similar meaning’ could refer to either

intensional-and-referential equivalence, intensional similarity or equivalence,

or referential (only) similarity or equivalence. To illustrate, queen and female

monarch are near-synonyms on intensional and referential criteria. Both mean

‘female sovereign heads of state’, and thus the referents of queen are also

referents of female monarch. (They are prevented from perfect synonymy as

they are really in a close hyponym relation; all queens are female monarchs, but

not all female monarchs are queens — but the example will have to suffice,

since perfect synonymy is exceedingly rare.) It is also possible to have inten-

sional near-equivalence without referential substitutability; Cruse (1986)

labelled this relation plesionymy. For example, while queen and empress both

have senses that include ‘female monarch’, they generally do not denote

the same set of people. In the referential-only case, Elizabeth II, the queen,

Defender of the Faith, and Harry’s paternal grandmother can all refer to the

same individual, but their referential similarity is not a product of the senses

of these expressions; the relations between them are coincident, but not

necessary.

Which of these interpretations of synonym is to be taken from the ‘sameness

of meaning’ definitions in general-use dictionaries? Assuming that the sense of

meaning in the AHD5 definition in (1) is present in the meaning entry in that

dictionary (shown in (9)), we are led to assume that synonymy is similarity of

sense (intensional meaning):

(9) mean�ing n.

1. Something that is conveyed or signified; sense or significance.

2. Something that one wishes to convey, especially by language: The

writer’s meaning was obscured by his convoluted prose.

3. An interpreted goal, intent, or end: ‘The central meaning of his pontifi-

cate is to restore papal authority’ (Conor Cruise O’Brien).

4. Inner significance: ‘But who can comprehend the meaning of the voice of

the city?’ (O. Henry).

Only two of the dictionaries consulted include a ‘referent of an expression’

sense of meaning (CED, MWOD). In each case, it is the final sense in the

entry and is treated as a specialist logical or philosophical term:

(10) 5. Philosophy

a. the sense of an expression; its connotation

b. the reference of an expression; its denotation. In recent philosophical

writings meaning can be used in both the above senses (CED)

What we talk about when we talk about synonyms 283

at University of C

alifornia, San Francisco on May 10, 2014

http://ijl.oxfordjournals.org/D

ownloaded from

Page 6: What We Talk About When We Talk About Synonyms: (and What it Can Tell Us About Thesauruses)

(11) 4

a: the logical connotation of a word or phrase

b: the logical denotation or extension of a word or phrase (MWOD)

Thus dictionary treatments of sense (1) of synonym can be understood to

favour the similarity-of-intension criterion for synonymy, since their definitions

of meaning are intensional in nature. They tend not to explicitly mention

referential equivalence, leaving room for plesionyms to be included in the

understanding of synonym and ignoring the possibility that, say, insect, para-

site, and pest could be considered synonyms. The intensional definition of

synonym found in dictionaries is also the general criterion used in the crafting

of thesaurus entries. Part of the purpose of the second corpus study here is to

determine whether people who look for synonyms are using these intensional

criteria.

3. Past work on dictionary users and synonyms

What do we know about people’s use of thesauruses or dictionaries to find

synonyms? While we can rely on some research about how people use diction-

aries, relatively little of it mentions using dictionaries for synonym searches.

The literature on thesaurus use is scant.

Since synonym searches are usually needed in language production, rather

than comprehension, they have been relatively neglected in user studies that

have assumed that dictionaries are used by readers. This bias has continued

into the electronic age (Dziemianko 2010), despite the fact that the act of

writing has increasingly become a computer-based activity, where lexicographic

tools are readily available. The reader-orientation in user studies echoes the

general tendency toward reader-orientation in lexicography: ‘only [a] few [dic-

tionaries] are really well suited for the productive task. Indeed, most diction-

aries have been built for the reader rather than for the writer’ (Zock et al. 2010:

205). Nevertheless, where researchers have looked for the evidence, they have

found that dictionaries are at least as important to users for productive tasks as

for reading. For instance, Kipfer (1985, cited in Kipfer 1987) found that high

school students used dictionaries more when writing than when reading, and

Siegel’s (2007) survey of undergraduate dictionary users at an American uni-

versity found that ‘students still consider dictionaries important tools, espe-

cially for their own writing’ (p. 24).

When usage surveys have asked about dictionaries as tools for language

production — and sometimes when they have not — it becomes clear that

finding a synonym is a common motivation for using lexicographical resources.

For instance, when Siegel (2007) used a multiple-choice question to ask native-

English-speaking university students why they use dictionaries, she did not

include synonym-finding as one of the choices, and therefore did not initially

284 M. Lynne Murphy

at University of C

alifornia, San Francisco on May 10, 2014

http://ijl.oxfordjournals.org/D

ownloaded from

Page 7: What We Talk About When We Talk About Synonyms: (and What it Can Tell Us About Thesauruses)

identify synonym-seeking as a key reason for using dictionaries. But when she

asked an open-ended question about how dictionaries could be made more

useful, she found that the majority of advanced students ‘want dictionaries

to display a wide choice of one-word synonyms, as if they were thesauruses

as well’ (2007: 24).

Desire for synonym information is found in non-native speakers’ use of

monolingual dictionaries as well. Asking French students about their use of

monolingual English dictionaries, Bejoint (1981) found that synonym searches

were the third most frequently reported use, after finding meanings and check-

ing grammatical patterns, and 68% of the subjects reported that they used

synonym information in dictionaries. This led Bejoint to comment: ‘The per-

centage of students using synonyms is important for lexicographers to note,

especially since this is not usually a developed feature of British monolingual

dictionaries’ (1981: 218). Battenburg (1991: 96) found that beginning language

learners tend to view monolingual dictionaries exclusively as sources for senses

and synonyms, commenting that their experience with bilingual reference tools

is to blame for the students’ ‘misguided notion’. As students in his study

became more advanced users of the language, they more often used dictionaries

for other purposes, but still 70% of the most advanced group used dictionaries

to look up synonyms ‘often’ or ‘sometimes’. Battenburg’s learner study pro-

vides an interesting contrast to Siegel’s findings. While Battenburg’s advanced

learner-users were less likely than beginners to seek synonym information in

dictionaries, Siegel’s advanced university students were more likely than their

freshman counterparts to want more explicit synonym information in diction-

aries. There might be several explanations for this difference, but one has to

wonder whether a key factor was that Battenburg’s subjects relied on print

dictionaries, whereas Siegel’s had easy access to electronic lexicographical tools

in their word processors. Siegel’s American undergraduate subjects noted the

difference, with comments like ‘I rarely need a physical dictionary. Thesauruses

are much more helpful in daily use’ (2007: 40) and ‘[dictionaries] are useful,

I just never use them. The online synonym creator in [Microsoft] Word, how-

ever, is very useful’ (2007: 40).

Because many electronic resources are on the internet, it is potentially easier

than ever for lexicographers to access information about dictionary and the-

saurus use by tracking users’ word searches (e.g. de Schryver and Joffe 2004).

But this is only helpful to a certain degree. We can tell whether someone has

looked for a particular string of characters and thereby identify common look-

up problems (e.g. misspellings), improve the electronic search process, or

decide whether new headwords or derivations should be included. But such

studies generally cannot tell us which information in the entry was found to be

relevant to the purpose of the search or whether the user was satisfied with

what they found. They also cannot tell us when people just do not bother to

consult a thesaurus even though they have a synonym query.

What we talk about when we talk about synonyms 285

at University of C

alifornia, San Francisco on May 10, 2014

http://ijl.oxfordjournals.org/D

ownloaded from

Page 8: What We Talk About When We Talk About Synonyms: (and What it Can Tell Us About Thesauruses)

Merriam-Webster collects some information about user motivations in their

online dictionary and thesaurus interface. There is a comments section on each

entry page, with the question prompt: What made you want to look up [word]?

Please tell us where you read or heard it (including the quote, if possible). This

again assumes that dictionary and thesaurus users are readers, and therefore

the site does not seem to encourage answers like ‘because I needed a synonym

for it’ or ‘I wanted to know how to spell it’. Reflecting this (and the fact that

the dictionary, rather than the thesaurus, is the default point of access on the

site), far fewer comments are left by users of the thesaurus than by users of the

dictionary. I examined the comment sections for ten headwords (randomly

selected from the 301 synonym search terms in the corpus described in §4)

and found that while all had substantial comment threads in on the dictionary

page, only one (gloomy) garnered any comments on the thesaurus page (which

did not, in any case, answer what made you want to look this up?).

Chon (2009) got around these problems by using a think-aloud writing task

to track dictionary use by Korean high-intermediate learners of English, noting

that ‘writing particularly favours the use of electronic dictionary resources’

(2009: 24). While Chon’s subjects were most likely to use a Korean-English

dictionary (70.2% of the time), the second most-accessed resource was the

thesaurus bundled in Microsoft Word. More than one fifth (21.6%) of the

learners’ dictionary use was aimed at finding ‘alternative words’ (i.e. synonyms)

for a word they had in mind, usually to avoid repetition (56.5%) but also to

increase the sophistication of their writing (41.9%), or to vary the style (1.6%).

They were more likely to encounter problems in their dictionary use when

looking for alternative words (problems in 54.8% of their attempts) than

when looking up an unknown word (34.3%), checking on the meaning of a

partially known word (35.3%), or checking an unknown spelling (7.1%).

Nearly 40% of the dictionary-based problems (DBPs) the learners faced were

secondary to the original word-based problem, potentially indicating problems

in the usability of the resources. For instance, in consulting the Microsoft

Word thesaurus, the learner finds candidate words, but not enough informa-

tion about the differences between them that would allow him to select one of

the candidates. While Chon recommends that users consider CD-ROM-based

resources like the Longman Language Activator instead of word-processor tools

or internet sites, the main recommendation to makers of electronic dictionaries

is that:

[. . .] thesaurus type dictionaries, which merely provide lists of words, need

to provide information on the differences of the near synonymous words.

For instance, using the thesaurus to seek a synonym to avoid repetition will

generate more DBPs than when dictionaries are used to seek spelling

information. Also, learners may be helped to distinguish between

(quasi-)synonyms when the dictionary or thesaurus carries more example

286 M. Lynne Murphy

at University of C

alifornia, San Francisco on May 10, 2014

http://ijl.oxfordjournals.org/D

ownloaded from

Page 9: What We Talk About When We Talk About Synonyms: (and What it Can Tell Us About Thesauruses)

sentences that demonstrate subtle differences (e.g. register, style, degree of

politeness) between words. (Chon 2009: 50)

In summary, the existing literature on synonym searches is scant and suffers

from a number of biases: (a) there is little specifically on synonym-searching

and thesaurus use, (b) dictionary use research often implicitly or explicitly

assumes that the user is engaged in receptive rather than productive tasks,

(c) investigations have often relied on retrospective surveys which are biased

both by the questions asked and by the user’s memory, and (d) the subjects of

the research are almost always exclusively students. Where investigators look

for (or leave open the possibility of) the importance of synonym-searches

among dictionary users’ needs, they tend to find it. The investigation discussed

in the following sections cannot overcome all these biases, but it does offer a

novel complement to existing studies.

4. Synonym commentary on theweb: methodology

While the internet has allowed for the development of new types of reference

material and new interfaces for accessing lexicographical resources, it has also

provided competitors to the established dictionary publishers, including self-

promoted language mavens, discussion forums, and crowd-sourced reference

works. Whereas fifty years ago, if you wanted to find a synonym, you needed

to be near either a reference book or a friend with a good vocabulary, these

days people wanting to know about word usage may well tweet their question

to hundreds of followers, ask it on an internet forum, or email it to a language

blogger. Sometimes they do so after encountering ‘dictionary-based problems’,

in Chon’s (2009) terms. Here I report on an experiment to determine what can

be gleaned from this publicly available evidence of writers’ and readers’ syno-

nym needs.

A corpus of synonym-search commentaries was gathered by retrieving syno-

nym-search-related word strings from a set of web domains that signal user-

driven content. A custom search tool (designed by Eric Kow) was used to

access the Microsoft Bing search engine (http://www.bing.com) and retrieve

up to fifty hits at a time for a series of searches involving each of the strings in

Table 1 at each of the web domains in Table 2.

The search strings were chosen after some experimentation with a larger

range of phrases (including need a synonym, want a synonym, etc.) in standard

web searches. The resulting six gave the least ‘noise’ and the most synonym-

search reflections and queries. These differed in form to some extent. For

example, I need a synonym was the most productive search string with need;

the I subject ensured that the results were mostly about individuals’ searches

for synonyms, rather than advice about writing or word-processing (e.g. If you

need a synonym, you simply have to click on the thesaurus tab). Other strings

What we talk about when we talk about synonyms 287

at University of C

alifornia, San Francisco on May 10, 2014

http://ijl.oxfordjournals.org/D

ownloaded from

Page 10: What We Talk About When We Talk About Synonyms: (and What it Can Tell Us About Thesauruses)

worked better with a bare verb: for example find a synonym allows for I need to

find a synonym, I want to find a synonym, I was trying to find a synonym,

Help me find a synonym, and so forth.

The searches were specialized to web domains that mostly comprise un-

edited writing by the general public. These included the two major blog

hosts, the most popular question-answer forum (according to the pilot results),

and Word Reference Forum. The productivity of these domains for this task

was established through some initial experimentation, and where pilot searches

resulted in relevant data, that data was included in the study as well. Thus,

in addition to the four deliberately searched domains, others are included in

Table 2, each of which counts for usually one, but as many as three, data

points.

Table 2: Web sources for data

Blog hosts

methodical search blogspot.com

wordpress.com

additional datum taprepeatedly.com

General question-answer forums

methodical search answers.yahoo.com

additional data ask.com/answers

funadvice.com

jishka.com

stackoverflow.com

wikianswers.com

Specialized forums

methodical search forum.wordreference.com

additional data englishtest.net/forum

englishforums.com

proz.com [translation forum]

songmeanings.net

tribalwar.com/forums [gaming forum]

Table 1: Phrasal strings searched on the World-Wide Web

best synonym for better synonym for

find a synonym looking for a synonym

I need(ed) a synonym what is a synonym for

288 M. Lynne Murphy

at University of C

alifornia, San Francisco on May 10, 2014

http://ijl.oxfordjournals.org/D

ownloaded from

Page 11: What We Talk About When We Talk About Synonyms: (and What it Can Tell Us About Thesauruses)

This process gave 873 instances of the search phrases. Search results were

then coded in order to include only those that involved individuals remarking

upon their desire to find a synonym or their experience using a thesaurus.

Discounted were any cases (a) that included a non-lexical sense of synonym,

(b) that comprised general discussion about synonyms, rather than about a

particular synonym search, (c) for which the search term was unclear, or (d)

that were clearly facetious (particularly and repetitively, What is a synonym for

synonym? or Need a synonym for ‘th[is] website sucks’). Examples that con-

stituted a synonym exercise set by a teacher (e.g. Find a synonym for each of the

following words) or users asking for synonyms for lists of words (which I

assumed to be homework assignments) were considered as categories, but

not analysed further. After duplicates were deleted, this left 301 contexts to

analyse in some depth.

The resulting data must be treated more qualitatively than quantitatively;

although numerical trends can be noted in the data, its collection was biased by

the vagaries of the search device and the self-selection, as it were, of the data

contributors, in that they are restricted to those who have chosen to remark

upon their synonym searches on the internet. Demographic data on the writers

is limited to what is explicitly mentioned on the websites and what can be

surmised from the URLs, for example whether their question appeared on

the Australian au.answers.yahoo com or the Indian in.answers.yahoo.com.

5. What we talk about whenwe talk about synonyms

The internet discussions of synonyms can be considered in terms of (a) the

activities that lead people to look for synonyms, (b) the reasons for wanting a

synonym within that activity, (c) the types of expressions for which synonyms

are required, and finally (d) the extent to which these needs are met by the-

sauruses. I briefly consider the first two of these before turning to the last two.

5.1 Synonym searchers and theiractivities

One third (101) of the derived examples do not mention (at least not in the

sampled string of text) why the synonym is needed. Within the remaining 200

examples, several synonym-requiring activities and therefore types of synonym

searcher (henceforth ‘SS’) can be identified: students, story-writers (e.g. fan-

fiction authors), fans of wordplay and word games, and a noticeable number of

job-seekers.

Student requests were motivated by two kinds of activity: synonym-based

homework activities and essay-writing. Some of the requests related to home-

work involved long lists of words for which synonyms—and often antonyms—

were requested. What is interesting about these is the frequent impossibility or

What we talk about when we talk about synonyms 289

at University of C

alifornia, San Francisco on May 10, 2014

http://ijl.oxfordjournals.org/D

ownloaded from

Page 12: What We Talk About When We Talk About Synonyms: (and What it Can Tell Us About Thesauruses)

irrelevance of the task, particularly with regard to antonyms. For example,

synonyms and antonyms of motif and atoll were requested. Similarly:1

(12) Okay so I’m doing this stupid Vocabulary Map thing for english home-

work and I can’t find a antonym or synonym for interjected. Can

anyone help me please? [answers.yahoo.com]

(13) okay i need a synonym for the word mistle toe..dont say PLANT!..then i

need 2 adjectives describing it! nothing like green or red and that i need

3 action words that describe it that end in (ing)

then i need a 4 word phrase describing it

haha thanks a lot! 10 pts! [sg.answers.yahoo.com]

It is not surprising that when students use thesauruses for help in essay-

writing, they seek to avoid repetition in their writing, and they want words

that will sound ‘more formal’ or ‘more sophisticated’. Those identifiable as

fiction writers also want to avoid repetition (particularly for speech-act verbs

in presenting dialogue) and to find le mot juste.

Wordplay SSs look for synonyms based on their form. These include cross-

word puzzle enthusiasts who need words that will fit a certain space (if they are

solving) or provide a particular challenge (if they are setting a clue). Other SSs

seek synonyms with particular form-based requirements, such as alliteration:

(14) i was looking for a synonym for the word ‘disappoint’ while writing my

letter to itunes. i wanted a word that began with an ‘i’ that meant

‘disappointedly’ so i could make it ‘iDisappointedly’– or something to

that effect. [presidentwishnack.blogspot.com]

(15) Since I was going to be reviewing books, I wanted to find a synonym for

‘review,’ and one that started with the letter ‘A,’ because I like alliter-

ation. [authoranalysis13.wordpress.com]

In contrast to the ten alliteration requests in the corpus, only one SS asked for a

rhyme for wordplay purposes.

The job-seekers writing cover letters and resumes are an interesting group to

consider because, unlike the other SSs considered so far, they may not be

regular writers or thesaurus users. Not having the resources (electronic or

print) that that writers, crossworders, and current students are likely to have,

they may rely more exclusively on resources that are available for free on their

computer or online, and they may have less experience in using them. They face

the challenge of presenting themselves as confident, competent, and profes-

sional, while avoiding hubris and repetitive phrasing.

Only a very small number of the SSs mention that they have looked in

a thesaurus or dictionary and not found the search word or a satisfactory

synonym. We cannot know whether the others are turning to internet

forums as a first or a last resort.

290 M. Lynne Murphy

at University of C

alifornia, San Francisco on May 10, 2014

http://ijl.oxfordjournals.org/D

ownloaded from

Page 13: What We Talk About When We Talk About Synonyms: (and What it Can Tell Us About Thesauruses)

5.2 The kinds of synonyms SSs want

Both the expressions for which the SSs seek synonyms (henceforth search

terms, STs) and the sought-after synonyms have stories to tell. While some

SSs will have turned to internet forums first (out of laziness or social motiv-

ations), for the most part we can assume that the STs considered here are ones

for which the SSs have either faced or anticipated dictionary-based problems.

In terms of thesaurus usability, we can ask: Are the expressions that inspire

synonym searches found in mainstream thesaurus resources? Are the synonyms

SSs seek found in thesauruses? And, if so, are the sought-after synonyms easily

identifiable?

In order to consider these questions, I have examined the 301 STs from

the corpus with reference to the sources in Table 3. I have limited these to

electronic sources, since we can be reasonably sure that the SSs are looking

for synonyms while at a computer. Most of these are based on — or related

to — resources that are also available in print. Two of these came with my

computer software, and with one exception the others are available free via

the internet. The free ones are not the only available on-line thesauruses,

but they were in the first page of search results when I searched for a

number of content words plus the word synonym (e.g. hope synonym)

using the Google search engine (http://google.co.uk) from a British internet

service provider and with a cleared browser history. Because it offers a very

different interface, I also considered Thinkmap’s Visual Thesaurus. Some

highly ranked sites that are not included here mirror the same source ma-

terial as sites that are included. For instance, while freedictionary.com’s

thesaurus comes very high in the search results, its material is from

Princeton WordNet (also the basis of Thinkmap Visual Thesaurus) and

Collins.2 Other than having set MS-Word to default to English-UK lan-

guage tools, I have not customized its search preferences, since it is unlikely

that the average SS will have done so. In addition to the properties

described in Table 3, all these resources allow the user to click-through to

definitions or other synonym entries.

Table 4 shows the grammatical categories of the STs. As is to be expected,

most are in the noun, verb, or adjective categories. These categories, as shown

in Table 4, include any expression that is headed by a noun, verb, or adjective,

not necessarily lexical expressions. So, for example, the VERB category includes

one-word lexical verbs like repent, phrasal verbs like end up, and entire verb

phrases like make me proud. Adjectives in Table 4 are organized with adverbs

and adverbial or adjectival prepositional phrases (e.g. at your earliest conveni-

ence) as a MODIFIER supercategory. Besides these major content categories are a

few sentences (e.g. it’s really cool; get well soon) and some grammatical and

discourse-level expressions (e.g. in my opinion, God forbid). I next discuss a

number of issues that arise in examining these synonym searches.

What we talk about when we talk about synonyms 291

at University of C

alifornia, San Francisco on May 10, 2014

http://ijl.oxfordjournals.org/D

ownloaded from

Page 14: What We Talk About When We Talk About Synonyms: (and What it Can Tell Us About Thesauruses)

Table

3:Synonym

resources

examined

Resource[abbreviation]

Accessedvia

Modality

Entrypresentation

CollinsEnglish

Thesaurus,

English

Worldwidesetting

[Collins]

collinsdictionary.com

Electronic,online;

free

Searchleadsto

headword

synonym

entry.

Macm

illanThesaurus,

American

English

setting

[Macm

illan]

Macm

illandictionary.com

Electronic,online;

free

Searchleadsto

headword

(orrun-on)

dictionary

entry.Synonyms

(‘orrelatedwords’)presentedbelow

definition,often

witha

linkto

athem

aticvocabulary

list.

Merriam-W

ebster

Online

Thesaurus

[M-W

]

merriam-w

ebster.com

Electronic,online;

free

Search

function

leadsto

headword

entry

orto

run-onentries.

Synonymsdistinguished

from

‘relatedwords’.

Microsoft

Word

Thesaurus

(Mac2011edn;English-U

K)

[MS-W

ord]

Included

withMicrosoft

Officesoftware

Electronic,offline

Searchfunctiongives

one-word

‘meanings’,

whichleadto

synonym

entries.

Oxford

AmericanWriter’s

Thesaurus

[OAWT]

Apple

Dictionary

Version

2.0.3

Electronic;included

withApple

personalcomputers

Search

function

leadsto

headword

syno-

nym

entry.

Roget’s

21st

Century

Thesaurus,

3rd

Edn

[R-TC]

thesaurus.com

Electronic,online;

free

Searchfunctionleadsto

headword

entry,if

oneexists,plusanyentriesin

whichST

occurs.

ThinkmapVisualThesaurus

[VT]

visualthesaurus.com

Electronic,online;

paid

subscription

Dynamic

spider-network

diagrams,

cen-

tring

on

ST.Colour-coded

by

part-of-

speech;sidebarand

click-through

links

todefinitions.

292 M. Lynne Murphy

at University of C

alifornia, San Francisco on May 10, 2014

http://ijl.oxfordjournals.org/D

ownloaded from

Page 15: What We Talk About When We Talk About Synonyms: (and What it Can Tell Us About Thesauruses)

5.2.1 Multiword expressions. Over 40% (127) of the expressions for which SSs

seek synonyms consist of more than one orthographic word. Fifty-four of these

are two-word constructions, forty-two are three-word constructions, and the

rest have between four and thirteen words. I use the word constructions here

purposefully: not all of these STs are phrases in the grammatical sense (which

entails grammatical constituency), for example let’s say (which was analysed as

a discourse expression) and not himself (analysed as adjectival). While many of

the constructions are lexicalized, many are not, and some are in the in-between

ground of being compositional idioms (e.g. contrary to popular belief). In some

cases, the multiword phrase might have been better spelled with a hyphen, but

this posed no difficulty when looking up the words in the electronic

Table 4: Search terms by grammatical category

Category Total Single

word

Multi

word

Major content categories 183 104 79

Noun 107 68 39

Verb 76 36 40

Modifiers 81 60 21

Adjective 74 58 16

Adverb 2 2 0

Prepositional Phrase

[adverbial/adjectival,

not discourse]

5 - 5

Closed-class / grammatical function 20 8 12

conjunctions 7 1 6

prepositions 5 1 4

pronouns 3 2 1

auxiliary verbs 2 2 0

intensifiers 2 1 1

negators 1 1 0

Extragrammatical 9 2 7

Discourse markers 7 0 7

goodbye 2 2 0

Other 8 0 8

Sentence 7 - 7

Other, non-constituent (an example of this is) 1 0 1

Total 301 174 127

What we talk about when we talk about synonyms 293

at University of C

alifornia, San Francisco on May 10, 2014

http://ijl.oxfordjournals.org/D

ownloaded from

Page 16: What We Talk About When We Talk About Synonyms: (and What it Can Tell Us About Thesauruses)

thesauruses, which all had means of leading the user to the hyphenated entry, if

one existed. (However, this does cause problems if the hyphenated expression is

a different part of speech than the multiword expression. For example, the

modal verb string would be, as in It would be great, versus adjectival would-

be in a would-be astronomer.) It is unclear whether SSs had used thesauruses

prior to asking online. While SSs might expect dictionaries —and possibly

thesauruses— to include lexicalized phrases like anal retentive and as well as,

they might know better than to look in a thesaurus for compositional phrases

like pleasantly challenged or team environment.

Of the 127 SSs who wanted synonyms for multiword STs, twenty-two expli-

citly state (within the text in the search frame) that they are looking for a single

word to replace the construction. Six more ask for synonyms with ‘fewer

words’. Those who explicitly asked for a one-word expression or ‘fewer

words’ generally were searching for unlexicalized expressions (e.g. dreaming

big, willingness to learn, feeling better than everyone else). Here the SS may

expect the resource to act as a reverse dictionary.

5.2.2 Connotation and scalarmeaning. Some phrasal STs offer insight into other

synonym needs. In the case of multiword, adjective-headed STs, half are mod-

ified adjectives, such as kinda pricey, barely alive, always right, most important,

deliberately frivolous. Such examples point out two ways in which SSs wish

synonyms to differ from the adjectives they can think of: by connotation or by

scalar position. The SS wanting a synonym for pleasantly challenged wants a

synonym for challenged with a positive connotation. The SS looking for a

synonym for kinda pricey wants a word that is at a particular point on the

scale that is described by words like cheap, pricey, and expensive. The desire for

different connotations and scalar positions is echoed in the explicit commen-

tary on other single-word and multiword STs.

Since thesaurus entries typically do not supply definitions of the words or

clearly indicate scalar positions or connotations, they rely on the SS’s famil-

iarity with the words to help them choose the best option. This, of course, can

result in the choice of inappropriate words for the context. Electronic resources

can overcome the problem of lack of information for making synonym choices,

since words in thesaurus entries can easily be linked to their dictionary entries.

In the resources I used, the thesaurus–dictionary interface seemed especially

clear in MS-Word, where selecting an item in a synonym list activates its dic-

tionary entry in a lower sub-window, so that the user can easily scan the

options (but note that Chon’s 2009 subjects either had a different version of

this tool or did not feel the definition window to be sufficient). Some of the

other electronic sources provide the definition of the ST on the same page as

the synonym listing, but none offer the synonyms’ definitions on the same

page. Instead, it is up to the SS to click through to the dictionary entry for

294 M. Lynne Murphy

at University of C

alifornia, San Francisco on May 10, 2014

http://ijl.oxfordjournals.org/D

ownloaded from

Page 17: What We Talk About When We Talk About Synonyms: (and What it Can Tell Us About Thesauruses)

each word. This puts a higher cognitive load on the SS, who has to click back

and forth between pages, keeping definitions in mind.

Other than linking to relevant definitions, the thesauruses examined do not

label synonyms according to differences in connotation. Where SSs commented

upon their need for words with different connotations, they generally did so in

terms of needing something more positive or more negative than the ST. (In my

sample, people often wanted more-positive adjectives to describe themselves

and more-negative nouns to describe others.) Grouping of positive and nega-

tive synonyms can be observed in all the thesauruses except M-W and R-TC,

which use alphabetical order. Where thesauruses do group according to con-

notation, the positive and negative tend not to be explicitly distinguished.

Consider the entry for the ST imitator from OAWT in (16):

(16) imitator

noun

1 she has many imitators copier, emulator, follower, mimic, plagiarist,

ape, parrot; informal copycat.

There is a discernible neutral-positive-negative ordering here, but the only item

that is set apart from the list is distinguished by its register, not its connotation.

We can examine visual presentation further with reference to scalar position,

considering the ST pricey. Collins and MS-Word have what appears to be

scalar ordering for pricey, but no indication that there is a qualitative difference

between costly or dear and exorbitant or extortionate. OAWT in this case

groups words with the most similar meanings within semicolons, allowing

the SS to surmise that over the odds and exorbitant are closer in meaning

than exorbitant and costly are. Macmillan has an indiscernible ordering for

the synonyms on the pricey page, but clicking a ‘more’ button leads to a

page of related words that are grouped by sense similarity.

One might think that the visual mapping of VT would provide a natural way

to group synonyms according to connotations and scalar relations, but this is

not clearly realized, largely because it is the victim of its source material,

WordNet. WordNet was initially conceptualized as a model of the mental

lexicon (Miller 1990), and it is shaped by its founders’ theoretical assumptions

(for more extensive discussion, see Sampson 2000, Murphy 2003). WordNet

consists of ‘synsets’, groups of words with similar meanings, which are orga-

nized by other semantic relations within part-of-speech sublexicons. The archi-

tectures of the noun and verb lexicons are based on inclusion relations, whereas

the primary relation in the adjective lexicon is antonymy. Because the inclusion

and antonymy relations hold between synsets, not between words, WordNet

does not work exactly like a thesaurus. In the case of adjectives, for example,

synsets are organized such that they are represented by ‘focal’ adjectives, which

are presented as having direct antonyms, and all other adjectives in the

What we talk about when we talk about synonyms 295

at University of C

alifornia, San Francisco on May 10, 2014

http://ijl.oxfordjournals.org/D

ownloaded from

Page 18: What We Talk About When We Talk About Synonyms: (and What it Can Tell Us About Thesauruses)

semantic field belong to synsets that (directly or indirectly) link to the focal

antonyms. Pricey belongs to the field for which expensive and cheap are the

focal adjectives. Figure 1 shows the result of a VT search for pricey.

VT shows the synset for pricey and thus offers just three synonyms and one

spelling variant for pricey, versus the thirty-nine synonyms that R-TC offers on

its one-page results (under the headwords costly and exorbitant). If one clicks

on the dot in the centre of the Figure 1 diagram, a link to expensive is shown. If

one clicks on that, the animated display settles down to Figure 2. (This is not

the same output as one would get by searching for expensive through the search

tool. That result shows the antonym relation to cheap.) While the display is

visual, it is not immediately obvious how items are related — why, for instance,

dearly-won is closer to overpriced than to dear. Items with different scalar

values are somewhat grouped together, but the ordering does not follow the

left-to-right ascension found in OAWT and Collins. Left-to-right ascension is

intuitive to readers of English, who are used to reading ascending numbers in

that direction. While VT also has the opportunity of vertical arrangement of

synonyms, higher prices are not necessarily correlated with higher position in

the diagram. Instead, adjectives indicating more expense are found to the left of

those indicating lower cost, and some of the ‘more expensive’ meanings are

lower than others. So, while there is certainly potential for scalar meaning

relations to be represented visually, this is not part of what VT does.

5.2.3 Register. Editors and educators (at least at the university level) roll their

eyes at obvious uses of thesauruses to sound more ‘sophisticated’ or ‘academic’,

and thus frequently advise students and writers that thesauruses should be

avoided. (For example, editor John McIntyre (2013) warns, ‘A reporter with

a thesaurus is like a toddler with a handgun’. See Zimmer 2012 for some history

of the ‘cliche’ of advising against thesaurus use.) But looking at the STs for

which SSs sought more formal or sophisticated expressions, it is hard to fault

most of them for wanting a more formal word, since many of these STs are

very informal. For example, the Americanisms you guys and awesome (in the

‘fantastic’ sense) and the gaming-slang noun try-hard (a ‘poser’) would not be

acceptable in formal writing. Others would be considered vulgar in most con-

texts, for example freaking (as a euphemism for another f-participle), balls

(in the ‘chutzpah’ sense), dirtball, and to bullshit.

I looked for six of these STs in their slang senses in the seven thesauruses.

(You guys is discussed in the next section.) Of the forty-two possible sense-

thesaurus combinations, nine entries were found, as summarized in Table 5.

As Table 5 shows, two of the STs are found in none of the thesauruses, and

two of the thesauruses (the two that were bundled with software packages)

have none of these slang STs. While American slang might be inappropriate for

the British thesauruses, that does not seem to be a major factor here, since the

296 M. Lynne Murphy

at University of C

alifornia, San Francisco on May 10, 2014

http://ijl.oxfordjournals.org/D

ownloaded from

Page 19: What We Talk About When We Talk About Synonyms: (and What it Can Tell Us About Thesauruses)

Figure 2: Pricey> expensive> expensive in Visual Thesaurus

Table 5: Coverage of six slang STs in seven thesauruses

Collins Macmillan M-W MS-Word OAWT R-TC VT

awesome – S R – – – –

balls ˇ ˇ – – – ˇ –

bullshit, v. – – – – – – –

dirtball – – – – – S M

freaking – ˇ R – – – –

try-hard – – – – – – –

ˇ =offers more formal synonyms; S=offers other slang synonyms;R=slang synonyms supplemented by non-slang ‘related words’ listing;M=metaphorical synonyms only; –=no entry for the relevant sense.

Figure 1: Pricey in Visual Thesaurus

What we talk about when we talk about synonyms 297

at University of C

alifornia, San Francisco on May 10, 2014

http://ijl.oxfordjournals.org/D

ownloaded from

Page 20: What We Talk About When We Talk About Synonyms: (and What it Can Tell Us About Thesauruses)

American resources are no better than the British ones in including these ex-

pressions. In three of the nine entries found, the slang ST is matched only or

overwhelmingly with other slang terms. For example, the Macmillan entry for

awesome (in its slang sense) links to a list ofWords used mainly by young people.

Not all of these are synonyms for awesome, but some are, and all of them

would be as unsuitable for formal writing as awesome is. In R-TC dirtball is

found in the entry for sleazebag, which includes some non-slang alternatives,

but the informality of the synonyms is only sometimes noted, as shown in (17).

It seems the asterisk marking informality is only applied to informal expres-

sions for which there is a hyperlink to another entry.

(17) Main Entry: sleazebag [sleez-bag] Show IPA

Part of Speech: noun

Definition: creep

Synonyms: crud, degenerate, deviant, dip, dirtbag, dirtball, per-

vert, pig*, scum*, scumbag, scuzzbag, sleaze, sleaze-

ball, slimebag, slimeball, slimebucket, slob, weirdo*

Of the slang STs, VT has only dirtball, and it is only linked to metaphorical

alternatives (insect, louse, worm). Again, it is questionable whether these alter-

natives would be appropriate to professional writing, where a euphemism

might be more appropriate.

While matching synonyms in terms of their slang status is appropriate for the

lexicological task of identifying ‘perfect’ synonyms, it is contrary to the needs

of most thesaurus users who want to find words that they can use when writing

schoolwork or professional documents. M-W provides a useful compromise

between these positions, although it only includes two of the STs. Only infor-

mal synonyms are offered under the heading ‘synonyms’, but directly below the

synonyms are ‘related words’, which list more job-application-friendly equiva-

lents. This is illustrated for freaking in (18).

(18) freaking

adjective

deserving of one’s condemnation or displeasure <this freaking assign-

ment is such a pain in the neck>

Synonyms accursed (or accurst), blasted, confounded, cotton-picking,

cursed (also curst), cussed, dang, danged, darn (also durn), darned

(also durned), deuced, doggone (or doggoned), freaking, goddamned

(or goddamn or goddam), infernal

Related Words atrocious, awful, bum, detestable, execrable, lousy, punk,

rotten, terrible, wretched; abominable, odious, vile; contemptible, despic-

able, miserable, nasty, pitiable, pitiful, scabby, scummy, scurvy, shame-

ful, sorry [. . .]

298 M. Lynne Murphy

at University of C

alifornia, San Francisco on May 10, 2014

http://ijl.oxfordjournals.org/D

ownloaded from

Page 21: What We Talk About When We Talk About Synonyms: (and What it Can Tell Us About Thesauruses)

5.2.4 Function words. About 10% of the STs do not belong to content cate-

gories. These are either closed-class expressions, such as pronouns and con-

junctions, longer paraphrases of those expressions, or expressions that interact

with sentences at the discourse rather than sentence level. While this is a small

proportion of the STs found, it is large if one considers that only about 1% of

an English speaker’s vocabulary is closed-class function words (Cutler 1993),

yet the single-word non-content subset is still nearly 7% of the data here. The

function words are interesting from a lexicographical perspective, in that the-

sauruses are semantically organized, yet these words have very little semantic

content.

Though functional expressions are semantically impoverished, SSs look for

synonyms for them for the same reasons that they embark on other synonym

searches: to avoid repetition, to sound more formal, or to put the same mean-

ing into fewer words. In the cases of the auxiliary verbs (was, would), the SSs,

who reported wanting to avoid repetition, might be better advised to rethink

sentence structure rather than word choice. But the desire for replacements for

STs like as well as, at least, and above and beyond is understandable to any

writer who has found such phrases popping up too often (or taking up too

much space) in their writing.

The twenty function-expression STs included nineteen unique STs (as well as

occurs twice). The best resource for function words (with thirteen of the nine-

teen) is R-TC. MS-Word and Macmillan do well with eleven and ten, respect-

ively. The rest include between four and six of these expressions. VT is the

weakest of these: although it includes five of the function expressions, it offers

very few synonyms for them. For example, the only result shown for never

is ne’er.

The intensifier very is the only functional expression for which synonyms are

given in all the thesauruses, offering between three (VT) and more than forty

(R-TC) synonyms. Three STs, including the two least lexicalized of the group,

were found in no thesaurus: just like, as seen in, and you guys. My, we, whether

or not, would, and in regards to were each found in just one. (Although two

others included in regard to, there was no link between that entry and the less

standard form.) Would leads to the entry for will in two thesauruses (M-W,

R-TC), but in both cases, the synonym entry for will relates to the lexical verb

(as in He wants to will his money to a cat), not the modal verb (I will go); thus

the result was confusingly irrelevant to the search for would.

For categories like conjunctions and intensifiers, English offers a lot of op-

tions, and some of them are more formal than others; for example, a writer

who feels she is using too many ands might want to be reminded of as well as, in

addition to, plus, moreover, as a consequence, and so forth. So although nouns,

verbs, and adjectives are the usual stuff of thesauruses, this finding calls for

more attention to the ‘non-content’ categories.

What we talk about when we talk about synonyms 299

at University of C

alifornia, San Francisco on May 10, 2014

http://ijl.oxfordjournals.org/D

ownloaded from

Page 22: What We Talk About When We Talk About Synonyms: (and What it Can Tell Us About Thesauruses)

5.3 Are dictionarydefinitions of synonymright?

At the start of this article, I examined uses of synonym in more general corpora

and pointed out ways in which use of the term by laypeople differs from the

definitions offered in standard dictionaries. In analysing the data from web

discussions of synonym searches, I have considered more specifically uses of

the word synonym by people who want to find synonyms. So, the final question

to ask is: are the synonymy relations sought by SSs the same as those repre-

sented in dictionary definitions, or do we see more evidence of ‘looser’ inter-

pretations of the term? Specifically, do these 301 contexts tell us anything about

whether synonymy is perceived as being a same-language phenomenon or not

and whether it is an intensional or extensional relation?

There was little evidence in the web data of cross-linguistic uses of synonym.

This may be an artefact of the search phrases: if a SS were inclined to call

translational equivalents synonyms, they are probably less likely to say I need a

synonym for X in French, than I need a French synonym for X, which would not

be found on a search for I need a synonym. Three items in the data involve

cross-dialectal synonyms between British and American English, as in (19).

(One of these, it turned out, was from my own blog.)

(19) As a Brit, I was too busy trying to find a synonym for sou’wester

(rexwordpuzzle.blogspot.com)

One could take such examples as evidence of a relaxation of the ‘same lan-

guage’ criterion, since the SS is looking for (an equivalent for) a word that is

not native to their own ‘mother tongue’. Then again, this might be evidence

that dialectal differences are considered much the same as registral differences

— forms of one’s own language that one does not use oneself.

On the intension/extension question, there are no examples of anyone asking

for a synonym that refers to the same thing but does not necessarily have the

same meaning. The most suspicious cases in this regard are two in which the SS

wants a synonym for a proper name, for example Canada. If we accept that

proper names do not have senses, then the similarity between Canada and any

possible synonym will be one of reference, rather than sense. Still, if the syno-

nym for Canada is a nickname for the country that is not used to refer to

anything else (e.g. The Great White North), only a semanticist with a strong

philosophical bent would argue that they do not mean the same thing (because

they do not ‘have meanings’). When SSs ask for help with synonyms, the help

they are offered is often from a thesaurus, and thus follows the expectation that

synonyms should be related by sense first, reference second. It might be inter-

esting to investigate whether (lay)people make a distinction between inten-

sional and extensional similarity by using other terminology. For example,

might I need a synonym for X request an answer with intensional similarity

300 M. Lynne Murphy

at University of C

alifornia, San Francisco on May 10, 2014

http://ijl.oxfordjournals.org/D

ownloaded from

Page 23: What We Talk About When We Talk About Synonyms: (and What it Can Tell Us About Thesauruses)

to X, and might I need another word for X indicate that intensional similarity is

not as important?

We can still question whether the same-language and similar-sense criteria of

the dictionary definitions are necessary conditions on synonymy or just proto-

typical ones. Nevertheless, the evidence here is that when people look for

synonyms, they are using the word as lexicographers do.

6. Conclusions

The rise of the knowledge economy has made more people responsible for

writing — particularly for writing formal prose (see Starke-Meyerring et al.

2011). At the same time, ease of access to thesauruses has increased dramatic-

ally. Thus it is reasonable to assume that more people than ever are using

thesauruses. This makes now a prime time for lexicographers to re-consider

thesauruses and other synonym-finding aids, in much the way that dictionary

usability has been attended to in recent decades.

This article has investigated what people say about synonyms and their

synonym needs with a view to how thesauruses might better meet those

needs. While it was found that synonym is sometimes used more loosely than

it is represented in dictionaries, the evidence found in web discussions of syno-

nym searches shows laypeople using the word synonym according to the

common definition that involves intensional similarity among expressions

within a language.

The web-based discussions of synonym searches offers insights into why

people want to find synonyms and the range of expressions for which syno-

nyms are sought. I have concentrated on aspects of those discoveries that

challenge the traditional notion of what a thesaurus or synonym dictionary

includes. Multiword expressions, including unlexicalized expressions, form a

significant proportion of the synonym searches. Examining these in the adjec-

tive category demonstrated users’ desires for synonyms that differ from the

search term in connotation or scalar position. Function expressions were

sought by synonym-seekers, but sometimes fall outside the semantic categories

devised by Roget (1852), which continue to influence modern thesauruses.

They also fall outside the remit of many relational databases, like WordNet,

and ontologies developed for Natural Language Processing. Extremely infor-

mal expressions were also noted among the synonym-search terms, and with

them a unanimous desire by writers to find more formal or euphemistic

alternatives.

Semanticists (myself included) tend to talk about ‘best’ synonyms in terms of

their substitutability in a particular context without change in denotative, con-

notative, or social meaning (see Murphy 2003, chapter 4). Perfect synonyms in

terms of substitutability rarely exist, and they are also not what most people

want when they seek a synonym while writing. Of the SSs who stated their

What we talk about when we talk about synonyms 301

at University of C

alifornia, San Francisco on May 10, 2014

http://ijl.oxfordjournals.org/D

ownloaded from

Page 24: What We Talk About When We Talk About Synonyms: (and What it Can Tell Us About Thesauruses)

reasons for looking for synonyms, only 13% wanted to avoid repetition and

about 5% were looking (perhaps in vain) for an expression with the same

meaning but different formal properties (e.g. for alliteration). The rest

wanted a ‘synonymous’ expression that differed from the search term in de-

notation, connotation, or register. As subjects in Chon’s (2009) study reported,

most existing tools are not designed to serve this need.

The shift away from print-based resources means that we can be very im-

aginative in how synonym information is found and presented. However, most

online resources have not moved much from the style of print dictionaries;

automatic searching and clickable links to definitions are sometimes the only

concessions to the electronic form. Visual Thesaurus offers the most imagina-

tive display, with mouse-over definitions and dynamic, graphic presentation of

word similarity. But it is limited both by its source material (Princeton

WordNet) and the automatic nature of the visual displays. While VT represents

closeness of meaning iconically, there is much more potential for the use of

vertical and horizontal spatial relationships to aid understanding of the vo-

cabulary presented.

So it is an exciting time to think about thesauruses (or relational information

in dictionaries) and how they might better serve writers of all types. Most

electronic thesauruses interact with a dictionary at some level, so a first,

simple suggestion is that any expression for which a dictionary definition

exists should also have synonym information. Thesaurus entries should also

be reviewed to ensure that they offer synonyms that differ in register and

connotation and that those differences are clearly marked.

But any reconsideration of thesaurus form should be accompanied by other

types of user (and usability) research. The unconventional web-corpus method

used here offers a look at expressions that people want synonyms for, whether

or not they have used a thesaurus to try to find them. This essay’s reflections on

the usability of the thesauruses for finding those expressions are mine alone.

While it would be beneficial to have the typical user studies that call on stu-

dents who are accustomed to writing essays in their first or second language, it

is also important to consider users who are writing other kinds of documents,

since the kinds of words that might impress a professor in the humanities are

not necessarily the same as those that would impress, say, a professor in the

sciences, stockholders reading a quarterly report, or employers reading job

applications.

Acknowledgements

My thanks to Fredric Dolezal, special issue editor; Eric Kow for the software

and documentation; the Research Committee of the School of English at the

University of Sussex, which funded that software; and Andy Holyer for clari-

fication regarding computing senses of synonym.

302 M. Lynne Murphy

at University of C

alifornia, San Francisco on May 10, 2014

http://ijl.oxfordjournals.org/D

ownloaded from

Page 25: What We Talk About When We Talk About Synonyms: (and What it Can Tell Us About Thesauruses)

Notes

1 I have not provided full URLs for the examples here in order to save space. Theexamples will remain searchable on the web for as long as they are maintained there.

Interested readers are welcome to request the example database from the author.

2 Not all thesaurus.com entries are marked for source, but most are from Roget’s 21st

Century Thesaurus. Occasional others are marked as ‘Concept Thesaurus, diction-

ary.com’. The items discussed in detail here are from Roget’s. Where searches gave

multiple pages of results, only the first page was examined.Date of last access for all electronic resources is 13 February 2013, unless otherwise

noted in References.

ReferencesA. Dictionaries and corpora

American Heritage Dictionary, 5th edn. 2011. Boston: Houghton-Mifflin. [AHD5]

Collins English Dictionary 2010. Glasgow: HarperCollins. [CED]

Collins English Thesaurus [online]. 2013. Glasgow: HarperCollins. http://www.collinsdic

tionary.com/english-thesaurus [Collins]

Davies, Mark 2004–. BYU-BNC: The British National Corpus. Available online at

http://corpus.byu.edu/bnc. [BNC; searched in June 2012]

Davies, Mark 2008–. The Corpus of Contemporary American English. Available at http://

www.americancorpus.org. [COCA; searched in June 2012]

EnTenTen08 Corpus 2008. Available via Sketch Engine http://sketchengine.co.uk

[Searched in June 2012]

Macmillan Thesaurus [online] 2009–2013. http://www.macmillandictionary.com/about_thesaurus.html [Macmillan]

Merriam-Webster Online Dictionary 2013. Springfield, MA: Merriam-Webster.http://www.merriam-webster.com/ [MWOD]

Merriam-Webster Online Thesaurus 2013. Springfield, MA: Merriam-Webster. http://

www.merriam-webster.com/ [M-W]

Microsoft Word for Macintosh 2011. Seattle Microsoft Corporation. [MS-Word]

Oxford Dictionary of English, 3rd edn. 2010. Oxford: Oxford University Press. [ODE]

Princeton University 2010. ‘About WordNet’. WordNet: Princeton University.

Random House Dictionary 2013. http://dictionary.com. [RHD]

Roget’s 21st Century Thesaurus, 3rd edn. 2013. The Philip Lief Group. http://thesaurus.com. [R-TC]

Thinkmap 1998–2013. Visual Thesaurus. http://www.visualthesaurus.com. [VT]

B. Other literature

Battenburg, J. D. 1991. English Monolingual Learner’s Dictionaries: A User Oriented

Study. (Lexicographica Series Maior 39.) Tubingen: Niemeyer.

Bejoint, H. 1981. ‘The Foreign Student’s Use of Monolingual English Dictionaries:A Study of Language Needs and Reference Skills’. Applied Linguistics, 2: 207–22.

Chon, Y. V. 2009. ‘The Electronic Dictionary for Writing: A Solution or a Problem?’.

International Journal of Lexicography, 22.1: 23–54.

Cruse, D. A. 1986. Lexical Semantics. Cambridge: Cambridge University Press.

Cutler, A. 1993. ‘Phonological Cues to Open- and Closed-Class Words in the Processing

of Spoken Sentences’. Journal of Psycholinguistic Research, 22.2: 109–31.

What we talk about when we talk about synonyms 303

at University of C

alifornia, San Francisco on May 10, 2014

http://ijl.oxfordjournals.org/D

ownloaded from

Page 26: What We Talk About When We Talk About Synonyms: (and What it Can Tell Us About Thesauruses)

de Schryver, G.-M. and D. Joffe. 2004. ‘On How Electronic Dictionaries are Really

Used’. Proceedings of EURALEX 2004, Lorient, France, 187–96.

Dziemianko, A. 2010. ‘Paper or electronic? The role of dictionary form in language

reception, production and the retention of meaning and collocations’. International

Journal of Lexicography, 23.3: 257–73.

Kilgarriff, A., P. Rychly, P. Smrz and D. Tugwell. 2004. ‘The Sketch Engine’.

Proceedings of EURALEX 2004, Lorient, France, 105–16.

Kipfer, B. A. 1985. ‘Dictionaries and the Intermediate Student: Communicative Needs

and the Development of User Reference Skills’. MA Dissertation, University of

Exeter.

Kipfer, B. A. 1987. ‘Dictionaries and the Intermediate Student: Communicative Needs

and the Development of User Reference Skills’. In Anthony P. Cowie (ed.),

The Dictionary and the Language Learner. Tubingen: Niemeyer, 44–54.

McIntyre, J. E. 2013. The Old Editor Says: Maxims for Writing and Editing. Baltimore:

Apprentice House.

Miller, G. A. (ed.) 1990. WordNet: An On-line Lexical Database. Special issue of

International Journal of Lexicography 3.4.

Murphy, M. L. 2003. Semantic Relations and the Lexicon. Cambridge: Cambridge

University Press.

Roget, P. M. 1852. Roget’s Thesaurus of English Words and Phrases. Reprinted 1998.

London: Longman.

Sampson, G. 2000. Review of C. Fellbaum (ed.), WordNet: an electronic lexical data-

base. International Journal of Lexicography 13.1: 54–59.

Siegel, M. E. A. 2007. ‘What Do You Do with a Dictionary: A Study of Undergraduate

Dictionary Use’. Dictionaries, 28: 23–47.

Starke-Meyerring, D., A. Pare, N. Artemeva, M. Horne and L. Yousoubova (eds.) 2011.

Writing in Knowledge Societies. Fort Collins, CO: The WAC Clearinghouse. http://

wac.colostate.edu/books/winks/.

Zimmer, B. 2012. ‘Word for word’. Lapham’s Quarterly (Spring). Available at: http://

www.laphamsquarterly.org/reconsiderations/word-for-word.php.

Zock, M., O. Ferret and D. Schwab. 2010. ‘Deliberate word Access: An Intuition, a

Roadmap and Some Preliminary Empirical Results’. International Journal of Speech

Technology, 13: 201–18.

304 M. Lynne Murphy

at University of C

alifornia, San Francisco on May 10, 2014

http://ijl.oxfordjournals.org/D

ownloaded from