Upload
khangminh22
View
0
Download
0
Embed Size (px)
Citation preview
Toronto Working Papers in Linguistics (TWPL), Volume 40
© 2018 Heather Yawney
On the right to be faithful: The irregular stress of Turkish
adverbials with -En Heather Lynn Yawney University of Toronto The stress pattern of Turkish adverbials with -En is an
interesting case of irregular stress that is understudied. Sezer
(1983) and Erdal (2009) offer brief descriptions of the stress
pattern and Inkelas (1999) acknowledges that the pattern exists.
Özçelik’s (2014) single grammar approach uses a single
constraint ranking under an Optimality Theory framework.
Turkish adverbials with -En posed some difficulty under his
pre-specified foot proposal, which accounts for several
surfacing Turkish irregular stress patterns. Following Idsardi’s
(2009) proposal that a single foot boundary is sufficient to
define a foot, a single constraint ranking that involves
faithfulness to a pre-specified right foot edge captures a unified
stress system that can extend to include Turkish adverbials with
-En.
1 Introduction
There is an interesting case of non-final main stress in Turkish which involves derived adverbs with
the -en/-an suffix.1 These adverbials are formed from Arabic nouns with the Arabic suffix -En and they
reject the regular main stress assignment rule which places main stress on the final syllable of the word.
This stress pattern presents a unique challenge when attempting to account for the irregular stress pattern
of Turkish adverbials since very little work has been done on this specific stress pattern in Turkish. Only
Sezer (1983) and Erdal (2009) provide simple and brief descriptions of the stress pattern. Additionally,
Inkelas (1999) recognizes the irregular stress pattern but chooses not to include it in her analysis because
the adverbial construction is not productive and the data is not consistent across speakers.2
I begin by illustrating the stress pattern with the following data. The stressed syllable is marked in
bolded type with an acute accent.
I would like to thank Peter Jurgec, the audiences at the CRC-Sponsored Summer Phonetics/Phonology Workshop
(2016) and the audiences at MOT (Montreal-Ottawa-Toronto) (2017) for the valuable feedback. I would also like to
thank an anonymous reviewer for their very helpful feedback. 1 Turkish is a language rich with vowel harmony across morphemes. The realisation of the -En suffix is dependent on
the frontness of the preceding vowel of the word. (E= {e [-back], a [+back]}). 2 An anonymous reviewer brought up an interesting question about why Turkish adverbials with -En should be
included in an analysis if Inkelas (1998) argues that its stress pattern lacks productivity and consistency. Along with
the regular and irregular stress patterns, including pre-stressing suffixes, stressed suffixes and stressed roots,
discussed in this paper, the stress pattern of Turkish adverbials with -En is another irregular stress pattern that
surfaces in Turkish. No matter the extent of its occurrence, one should be able to account for the system in relation
to the stress patterns already thoroughly discussed and analysed in Turkish. Turkish adverbials with -En is one of
the understudied stress patterns in Turkish and deserves more attention. One should not ignore that a lesser known
pattern exists.
HEATHER LYNN YAWNEY
2
(1) Common bisyllabic Turkish adverbials with -En3
a. bazen ‘sometime’ h. manen ‘inwardly’
b. ʤidden ‘really’ i. naklen ‘by transfer’
c. halen ‘currently’ j. nakten ‘in cash’
d. hykmen ‘forfeit’ k. ɾesmen ‘officially’
e. kasten ‘deliberately’ l. ʃahsen ‘personally’
f. kɯsmen ‘partially’ m. seklen ‘formwise’
g. lytfen ‘please’ n. zaten ‘already’
(2) Common polysyllabic Turkish adverbials with -En
a. aʤilen ‘urgently’ i. myʃtereken ‘mutually’
b. nadiɾen ‘rarely’ j. mytemadijen ‘on end’
c. nisbeten ‘proportionally’ k. tekeffylen ‘by surety’
d. emaneten ‘safekeeping’ l. tesadyfen ‘by chance’
e. mynhasɯɾan ‘specially’ m. iktisa:den ‘economically’
f. muhtemelen ‘probably’ n. istina:en ‘based on’
g. muntazaman ‘regularly’ o. kaza:en ‘accidentally’
h. muvakkaten ‘temporarily’ p. esa:sen ‘basically’
(Erdal, 2009: 20; Sezer, 1983: 62–63)
Numerous adverbials have initial stress when they are bisyllabic (1). However, in polysyllabic words, stress
is moved one syllable to the left of where the base had it. Stress is retracted to the antepenultimate (2a–l),
unless the penultimate syllable has a long vowel (2m–p).
Sezer (1983) and Erdal (2009) vary slightly in their descriptions of the irregular stress pattern of the
Turkish adverbials with -En. On one hand, Sezer (1983) refers to the strong and weak syllable distinction
when discussing the stress pattern. A strong syllable is one of (C)VC, (C)VCC or (C)V, and a weak syllable
is one ending with a short vowel, (C)V. With derived adverbials with -En, main stress falls on the
penultimate syllable if it is strong or on the antepenultimate syllable if the penultimate syllable is not strong.
Inkelas (1999) follows Sezer (1983) and adds that this pattern can be analysed as a moraic trochee with an
extrametrical final syllable, as observed by Hayes (1995). On the other hand, Erdal (2009) observes that
vowel length determines where main stress falls. The stress of the adverbials with -En falls on the
antepenultimate syllable unless the penultimate has a long vowel.
Özçelik (2014) offers an account of several other stress patterns (regular and other irregular stress
patterns) in Turkish that captures irregular stress as involving faithfulness to the right edge of a pre-specified
foot (either one or two syllables in length) in the underlying representation using an Optimality Theory
framework. He argues for (the edges of) a foot being pre-specified in the input and not stress or a stressed
syllable itself. This foot does not need to be well-formed or be the one that surfaces, in accordance with
Prince & Smolensky’s (1993) Richness of the Base. Özçelik’s (2014) single grammar account unifies
regular stress and irregular stress, including pre-stressing suffixes, stressed suffixes and stressed roots, using
a single constraint ranking.4 The only difference between regular and irregular stress is that the former does
not involve foot structure and the latter is footed in the input, all while being subject to the same constraints.
3 The in-text abbreviations found throughout this paper are defined as followed: 1/2/3 = first/second/third person, ADV
= adverb, CONN = connector, EP.COP = epistemic copular, NEG = negative, PL = plural, PRS = present progressive,
PST = past, QUES = yes/no question marker, RECIP = reciprocal, SG = singular. 4 Özçelik (2014) does not treat final (regular word-final stress) and exceptional (irregular non-word-final stress) as
different systems belonging to different co-phonologies.
ON THE RIGHT TO BE FAITHFUL: THE IRREGULAR STRESS OF TURKISH ADVERBIALS WITH -EN
3
No one has attempted to provide a constraint-based analysis of the irregular stress pattern of Turkish
adverbials with -En. Therefore, the goal of this paper is to provide an analysis using Optimality Theory to
account for the stress pattern of Turkish adverbials with -En. Applying an aspect from Idsardi (2009)
whereby a single left or right boundary is sufficient to define a grouping5,6, and promoting the faithfulness
to the right edge of the foot constraint to a higher ranking from Özçelik’s (2014) original analysis, I propose
that irregular stress patterns involving a pre-specified right edge foot bracket will maintain a single grammar
approach.7 I assume a quantity sensitive system (c.f. Hayes, 1995) whereby only long vowels influence the
weight of the syllable and, by extension, where stress falls. This is contra some previous proposals (c.f.
Inkelas, 1999; Sezer, 1983) that favour coda consonants also contributing to syllable weight. Since words
of Turkish origin do not contain long vowels, vowel weight is the main factor to consider because it is only
long vowels blocking the stress retraction when forming Turkish adverbials from Arabic nouns and using
the Arabic -En suffix. Özçelik (2014) also makes it very clear that syllable weight in terms of syllables with
coda consonants do not affect where main stress falls in instances of other irregular stress patterns.
Alongside syllable weight, I argue for a moraic trochee surfacing in the input after generating the
foot. Turkish adverbials with -En have stress falling on the syllable preceding the -En suffix. Stress falls on
the antepultimate syllable unless the penultimate contains a long vowel. The other irregular stress patterns
also accounted for in this paper never have stress falling on the final syllable or on a syllable following the
presence of one of the irregularly patterned morphemes (suffixes and roots). First, the pre-stressing suffixes
have stress falling on the preceding syllable and are never post-stressing (§2.2.1). Second, the bisyllabic
stressed suffixes always have stress falling on the first syllable and never on the second (§2.2.2). Third,
stressed roots are variable in terms of whether stress falls on the penultimate or the antepenultimate syllable
but stress is never on the final syllable (§2.2.3). In addition, when two different irregular stress patterns
interact, the leftmost stress wins as the main stress surfacing on the word.8
This paper will be structured as follows. Section 2 offers a description of the different stress patterns
to be discussed. Since I am adding to a previous proposal using a constraint-based system, it is important
to also account for the regular and the other irregular stress patterns found in Turkish. Section 3 provides
an analysis, within the Optimality Theory framework, which uses a trochaic foot for the entire grammar.
Section 4 will summarise the important points this paper puts forth to account for Turkish adverbials with
-En.
2 Turkish stress patterns
Turkish has been discussed in the literature for a long time. Different parts of the system have been
analysed but no one has analysed the full system. In this section, I will review the regular stress pattern
(final main stress) and several other instances of irregular stress patterns, which include pre-stressing
suffixes, stressed suffixes, stressed roots and adverbials with -En. The domain with which this paper is
concerned is the word level.9 It is necessary to introduce these other stress patterns to support a unified
stress analysis that extends to include Turkish adverbials with -En, the primary goal of this paper.
5 After parsing syllables, a grouping is a built foot. See Idsardi (2009) for further details. 6 Idsardi (2009) categorizes Turkish in two possible groups of languages that pattern with a left edge bracket specified
preceding either the ultimate or the penultimate syllable. He includes a question mark beside Turkish under the
possible languages of each group, thus demonstrating uncertainty where to categorize Turkish. However, this paper
proposes that Turkish patterns with languages with a right edge bracket. 7 A single grammar approach involves a single constraint ranking that can account for several different surface patterns. 8 This will not be discussed further as it is beyond the scope of this paper. 9 Other Turkish stress phenomena, such as compound stress, derived diminutive adjective stress, vocative stress and
secondary stress, exist but will not be dealt with in this paper. Including all patterns is a very ambitious project and
not within the scope of this paper. The goal is to take one step closer to a unified Turkish stress analysis by showing
that an understudied stress pattern, the Turkish adverbials with -En, may be included with the stress patterns that
have already been thoroughly discussed.
HEATHER LYNN YAWNEY
4
2.1 Regular main stress
Turkish has a very simple stress assignment rule which assigns main stress on the final syllable
irrespective of the length of the word, the weight of the syllable or other rhythmic considerations (see Lees,
1961; Lewis, 1967; Underhill, 1976; Sezer, 1983; van der Hulst & van de Weijer, 1991; and many others).
Consider the following example. The stressed syllable is marked in bolded type and stress is marked with
an acute accent.
(3) Turkish regular stress
a. kitáp ‘book’
b. kitaplɯk ‘bookcase’
c. kitaplɯklaɾ ‘bookcases’
d. kitaplɯklaɾɯm ‘my bookcases’
e. kitaplɯklaɾɯmɯz ‘our bookcases’
f. kitaplɯklaɾɯmɯzdán ‘from our bookcases’
(Kabak & Vogel, 2001: 316)
With a series of suffixes attached to the stem, stress moves rightward each time a new suffix is added. This
tendency is spread through word classes, whether such forms are simple or derived, native or borrowed.
The right edge will always bear main stress and this pattern is not disturbed by a large number of nominals
and verbal suffixes (Kamali, 2011).
Studies in Turkish phonology use the cover term ‘stress’ to describe the shared perceptually salient
phenomenon of prominence and the fundamental frequency-related pitch accent. According to Kamali
(2011), Turkish only uses pitch as the most reliable cue to signal prominence within a word. This has been
observed as early as Underhill (1976). Levi (2005) ascribes Turkish to the class of pitch accent languages.
However, languages differ with their systematic use of other acoustic means, such as intensity and
tone/pitch.
2.2 Irregular main stress
In addition to the regular stress assignment pattern, there are various other stress assignment patterns
that demonstrate instances of non-final stress. These patterns are considered irregular. These irregular
patterns override final stress with main stress falling on a syllable other than word-final. As mentioned
above, I will introduce pre-stressing suffixes, stressed suffixes, stressed roots and finally, adverbials with -
En.
2.2.1 Pre-stressing suffixes
A pre-stressing effect arises when certain suffixes are attached to a word. Consider the following list
of pre-stressing suffixes.
(4) Pre-stressing suffixes10
a. -mE negative
b. -CA, -leyin, -In adverb/adjective-deriving suffixes
c. -de post-clitic coordinator
10 Turkish is a language rich with vowel harmony across morphemes. There are two types of harmony. First, frontness
harmony is represented by E ([e] = [-back], [a] = [+back]). Second, frontness and roundedness is represented by I
([i] = [-back, -round], [y] = [-back, +round], [ɯ] = [+back, -round], [u] = [+back, +round]}). Turkish also has
consonant voicing assimilation. There are two different alternations in this table, C ([ʤ] = [+voice], [ʧ] = [-voice])
and D ([d] = [+voice], [t] = [-voice]).
ON THE RIGHT TO BE FAITHFUL: THE IRREGULAR STRESS OF TURKISH ADVERBIALS WITH -EN
5
d. -ki complementiser
e. -DIr epistemic copula
f. -j/-Ø copular clitic (full form: i)
g. -mI yes/no question clitic
h. -(j)le commutative/instrumental (full form: ile)
i. -(j)ken ‘when’ (full form: iken)
j. -gil suffixes that derive family names from nouns
(modified from Kabak & Vogel, 2001: 328)
Main stress is shifted to the syllable immediately preceding a pre-stressing suffix irrespective of
rhyme structure (Kabak & Vogel, 2001; Özçelik, 2014). To illustrate pre-stressing suffixes in action,
consider the following examples. The stressed syllable is marked in bolded type with an acute accent and
the pre-stressing suffix is underlined.11
(5) a. gel -di -niz
come -PST -2PL
‘You came’
b. gel -me -di -niz
come -NEG -PST -2PL
‘You didn’t come’
c. gel -di -niz -mi
come -PST -2PL -QUES
‘Did you come?’
(6) a. sakla -n -dɯ -laɾ
hide -RECIP -PST -3PL
‘They hid (themselves)’
b. sakla -n -dɯ -laɾ -da
hide -RECIP -PST -3PL -CONN
‘They also hid (themselves)’
c. sakla -n -mɯʃ -tɯɾ -dı -lar -da
hide -RECIP -EVID -EP.COP -PST -3PL -CONN
‘They might also have hidden (themselves)’
(Kabak & Vogel, 2001: 317)
The (a) examples above illustrate the regular Turkish main stress pattern. The (b) and (c) examples show
stress falling on the syllable preceding the presence of a pre-stressing suffix. Stress placement is prevented
on or following these suffixes, thus resulting in non-final stress.
2.2.2 Stressed suffixes
Stressed suffixes are pre-specified by being marked with stress. Consider the following list of
lexically stressed suffixes. The stressed syllable is marked in bolded type with an acute accent.
11 The leftmost stress surfaces as the primary stress of the word when two or more irregular stress patterns combine
with each other (seen in (6c)). This interaction is not discussed here since it goes beyond the scope of this paper.
HEATHER LYNN YAWNEY
6
(7) Stressed suffixes
a. -Íjor ‘progressive’
b. -Éɾek ‘by’
c. -ÍndʒE ‘when’
(Inkelas & Orgun, 2003: 141)
Main stress falls on the first syllable of the bisyllabic stressed suffixes (Inkelas & Orgun, 2003; Özçelik
2014). Regardless of the rhyme structure, the main stress of a word is required to fall on the stressed suffix.
To illustrate stressed suffixes in action, consider the following examples. The stressed syllable is marked
in bolded type with an acute accent.
(8) a. yap -arak -mi
do -ADV -QUES
‘Is it by doing?’
b. yap -iyor -mu?
do -PRS -QUES.3SG
‘Is he/she/it doing?’
(Inkelas & Orgun, 2003: 142)
The presence of a stressed suffix prevents stress to fall on any following suffixes, thus resulting in non-
final stress.
2.2.3 Stressed roots
Stressed roots are pre-specified for main stress. Consider the following list of stressed roots.12 The
stressed syllable is marked in bolded type with an acute accent.
(9) Stressed roots
c. ankaɾa ‘Ankara’ h. adʒaba ‘one wonders’
d. yskydar ‘Üsküdar’ k. akvaɾjum ‘aquarium’
i. beltʃika ‘Belgium’ q. negatif ‘negative’
j. kastamonu ‘Kastamonu’ r. tʃanta ‘bag’
k. baɾbaɾa ‘Barbara’ s. amdʒa ‘uncle’
l. mandela ‘Mandela’ t. baɾaka ‘shed’
m. fakat ‘but’ u. pendʒeɾe ‘window’
(Kabak & Vogel, 2001: 316; Özçelik, 2014)
The position of main stress varies but usually falls on the penultimate or antepenultimate syllable.
There is a certain tendency towards quantity sensitivity to describe a category of irregularly stressed roots
consisting of loanwords and place names (Sezer, 1983). These items have come to be known as Sezer roots.
The rule is formulated as followed.
(10) Sezer stress rule
If the antepenultimate syllable is heavy and the penultimate syllable is light, stress the
antepenultimate; otherwise stress the penultimate syllable.
12 This list is not exhaustive.
ON THE RIGHT TO BE FAITHFUL: THE IRREGULAR STRESS OF TURKISH ADVERBIALS WITH -EN
7
However, there are exceptions to this rule, even though this correctly predicts many irregularly
stressed roots in Turkish. A dictionary survey of polysyllabic words by Cakır (2000) revealed that, of 206
irregularly stressed place names that contain a heavy antepenultimate syllable and a light penultimate
syllable, 155 have main stress on the penultimate syllable and only 51 have main stress on the
antepenultimate syllable. These results do not favour the pattern predicted by the Sezer rule. The Sezer
stress pattern is interesting and worth mentioning since several authors treat Sezer roots as a separate stress
pattern from other stressed roots. (c.f. Inkelas, 1999; Inkelas & Orgun, 2003; Kabak & Vogel, 2011; Sezer
1983). However, following Kabak & Vogel (2001) and Özçelik (2014), I treat stressed roots, whether Sezer
roots or not, by grouping them into one big category having an irregular stress pattern without splitting
irregularly stressed roots into different sub-categories.
2.2.4 Turkish adverbials with -En
Adverbials with the originally Arabic element -En have a variant stress position depending on the
presence of a long vowel.13
(11) Common bisyllabic Turkish adverbials with -En
a. naklen ‘by transfer’ g. muvakkaten ‘temporarily’
b. nakten ‘in cash’ h. nisbeten ‘proportionally’
c. seklen ‘formwise’ i. iktisa:den ‘economically’
d. mynhasɯɾan ‘specially’ j. istina:den ‘based on’
e. myʃtereken ‘mutually’ k. kaza:en ‘accidentally’
f. tekeffylen ‘by surety’ l. esa:sen ‘basically’
(Sezer, 1983: 62–63)
As mentioned in the introductory section, the Turkish adverbial with -En data shows that disyllabic words
have initial stress (8a–c) and polysyllabic words have stress on the antepenultimate syllable (8d–h) unless
the penultimate syllable contains a long vowel (8i–l). Erdal (2009) describes this as stress being retracted
to the third syllable from the end unless the subsequent one possesses a long vowel. He generalises that
adverbial stress retraction stops at the third syllable from the end.14 Sezer (1983) describes this pattern by
making a clear distinction between strong and weak syllables. Strong syllables are ones that either end with
one or more consonants or with a long vowel. This includes the possible following shapes: (C)VC, (C)VCC
or (C)V:. A weak syllable is one which ends with a short vowel, (C)V. The main stress of Turkish adverbials
with -En falls on the antepenultimate syllable if the penultimate syllable is weak. Sezer formulates the
pattern as the following:
(12) In the adverbials derived with -En, primary stress is on the penult if it is strong. If the penult is not
strong, then primary stress is on the ante-penult (Sezer, 1983: 63).
This means that the stressed antepenultimate syllable does not need to be strong.
13 This section discusses what has been said in previous literature. The following section addresses that only long
vowels contribute to syllable weight. 14 This generalisation includes the antepenultimate and the penultimate syllables. Erdal (2009) discusses this
generalisation by illustrating that, if adverbials have irregular stress, stress will either fall on the penultimate or the
antepenultimate syllable and will not fall on any other syllable further away from the end of the word. He illustrates
this generalisation by including other examples of adverbs that don’t fall under the Turkish adverbials with -En
umbrella.
HEATHER LYNN YAWNEY
8
2.3 Summary
Now that I have introduced several Turkish stress patterns to be accounted for in this paper (regular
main stress and several irregular main stress patterns including pre-stressing suffixes, stressed suffixes,
stressed roots and adverbials with -En), the following section will account for the stress pattern facts using
an Optimality Theory framework. The analysis discusses how the regular and irregular stress patterns are
not associated with different co-phonologies. Özçelik’s (2014) original proposal demonstrated that prosodic
faithfulness to the right edge of a pre-specified foot in underlying representation can account for regular
and irregular stress phenomena under a single grammar approach. I propose that modifying the analysis to
involve only a pre-specified right edge foot bracket in underlying representation will allow a single
grammar approach of Turkish stress patterns to extend to the stress pattern surfacing for Turkish adverbials
with -En.
3 OT analysis
This section tackles an OT analysis that accounts for the different stress patterns discussed above.
Özçelik (2014) proposes a single grammar approach using prosodic faithfulness to a pre-specified foot’s
right edge in underlying representation to account for regular stress, pre-stressing suffixes, stressed suffixes
and stressed roots. The pre-specified foot guarantees a foot to be interpreted in output. I propose that a
single grammar approach can be extended to Turkish adverbials with -En if I modify the analysis to involve
only a pre-specified right edge foot bracket (à la Idsardi, 2009) in underlying representation. This also
guarantees a foot to be constructed in output, and by extension stress assignment on that foot. In this analysis,
I support a quantity sensitive system (c.f. Hayes, 1995) whereby only long vowels contribute to the
syllable’s weight. The reasoning behind this is because when stress falls on the antepenultimate syllable,
we find either an open (C)V or a closed (C)VC syllable. Also, the Sezer stress rule, which is sensitive to
syllable weight, introduced in §2.2.3 when discussing the stressed roots, cannot withstand the argument
that coda consonants contribute to syllable weight. Through the evaluation process of the generated optimal
candidates within the OT framework, irregular stress patterns can be explained by classifying Turkish as a
trochaic language (c.f. Inkelas, 1999; Inkelas & Orgun, 1998; Özçelik 2014).
The regular and irregular stress patterns are not associated with different co-phonologies (c.f. Inkelas
& Orgun 1998). Through a single grammar approach, this account of different Turkish stress patterns can
be captured using a single constraint ranking. Turkish is a trochaic language with PARSE-σ being ranked
low.
(13) TROCHEE: Align stress to the left edge of the foot.
(14) PARSE-σ: Syllables are parsed by feet (McCarthy & Prince, 1993).
In the case of regular stress, the absence of a pre-specified foot edge in underlying representation means
that TROCHEE does not apply and is vacuously satisfied.15 A higher ranked constraint, FINAL PROMINENCE,
means that stress (known as intonational prominence in the case of regular stress) will fall on the final
syllable of a prosodic word, resulting in regular word-final main stress.16
(15) FINAL PROMINENCE: Put a boundary tone at the end of a prosodic word (Özçelik, 2014).
15 Regular stress does not involve a pre-specified foot edge in the input and thus, does not result in foot construction
in the output. Since TROCHEE does not apply unless there is a foot, the candidate evaluation does not incite any
violations for TROCHEE. With no violation marks, TROCHEE is satisfied. 16 In §3.1 will offer a brief discussion behind Özçelik’s (2014) reasons behind FINAL PROMINENCE. See §3.1 in Özçelik
(2014) for a more in-depth discussion on why he analyses Turkish final stress as final (intonational) ‘prominence’
falling on the last syllable of a prosodic word.
ON THE RIGHT TO BE FAITHFUL: THE IRREGULAR STRESS OF TURKISH ADVERBIALS WITH -EN
9
In cases of irregular stress, the pre-specified right edge foot bracket is specified in the input and will have
to be parsed in the output because of the high-ranking ANCHOR-R prosodic faithfulness constraint.
(16) ANCHOR-R: The right edge of every foot in the input corresponds to the right edge of some foot in
the output (McCarthy & Prince, 1995).
With a foot generated in the output (the left and right edge), TROCHEE takes effect, allowing the irregular
stress to surface. FT-BIN will parse a foot of the type (ˈL L) and (ˈH). Moras are not sensitive to codas, thus,
coda consonants do not participate in syllable weight.
(17) FT-BIN: Feet must be binary under syllabic or moraic analysis (McCarthy & Prince, 1993).
Crucial to the overall analysis, it is the right edge foot bracket that is pre-specified in the underlying
representation and not the location of stress or a pre-specified well-formed foot in the underlying
representation.
The only difference between regular and irregular stress is whether the word has a pre-specified right
edge foot bracket in the input or not. Both patterns are subject to the same constraint ranking but are
different with regards to irregular non-final stress involving a right edge foot bracket in underlying
representation and regular final stress not involving a pre-specified foot bracket. All the while, the two
patterns involve the same grammar. Consider the following constraint ranking.
(18) ANCHOR-R >> TROCHEE, FT-BIN >> FINAL PROMINENCE >> PARSE-σ
This constraint ranking results in optimal outputs that account for the regular and irregular stress patterns
in Turkish. This includes Turkish adverbials with -En. The following subsections will demonstrate how the
proposed ranking results in all the surface stress patterns accounted for in this paper.
3.1 Regular stress
The regular stress analysis follows Özçelik’s (2014) proposal that treats final main stress in Turkish
as an intonational prominence at the end of a phonological word.17 This is the result of associating different
phonetic cues to the prominences of regular versus irregular stress. Irregular stress patterns seem to be a
true foot-based stress with a sharp F0 rise and greater intensity, while regular stress with final prominence
is marked with only a slight F0 rise (Konrot, 1981, 1987; Levi, 2005; Pycha, 2006).18 Final stress is not the
result of a trochaic stress but the effect of FINAL PROMINENCE. This constraint comes into play by placing
prominence on the final syllable of a phonological word with the absence of a pre-specified right edge
bracket in underlying representation. Consider the following constraint ranking.
(19) TROCHEE, FT-BIN >> FINAL PROMINENCE >> PARSE-σ
Now, consider the following tableau.
17 See Inkelas (1999) for an alternate analysis which proposes final catalexis of the trochaic foot in the regular main
stress pattern. 18 For some speakers, regular stress is marked with a plateau with no rise at all (Levi, 2005).
HEATHER LYNN YAWNEY
10
(20) Regular main stress19,20
/kitap-lɯk/ TROCHEE FT-BIN FINAL PROMINENCE PARSE-σ
a. ki.('tap.lɯk) *! *
b. ki.(tap.'lɯk) *! *
c. ki.tap.('lɯk) *! **
d. H%
|
ki.tap.lɯk
***
Candidate (20a) fatally violates FINAL PROMINENCE although it is trochaic. Candidate (20b) fatally violates
undominated TROCHEE since it contains an iambic foot. Candidate (20c) fatally violates UNDOMINATED FT-
BIN and satisfies FINAL PROMINENCE. Candidate (20d) wins because none of the high-ranking constraints
have been violated; it only incurs violations to the lowest ranking constraint, PARSE-σ. For regular word-
final main stress in Turkish, it is better not to have a pre-specified right edge foot bracket than to have one
and violate well-formedness constraints or trochaicity.21
3.2 Irregular stress
Irregular stress patterns differ from the regular stress pattern in that they are pre-specified with a right
edge foot bracket in the input. This is where the higher-ranking ANCHOR-R faithfulness constraint is
necessary to guarantee that the outputs respect the right edge of the foot. Consider the following new
constraint ranking.
(21) ANCHOR-R >> TROCHEE, FT-BIN >> FINAL PROMINENCE >> PARSE-σ
In the next few subsections, I will look at the different irregular stress patterns, including pre-stressing
suffixes (§3.2.1), stressed suffixes (§3.2.2), stressed roots (§3.2.3) and adverbials with -En (§3.2.4), which
all conform to the above constraint ranking.
3.2.1 Pre-stressing suffixes
Main stress falls on the syllable directly preceding the pre-stressing suffix. The right edge bracket
will ensure that TROCHEE and FT-BIN are respected in the output. If these two constraints are violated, a
monosyllable could surface with the stress surfacing on the pre-stressing suffix itself. Given that the crucial
characteristic of these suffixes is their pre-stressing effect, and given the trochaic nature of Turkish, a
bisyllabic foot is necessary for stress to fall on the syllable preceding the pre-stressing suffix, all while
maintaining the right edge of the foot. Consider the following tableau.
19 Özçelik (2014) illustrates that candidates similar to (20c) with a monosyllabic/monomoraic foot candidate are all
trochaic. Even though he does not provide an explanation for this, stress can still be argued as aligning with the left
edge of the foot, under my definition of TROCHEE, seen in (13). 20 H% and the association line represent the intonation prominence at the end of a phonological word in the absence
of a pre-specified bracket edge. 21 This is inspired by Özçelik’s (2014) original argument that not having a foot is better than having one and
violating well-formedness constraints or trochaicity.
ON THE RIGHT TO BE FAITHFUL: THE IRREGULAR STRESS OF TURKISH ADVERBIALS WITH -EN
11
(22) Pre-stressing suffixes
/gel-me)-di/ ANCHOR-R TROCHEE FT-BIN FINAL PROMINENCE PARSE-σ
a. ('gel.me).di * *
b. (gel.'me).di *! * *
c. gel.('me.di) *! * *
d. gel.('me).di *! * **
e. H%
|
gel.me.di
*! ***
Candidate (22a) wins with the only violations incurring on the lower ranking constraints. The highest
ranked constraint ANCHOR-R is satisfied by maintaining the right edge of the foot in the output. Candidate
(22b) fatally violates TROCHEE since it contains an iambic foot. Candidate (22c) fatally violates
undominated ANCHOR-R. Candidate (22d) fatally violates FT-BIN. The bracketless candidate (22e) fatally
violates undominated ANCHOR-R.
3.2.2 Stressed suffixes
Stressed suffixes place main stress on the first syllable of these two syllable suffixes. Given TROCHEE,
main stress falls on the first syllable. The anchoring constraint is satisfied if the right edge of the foot is
maintained from input to output. Consider the following tableau.22
(23) Stressed suffixes
/gel-ince)/ ANCHOR-R TROCHEE FT-BIN FINAL PROMINENCE
a. gel.('in.ce) *
b. gel.(in.'ce) *!
c. ('gel.in).ce *! *
d. gel.in.('ce) *!
e. H%
|
gel.in.ce
*!
Candidate (23a) wins with violations incurred only on lower ranked constraints. The highest ranked
constraint ANCHOR-R is satisfied by maintaining the right edge of the foot in the output. Candidate (23b)
fatally violates TROCHEE since it contains an iambic foot. Candidate (23c) fatally violates the undominated
ANCHOR-R. Candidate (23d) fatally violates FT-BIN, while still satisfying FINAL PROMINENCE. The
bracketless candidate (23e) fatally violates the undominated ANCHOR-R.
3.2.3 Stressed roots
The syllable on which stress falls is pre-specified for stressed roots. As in the previous two sections,
the right edge foot bracket in the input will ensure that TROCHEE and FT-BIN are respected in the output.
However, these stressed roots show variation in terms of where the stress falls. Stressed roots can be divided
into two categories.
22 Henceforth, PARSE-σ is dropped from all following tableaux. This constraint incurs violations with no effect on
determining the winning candidate. I have done this to simplify the tableaux.
HEATHER LYNN YAWNEY
12
1. Stressed roots with penultimate stress
2. Stressed roots with antepenultimate stress
The differing surface stress placement may seem confusing for stressed roots. I argue that this is dependent
on the position of the pre-specified right edge foot bracket.
First, consider the following tableau for stressed roots with penultimate stress. The pre-specified right
edge foot bracket falls at the end of the word.
(24) Stressed roots with penultimate stress
/fabɾika)/ ANCHOR-R TROCHEE FT-BIN FINAL PROMINENCE
a. fa.('bɾi.ka) *
b. fa.(bɾi.'ka) *!
c. ('fa.bɾi).ka *! *
d. fa.bɾi.('ka) *! *
e. H%
|
fa.bɾi.ka
*!
Candidate (24a) wins with violations incurred only on lower ranked constraints. Candidate (24b) fatally
violates TROCHEE since it contains an iambic foot. Candidate (24c) fatally violates the undominated
ANCHOR-R. Candidate (24d) fatally violates FT-BIN. The bracketless candidate (24e) fatally violates the
undominated ANCHOR-R.
Second, consider the following tableau for stressed roots with penultimate stress. The right edge foot
bracket falls within the word instead of at the right word boundary.
(25) Stressed roots with antepenultimate stress
/'nega)tif/ ANCHOR-R TROCHEE FT-BIN FINAL PROMINENCE
a. ('ne.ga).tif *
b. (ne'ga).tif *!
c. ne.('ga.tif) *! *
d. ne.('ga).tif *! *
e. H%
|
ne.ga.tif
*!
The same explanations apply here in (25) as they did for (24). Candidate (25a) wins with violations incurred
only on lower ranked constraints. Candidate (25b) fatally violates TROCHEE since it contains an iambic foot.
Candidate (25c) fatally violates the undominated ANCHOR-R. Candidate (25d) fatally violates FT-BIN. The
bracketless candidate (25e) fatally violates the undominated ANCHOR-R.
The above tableaux for stressed roots demonstrates that the same constraint ranking will account for
the surfacing stress facts. Stressed root with penultimate or antepenultimate stress surfaces depending on
the position of the pre-specified right edge foot bracket within the word.
3.2.4 Turkish adverbials with -En
Now that I have shown how an OT analysis can account for other Turkish stress patterns, I can finally
demonstrate how this system can be extended to the Turkish adverbials with -En. The pre-specified right
edge foot bracket can be extended to the stress phenomenon observed by Sezer (1983) and Erdal (2009).
ON THE RIGHT TO BE FAITHFUL: THE IRREGULAR STRESS OF TURKISH ADVERBIALS WITH -EN
13
As mentioned in §1, this analysis is inspired by Özçelik’s (2014) original single grammar proposal whereby
prosodic faithfulness to the right edge of the foot was discussed. By promoting the right edge anchoring
constraint (with a left edge anchoring constraint being irrelevant here because I do not propose a left edge
bracket in the input) and emphasizing the crucial characteristic to this analysis being respect for the right
edge of the foot, I am able to extend a single grammar approach to Turkish adverbials with -En.
While Sezer (1983) treats the stress pattern of Turkish adverbials with -En in terms of strong and
weak syllables, and a quantity sensitive system, it seems that it is only the length of the vowel that affects
on which syllable stress falls, shown by Erdal (2009). I also support a quantity sensitive system but only
one whereby vowels contribute to the weight of the syllable. This is maintained by the FT-BIN constraint
put forth at the beginning of this section. Hayes (1995) demonstrates that moraic theory involves some
variation. It is clear that (C)V is monomoraic and (C)V: is bimoraic. However, whether (C)VC is light or
heavy varies across languages. Therefore, it is necessary to specify the parameter set of what constitutes a
light and heavy syllable. Since Erdal (2009) observes that it is the long vowel that blocks stress shift, I
propose that only long vowels (bimoraic) in Turkish affect syllable weight and coda consonants have no
effect. Thus, Sezer’s (1983) strong syllable, a closed syllable, is monomoraic in this analysis. This favours
only the vowel contributing to the weight of the syllable.23
Identical to the other irregular stress patterns discussed in this paper (pre-stressing suffixes, stressed
suffixes and stressed roots), the right edge bracket will ensure TROCHEE and FT-BIN are respected in the
output. However, in the case of Turkish adverbials with -En, I argue for the pre-specified right edge bracket
falling to the left of the -En suffix: -)En. I have divided Turkish adverbials with -En into three categories.
1. Polysyllabic adverbials with main stress on the antepenultimate syllable (see the tableau in (26),
2. Polysyllabic adverbials with main stress on the penultimate syllable (see the tableau in (27))
3. Bisyllabic adverbials with penultimate stress (see the tableau in (28))
It may be confusing with what seems to be different surfacing stress patterns on the Turkish adverbials with
-En but I will demonstrate that the same constraint ranking will account for all the facts. First, consider the
following tableau for polysyllabic adverbials with main stress falling on the antepenultimate syllable.
(26) Polysyllabic adverbials with main stress on the antepenultimate syllable
/münhası)ɾen/ ANCHOR-R TROCHEE FT-BIN FINAL PROMINENCE
a. mün.('ha.si).ɾen *
b. mün.(ha.'si).ɾen *!
c. ('mün.ha).si.ɾen *! *
d. mün.ha.('si).ɾen *! *
e. H%
|
mün.ha.si.ɾen
*!
23 The original Arabic nouns before affixing the adverbializing -En suffix had regular word-final stress, whether the
word ended in a closed (C)VC or an open syllable with a long vowel (C)V:. There were no instances of the Arabic
nouns ending in an open syllable with a short vowel (C)V. It is difficult to show that only long vowels affect
syllable weight if one considers only Turkish adverbials with -En, and that resyllabification results in open
penultimate syllables with a short vowel. In §2.2.3 on stressed roots, I introduced the Sezer stress rule, but there
are exceptions to this rule. This leads me to oppose coda consonants contributing to syllable weight and stress
assignment, and to follow Kabak & Vogel (2001) and Özçelik (2014) in treating stressed roots under one group.
This paper follows the data introduced in Sezer (1983) and Erdal (2009). Inkelas (1999) discusses the
inconsistency and the lack of productivity of Turkish adverbials with -En. Another dictionary survey and further
elicitation would hopefully be revealing. This goes beyond the theme of this volume and is for future research.
HEATHER LYNN YAWNEY
14
Candidate (26a) wins with the only violation incurring on lower ranking constraints. Candidate (26b) fatally
violates TROCHEE since it contains an iambic foot. Candidate (26c) fatally violates the undominated
ANCHOR-R. Candidate (26d) fatally violates FT-BIN. The bracketless candidate (26e) fatally violates the
undominated ANCHOR-R.
Second, consider, the following tableau for polysyllabic adverbials with main stress falling on the
penultimate syllable.
(27) Polysyllabic adverbials with main stress on the penultimate syllable
/kaza)en/ ANCHOR-R TROCHEE FT-BIN FINAL PROMINENCE
a. ka.('za).en *
b. ('ka.za).en *! *
c. (ka.'za).en *! *
d. H%
|
ka.za.en
*!
Candidate (27a) wins with violations incurring only on lower ranked constraints. Candidate (27b) fatally
violates FT-BIN and has more violation marks than candidate (27a). Candidate (27c) fatally violates
TROCHEE since it contains an iambic foot. The bracketless candidate (27d) fatally violates undominated
ANCHOR-R.
Third, consider the following tableau for bisyllabic adverbials.
(28) Bisyllabic adverbials
/nak)ten/ ANCHOR-R TROCHEE FT-BIN FINAL PROMINENCE
a. ('nak).ten * *
b. ('nak.ten) *! *
c. H%
|
nak.ten
*!
Candidate (28a) wins with the violation incurring only on lower ranked constraints. Candidate (28b) fatally
violates the undominated ANCHOR-R. The bracketless candidate (28c) fatally violates the undominated
ANCHOR-R. It may seem unsettling that FT-BIN is violated, when it usually results in a fatal violation. But
this reveals that the system favours violations of FT-BIN over ANCHOR-R. If the choice were given between
candidates with a single violation of one of the higher ranked constraints, maintaining a foot edge is more
valuable than foot binarity when it comes to stress assignment. There is no other possibility than to assign
the left edge foot bracket and form a single foot in the output for stress to surface.
3.3 Summary
This section provided an OT account of the regular stress pattern and several irregular stress patterns
including pre-stressing suffixes, stressed suffixes and stressed roots in Turkish. I offered an analysis that
unified Turkish adverbials with -En with the regular stress pattern and the irregular stress patterns discussed
in this paper under a single constraint ranking by proposing a pre-specified right edge foot bracket. The
following section will briefly conclude the findings.
ON THE RIGHT TO BE FAITHFUL: THE IRREGULAR STRESS OF TURKISH ADVERBIALS WITH -EN
15
4 Conclusion
To conclude, the goal of this paper was to offer an OT analysis that accounts for the regular and
several irregular stress patterns, including pre-stressing suffixes, stressed suffixes, stressed roots and
adverbials with -En in Turkish. This account captures all the surface stress patterns by being subject to a
single constraint ranking.
(29) ANCHOR-R >> TROCHEE, FT-BIN >> FINAL PROMINENCE >> PARSE-σ
Inspired by Özçelik’s (2014) single grammar approach whereby all surface stress patterns can be explained
using a single constraint ranking, and following Idsardi’s (2009) single bracketing approach whereby a
single boundary is enough to define a foot grouping, I was able to propose a single constraint ranking that
involves faithfulness to a pre-specified right edge of a foot bracket. This ensured that the right edge in the
input is respected in the output. My approach disregarded co-phonology proposals and treated the system
as a whole. This allowed for a unified system of stress in Turkish that could be used across different surface
stress patterns and extended to include the Turkish adverbials with -En.
References
Cakır, M.C. (2000). On non-final stress in Turkish simplex words. In A. Göksel & C. Kerslake (eds.) Studies
on Turkish and Turkic languages: Proceeding of the ninth international conference on Turkish
Linguistics (pp. 3-10). Harrassowitz: Wiesbaden. 3-10.
Erdal, M. (2009). Stress and the Turkish adverb. In S. Ay, Ö. Aydın, I. Ergenç, S. Gökmen, S. Işsever &
D. Peçenek (Eds.), Essays on Turkish linguistics (pp. 17–22). Harrassowitz Verlag: Wiesbaden.
Hayes, B. (1995). Metrical stress theory: Principles and case studies. Chicago, IL: The Chicago University
Press.
van der Hulst, H. & van de Weijer, J. (1991). Topics in Turkish phonology. In H. Boeschoten & L.
Verhoeven (Eds.), Turkish linguistics today (pp. 11–59). Leiden, Netherlands: Brill.
Idsardi, W.J. (2009). Calculating metrical structure. In C. Cairns & E. Raimy (Eds.), Contemporary views
on architecture and representations in phonological theory (pp. 191–211). Cambridge, MA: MIT
Press.
Inkelas, S. (1999). The exceptional stress-attracting suffixes in Turkish: Representations versus the
grammar. In R. Kager, H. van der Hulst & W. Zonneveld (Eds.), The prosody–morphology interface
(pp. 134–187). Cambridge, UK: Cambridge University Press.
Inkelas, S. & Orgun, C.O. (1998). Level (non)ordering in recursive morphology: Evidence from Turkish.
In S.G. Lapointe, D.K. Brentari & P. M. Farrell (Eds.), Morphology and its relation to phonology
and syntax (pp. 360–410). Stanford: CSLI.
Inkelas, S. & Orgun, C.O. (2003). Turkish stress: A review. Phonology, 20(1), 139–161.
Kabak, B. & Vogel, I. (2001). The phonological word and stress assignment in Turkish. Phonology, 18,
315–360.
Kabak, B. & Vogel, I. (2011). Exceptions to stress and harmony: Co-phonologies or prespecification? In
H. J. Simon & H. Wiese (Eds.), Expecting the unexpected: Exceptions in grammar (pp. 59–94).
Berlin/New York: De Gruyter Mouton.
Kamali, B. (2011). Topics at the PF interface in Turkish. (Doctoral dissertation). Harvard University,
Cambridge, MA.
Konrot, A. (1981). Physical correlates of linguistic stress in Turkish. University of Essex Language Centre
Occasional Papers, 24, 26–53.
HEATHER LYNN YAWNEY
16
Konrot, A. (1987). Stress in Turkish: Is it determined phonologically or morphologically? In H.E.
Boeschoten & L. T. Verhoeven (Eds.), Studies on modern Turkish: Proceedings of the 3rd conference
on Turkish linguistics (pp. 3–12). Tilburg, Netherlands: Tilburg University Press.
Lees, R. (1961). The phonology of Modern Standard Turkish. Bloomington, IN: Indiana University Press.
Levi, S.V. (2005). Acoustic correlates of lexical accent in Turkish. Journal of the International Phonetic
Association, 35, 73–97.
Lewis, G. (1967). Turkish grammar. Oxford, UK: Clarendon.
McCarthy, J. & Prince, A. (1993). Generalized alignment. In G. Booij & J. van Marle (Eds.), Yearbook of
morphology. Dordrecht: Kluwer.
McCarthy, J. & Prince, A. (1995). Faithfulness and reduplicative identity. In J.N. Beckman, L. Walsh
Dickey & S. Urbanczyk (Eds.), Papers in Optimality Theory (pp. 249–384). Amherst, MA: GLSA.
Özçelik, Ö (2014). Prosodic faithfulness to foot edges: The case of Turkish stress. Phonology, 31, 229–269.
Prince, A. & Smolensky, P. (1993) Optimality Theory: constraint interaction in generative grammar. Ms,
Rutgers University & University of Colorado, Boulder. Published 2004, Malden, Mass. & Oxford:
Blackwell.
Pycha, A. (2006). A duration-based solution to the problem of stress realization in Turkish. UC Berkeley
Phonology Lab Annual Report, 141–151.
Sezer, E. (1983). On non-final stress in Turkish. Journal of Turkish Studies, 5, 61–69.
Underhill, R. (1976). Turkish grammar. Cambridge, MA: MIT Press.