Violation of Onset Constraint in Word Initial Syllable (Analysis in English and Lithuanian)

Lingvistika / Linguistics ISSN 1392-8600 E-ISSN 1822-7805 Žmogus ir žodis / Svetimosios kalbos Man and the Word / Foreign Languages 2015, t. 17, Nr. 3, p. 54 65 / Vol. 17, No. 3, pp. 54 65, 2015 Violation of Onset Constraint in Word Initial Syllable (Analysis in English and Lithuanian) Edita Katinaitė-Kalčiūnienė Lithuanian University of Educational Sciences, Faculty of Philology, Department of English Didactics, Studentų St. 39, LT-08106 Vilnius, edita.kalciuniene@leu.lt Abstract. ONSET of the word-initial syllables in English and Lithuanian has been investigated in the theoretical framework of Optimality Theory (1397 word-initial syllables in Lithuanian and the same number of word-initial syllables in English). The results of the investigation have shown that the Lithuanian language exhibits fewer cases of ONSET violations in word-initial syllable than the English language. Thus, the Lithuanian language is less marked in this respect. Accordingly, ONSET constraint in Lithuanian may prove to be occupying a higher position in the constraint hierarchy of the Lithuanian language. The hierarchy of constraints in syllable formation in Lithuanian is not established; the present investigation is a contribution towards this end. Keywords: Optimality Theory, onset, syllable, constraint hierarchy, sonority. Introduction All major approaches to phonology have recognised the syllable as the fundamental unit in phonological analysis. However, in the article the discussion is confined to the treatment of the syllable in the theory of generative linguistics, i.e. Optimality Theory (OT) which was introduced in the 1990s (Prince & Smolensky, 1991; McCarthy & Prince, 1993, 1996; Archangeli, 1998; Hammond, 1998). This theory was introduced in response to the universal grammar of the 1960 s that claimed language to be rule-governed. Accordingly, the syllable formation process was treated as a process with its rules, without analysing the reasons of the syllable being closer or further off the universal pattern. Contrary to 54 DOI: http://dx.doi.org/10.15823/zz.2015.11

ISSN 1392 8600 Lingvistika / Linguistics the rule-based approach, OT totally rejects the rule component. Instead, it introduces universal constraints that are violable. They are present in the constraint hierarchy that is language- specific. A lower-ranked constraint may be violated in order to satisfy a higher-ranked constraint. Thus, ONSET constraint and its impact on the formation of word-initial syllables in English and Lithuanian is the object of the investigation. The analysis of ONSET constraint is a contribution to the establishment of the constraint hierarchy in Lithuanian syllabification. The Constraint Hierarchy in Syllabification Within the scope of OT the typical properties of syllables were pointed out by Archangeli (1998): syllables begin with a consonant; syllables have one vowel; syllables end with a vowel; syllables have at most one consonant at an edge; syllables are composed of consonants and vowels. It has to be underlined that the statements presented above are only general tendencies, not unquestionable laws. The universal properties of syllables made it possible to formulate the universal constraints in syllabification: ONSET, PEAK, NOCODA, *COMPLEX, FAITHFULNESS (FAITH and FAITHV). *COMPLEX stands for complex onsets and complex codas are unacceptable. Every syllable has a peak, the most sonorous segment within the syllable. Thus, the PEAK constraint requires that syllables have one vowel. Segments preceding a peak constitute a unit called an onset. Accordingly, the ONSET constraint requires that syllables begin with a consonant. Segments following a peak form the phonological unit a coda. Obviously then, the NOCODA constraint requires that syllables end with a vowel. There are syllables in languages that begin with a consonant cluster, or end with a consonant cluster. Then the *COMPLEX constraint indicates that syllables have at most one consonant at an edge. The symbol *, as has been noted above, is used to indicate unacceptability. Finally, The FAITHFULNESS constraint requires that input and output in generation coincide. OT, like other models of linguistics, proposes input and output and the relation between them. In the theory, the relation between input and output is arbitrated by two formal mechanisms: GEN (for Generator) and EVAL (for Evaluator). EVAL selects the optimal candidate (s) produced by GEN, making use of the language-specific constraint hierarchy, i.e. particular ranking of CON (the universal set of constraints) (Steponavičienė, Katinaitė, 2001, 42). 55

Lingvistika / Linguistics ISSN 1392 8600 The typical properties of a syllable given above are viewed as constraints on specific aspects of a syllable. Each of them expresses a strong universal tendency. For instance, the Lithuanian words arklys, upė prove that not all languages require onsets (ONSET). On the other hand, every language allows onsets. The existence of syllable initial consonant clusters br, pl or the existence of syllable-final consonant clusters mp, lk in English obviously prove the discussed language to be violating the *COMPLEX constraint. However, the English words bring, play, camp, milk as well as the Lithuanian words arklys, upė are still acceptable words in these languages. The possible violation of these constraints leads to two important results. First of all, such violations allow language specific patterns and variation between languages to be admitted into the model. Second, via constraint violations markedness is also admitted into the model. Most importantly, introducing constraints the model at the same time addresses the central issues in linguistics which previous models failed to address. To go further, for satisfying universal constraint requirements, languages allow different constraint violations, and thus create different constraint hierarchies. Concrete examples will show more explicitly how differences between constraint rankings result in different patterns, giving rise to systematic variation between languages. A good illustration is, e.g. a consonant sequence that appears in languages in word derivation or inflection. To illustrate, mirk, the verb root for the Lithuanian word, mirkti with sta, marking the present tense third person singular, would create a four-consonant sequence, which would be a violation of *COMPLEX constraint, because both the coda of the first syllable as well as the onset of the second syllable would have a complex unit (mirk.sta). Archangeli (1998) gives a similar example in Yawelmani, a native American language. Logw, the verb root for pulverize with hin, marking the past tense, would create a three-consonant sequence, which would also be a violation of *COMPLEX constraint because either the coda of the first syllable or the onset of the second syllable would be bound to have a complex unit as well (either logw.hin or log.whin). The analysis of the Yawelmani language proves the discussed constraint to be highly ranked. Then, the syllables of this type do not appear in the language. The second possible violation would be FAITHC, effected by deleting one of the consonants (e.g. log.hin) which actually does not happen in the Yawelmani language either. The third possible solution might be the violation of the PEAK constraint, creating the third syllable containing not a vowel, but a consonant (like log.w.hin). However, the Yawelmani language does not behave in this way either. This language violates FAITHV constraint by including an extra vowel (i.e., lo.giw.hin) (Steponavičienė, Katinaitė, 2001, 42). Further investigation of this language made it possible to construct the analysed constraint hierarchy in Yawelmani (Archangeli, 1998, 20): *COMPLEX, FAITH, PEAK, FAITHV. Accordingly, FAITHV is the most prone to violations and its position in the hierarchy of the Yawelmani language secures more-high ranking constraints. 56

ISSN 1392 8600 Lingvistika / Linguistics The constraint hierarchy is different in English syllabification. The ranking of the four constraints, analysed in Yawelmani, appears to be: FAITHV, FAITHC, PEAK, *COMPLEX (Archangeli, 1998, 23). Obviously, *COMPLEX is violated by allowing, e. g., limp.ness, which is a complex coda. If the word became either lim.pi.ness or lim.ness, or lim.p.ness. it would undoubtedly be violations of FAITHV, FAITHC, PEAK. Hammond (1998) presents a full analysis of the constraint hierarchy in English syllabification. The constraint hierarchy based on his analysis looks like this: PEAK, LICENS- ING, SONORITY FAITHFULNESS > ONSET, NOCODA, *COMPLEX. By LICENSING Hammond (1998) means the composition of words of syllables (which was pointed out by Hooper, 1972; Kahn, 1976; Ito, 1989 et al). In English, there is an additional requirement, namely, that of a PEAK. These two constraints cannot be violated. Consequently, they are placed at the top of the constraint hierarchy. Then, violations of ONSET and NOCODA are tolerated in English in order to satisfy FAITFULNESS. Ultimately, then, FAITFULNESS is ranked above ONSET and NOCODA. Earlier in the paper some examples of English words (bring, play, camp, milk) were presented. The following examples prove English to be tolerating more than one consonant in onset and coda position, which means that English allows the violation of *COMPLEX. Then the question arises: are all sequences of consonants acceptable in word-initial and word-final positions? For example, can we use the word mikl instead of milk and still claim it to be a well-formed word of English? The answer is definitely no. Possible sequences of consonants at the edges of syllables is restricted by SONORITY. It requires the sequences of consonants demonstrate a proper sonority profile. The constraint is termed SONORITY constraint. Sievers (1881), Jespersen s (1904) observations led to the Sonority Sequencing Generalization requiring a sonority rise or plateau between any member of a syllable and the syllable peak. Then, in the onset position the sonority is supposed to rise, in the coda position the sonority profile is going downwards. As it is seen, ONSET is in the next-to-last position in the constraint hierarchy in English, thus possibly violable in the majority of cases. As it has been mentioned previously, the constraint hierarchy of the Lithuanian language has not been created yet, therefore, there is no evidence that ONSET constraint is also ranked low in the Lithuanian language. Accordingly, the tasks set for the contrastive analysis are as follows: 1. To determine violations of ONSET constraint in word-initial syllables of English and Lithuanian. 2. To compare the distribution of violations of ONSET constraint in word-initial syllables of the two languages: a) quantitative analysis; b) distributional analysis: establishment of the sequences that violate ONSET constraint and their sonority analysis. 3. To find out which of the two languages exhibits a stronger universal tendency. 57

Lingvistika / Linguistics ISSN 1392 8600 The method of the investigation and the obtained results are presented below. Method As linguistic material for analysis, passages from French Lieutenant s Woman by Fowles (1969) and Sodybų Tuštėjimo metas by Avyžius (1970) were chosen. All the words in the two passages were syllabified. The English words were also transcribed. The English syllabification was based on the view presented by Wells (1998): a syllable boundary is found wherever there is a word boundary, and also coincides with the morphological boundary between elements in a compound; affricates cannot be split; where the first two restrictions and the phonotactic constraint allow, consonants are syllabified with whichever of the two adjacent vowels is more strongly stressed, or, if they are equally stressed, with the leftward one. Degrees of stress, for this purpose, are ranked in decreasing order of strength, i.e. strong vowel, weak vowel. For example, the word happy is being analyzed as hæp. i. The Lithuanian syllabification was based on the functional syllable theory presented by Girdenis (1981). The theory claims that the syllable is to begin with the maximal consonant cluster, the model of which is allowed at the beginning of the word. Accordingly, e.g. the Lithuanian words mirksta, kalnas, vėtra were divided into syllables in the following way: mirk-sta, kal-nas, vė-tra. If the words were divided into syllables in some other way, it would either result in the failure to cover the maximal consonant cluster allowed at the beginning of the word or would violate the phonotactic constraints of the language. As the material for the contrastive analysis, 1397 word-initial syllables in English and 1397 word-initial syllables in Lithuanian were obtained. All the initial syllables were grouped into the following types: 1) the ones that have no onset, i.e. begin with a vowel; 2) the ones that have a simple onset, i.e. begin with a consonant; 3) the ones that have a complex onset, i.e. begin with a consonant cluster. The distributional analysis aiming at establishing the sequences that violate ONSET constraint and their sonority analysis was done, following Ladefoged s (1996, 222) sonority ranking for English and Blevin s (1996, 211) working universal scale. Results The quantitative analysis of violations of ONSET and *COMPLEX constraint in English and Lithuanian word-initial syllables revealed the following data. In English (table 1, chart 1) out of 1397 word-initial syllables, 321 syllables exhibit violations of ONSET constraint, i.e. begin with a vowel. 990 word-initial syllables have 58

ISSN 1392 8600 Lingvistika / Linguistics a simple onset, i.e. begin with a consonant. Out of 86 syllables demonstrating a complex onset, 80 cases appear to be two consonant long and 6 cases three consonant long. Table 1. Onset in English word-initial syllables The total number of word-initial No Onset Simple Onset Complex Onset syllables 80 (2 cons.) 1397 321 990 86 6 (3 cons.) 990 (71,79%) 80 (5,8%) 86 (6,24%) 321 (23,28%) 6 (0,44%) No Onset Simple Onset Complex Onset (2 cons.) Complex Onset (3 cons.) Chart 1. Onset in English word-initial syllables The Lithuanian language (table 2, chart 2), on the other hand, exhibits fewer instances of violations of ONSET constraint than the English language, thus has maximal claims to universality. Out of 1397 word-initial syllables analyzed there appeared 276 syllables beginning with a vowel, i.e. having no onset. 965 syllables have simple onset, i.e. begin with a consonant and 156 have a complex onset consisting of two-consonants. Word-initial syllables beginning with a consonant cluster consisting of three consonants did not surface in Lithuanian. Table 2. Onset in Lithuanian word-initial syllables The total number of word-initial syllables No Onset Simple Onset Complex onset 1397 276 965 156 (2 cons.) 59

Lingvistika / Linguistics ISSN 1392 8600 [VALUE] 156 (11,17%) [VALUE] 276 (19,76%) [VALUE] 965 (69,08%) No Onset Simple Onset Complex onset (2 cons.) Chart 2. Onset in Lithuanian word-initial syllables As it has been illustrated above, the results obtained for the two languages were also expressed in percentage. As has been noted, English has fewer claims to universality as it demonstrates more instances of violations of ONSET constraint than Lithuanian. 23.28% of the analysed English syllables have no onset, i.e. begin with a vowel, whereas in the Lithuanian sample 19.76% of the word-initial syllables have no onset, i.e. end with a vowel. Consequently, the percentage of the analysed syllables that have an onset is also different in the two languages: 80.24% of the word-initial syllables that have an onset in Lithuanian, while in English there are 78.03% of word-initial syllables that have an onset. Out of 80.24% Lithuanian word-initial syllables that have an onset, 69.08% exhibit instances of a simple onset, 11.17% of the analysed syllables begin with a consonant cluster, i.e. have a complex onset consisting of two consonants. Consonant clusters consisting of three consonant were not surfaced in the Lithuanian language. The analysis of the English language yielded different results. From the total amount of the word-initial syllables in the English sample, 71.79% have a simple onset, 5.80% begin with a consonant cluster consisting of two consonants. The ones that begin with a consonant cluster consisting of three consonants constitute 0.44%. This is shown graphically in chart 3. 60

ISSN 1392 8600 Lingvistika / Linguistics 1200 1000 [VALUE] 1121 (80,24%) [VALUE] 1076 (78,03%) 800 600 [VALUE] 276 400 (19,76%) 200 0 Lithuanian English No Onset 276 321 Axis Title Onset 1121 1076 Onset No Onset [VALUE] 321 (23,28%) Chart 3. Violations of ONSET constraint in word-initial syllables in English and Lithuanian The second task of the present investigation was the distributional analysis, aiming at establishing the sequences that violate ONSET constraint in English and Lithuanian as well as their sonority analysis. A complex onset, i.e. consonant clusters, appeared in 86 (6.24%) word-initial syllables in English. In Lithuanian 156 (11.17%) cases of a complex onset were found. Furthermore, in Lithuanian there were only two-consonant clusters, whereas in English out of 86 complex onsets there were 6 three-consonant clusters. The obtained results are explicitly shown in tables 3, 4, 5 below. Table 3. The two-consonant clusters in Lithuanian (156 in total). bl 4 2,56% br 7 4,49% fr 2 1,28% dr 4 2,56% dv 7 4,49% gn 1 0,64% gr 8 5,13% kl 2 1,28% kr 14 8,97% kv 6 3,85% pl 5 3,21% pr 37 23,72% 61

Lingvistika / Linguistics ISSN 1392 8600 tr 10 6,41% tv 4 2,56% sk 3 1,92% sl 1 0,64% sm 1 0,64% sn 1 0,64% sp 2 1,28% st 14 8,97% sv 7 4,49% šl 3 1,92% šn 1 0,64% šr 1 0,64% šv 1 0,64% zv 1 0,64% žm 5 3,21% žv 4 2,56% 156 100% Table 4. The two consonant clusters in English (86 in total) bl 3 3,75% br 3 3,75% dr 3 3,75% fr 6 7,50% fl 1 1,25% gl 3 3,75% gr 4 5,00% hj 1 1,25% kj 1 1,25% kl 4 5,00% kr 1 1,25% kw 1 1,25% pl 3 3,75% pr 10 12,50% sl 2 2,50% sm 6 7,50% sp 4 5,00% st 15 18,75% tr 6 7,50% vj 3 3,75% 80 100% 62

ISSN 1392 8600 Lingvistika / Linguistics Table 5. The three consonant clusters in English (6): str 3 50,00% skw 1 16,67% spr 2 33,33% 6 100,00% The obtained results of the investigation show that both English and Lithuanian allow more than a single consonant in the onset position, i.e. have complex onsets. As it has been mentioned previously in the article, in the onset the sonority profile is upward. The majority of the complex onsets in the samples of the two languages satisfy SONORITY constraint, i.e. demonstrate a proper sonority profile. However, there were 19 cases among the two-consonant clusters (80 in total) and all 6 cases among the three-consonant clusters in English obviously violating SONORITY constraint. For example, the fall in sonority from s to t or from s to p in steps, stare, spontaneous, spin, etc. is not satisfactory for an onset. All the three types of complex onsets demonstrating a fall in sonority that were found in English fail to satisfy SONORITY constraint as well, e.g. strange, spring, squall (skw-). Only 3 types of complex onsets demonstrating a fall in sonority were found in the Lithuanian sample. The complex onset consisting of the constrictive voiceless consonant s and the occlusive voiceless consonant k, e.g. skuduras, skubiai. Also the complex onset consisting of the constrictive voiceless consonant s and the occlusive voiceless consonant t, e.g. storos, stumtelėjo fail to satisfy SONORITY constraint in Lithuanian. The third type of the complex onset violating SONORITY constraint is a fall in sonority between the constrictive voiceless consonant s and the occlusive voiceless consonant p, e.g. spoksojo, spaudė. Thus, though there are fewer *COMPLEX ONSET violations in Lithuanian in general, these violations show a rather high percentage of sonority profile violations (12.17%). In English there are far more *COMPLEX ONSET violations. They are at the same time sonority profile violations in 23.75% of cases in two-consonant onsets. In three-consonant onsets sonority violations seem to be numerous. However, a greater amount of data is necessary to draw relevant conclusion. This issue, like numerous other issues that arose in the discussion, can be successfully resolved in further linguistic research. Conclusions 1. The results of the analysis of violations of ONSET constraint in word-initial syllable show that: a) The English language exhibits more cases of ONSET violations in word-initial syllable that the Lithuanian language. 63

Lingvistika / Linguistics ISSN 1392 8600 b) The distributions of consonants in English and Lithuanian onsets show that the majority of cases in English and Lithuanian satisfy SONORITY constraint, i.e. exhibit an appropriate sonority profile. 2. As the cases of ONSET violations in word-initial syllables of English are more frequent than in Lithuanian, in this respect the English language is more marked and the Lithuanian language expresses a stronger universal tendency. 3. Violations of ONSET constraint are tolerated in English due to the low ranking of the constraint in the constraint hierarchy of the English language. The constraint hierarchy of the Lithuanian language is not known and, thus the place of the constraint within the hierarchy is not defined. However, the analysis of its nature is a contribution to the future establishment of the constraint hierarchy in the Lithuanian language. References Archangeli, D. (1998). Optimality Theory: An Introduction to Linguistics in the 1990 s. In D. Archangeli, D. T. Langendoen (Eds.). Optimality Theory. An Overview (pp. 1-32). Oxford UK: Blackwell Publishers. Avyžius, J. (1970). Sodybų tuštėjimo metas. Vilnius: Vaga. Blevins, J. (1996). The Syllable in Phonological Theory. In J.A. Goldsmith (Ed.). The Handbook of Phonological Theory (pp. 206 244). Cambridge MA, Oxford UK: Blackwell Publishers. Clements, G.N. (1990). The Role of the Sonority Cycle in Core Syllabification. In J. Kingston, M. Beckman (Eds.). Papers in Laboratory Phonology I: Between the Grammar and Physics of Speech (pp. 283 333). Fowles, J. (1969). French Lieutenant s Woman. Toronto: Little Brown and Co. Girdenis, A. (1981). Fonologija. Vilnius: Mokslas. Hammond, M. (1998). Optimality Theory and Prosody. In D. Archangeli, D.T. Langendoen (Eds.). Optimality Theory. An Overview (pp. 33 58). Oxford UK: Blackwell Publishers. Jespersen, O. (1904). Lehrbuch der Phonetic. Leipzig and Berlin: B.G. Teubner. Ladefoged, P. (1982). A Course in Phonetics. New York: Harcourt Brace Jovanovich. McCarthy, J., Prince, A. (1993). Prosodic I: Constraint Interaction and Satisfaction. MS, University of Massachusetts, Amherst and Rutgers University. McCarthy, J., Prince A. (1996). Prosodic Morphology. In J. A. Goldsmith (Ed.). The Handbook of Phonological Theory. Oxford UK: Blackwell Publishers. Prince, A., Smolensky, P. (1991). Optimality. Paper presented at the Arizona Phonology Conference. Tucson: University of Arizona. Sievers, E. (1881). Grundzuge der Phonetic. Leipcig: Breitkopf and Hartel. Steponavičienė, S., Katinaitė, E. (2001). Violation of Nocoda Constraint in Word-initial Syllable (Analysis in English and Lithuania). Kalbotyra, 50 (3), 41 46. Wells, J.C. (1998). Longman Pronunciation Dictionary. UK: Longman Group Limited. 64

ISSN 1392 8600 Lingvistika / Linguistics Onseto apribojimo (angl. onset constraint) pažeidimai pirmajame žodžio skiemenyje (anglų ir lietuvių kalbų analizė) Edita Katinaitė-Kalčiūnienė Lietuvos edukologijos universitetas, Filologijos fakultetas, Anglų kalbos didaktikos katedra, Studentų g. 39, 08106 Vilnius, edita.kalciuniene@leu.lt Santrauka Straipsnyje pristatomas onseto tyrimas pirmajame anglų ir lietuvių kalbų žodžių skiemenyje (1397 lietuvių kalbos žodžių pirmieji skiemenys ir tiek pat anglų kalbos žodžių pirmųjų skiemenų). Tyrimas atliktas remiantis generatyvinės lingvistikos Optimalumo teorija, kuri atsirado XX a. pabaigoje kaip atsakas generatyvinei gramatikai, teigusiai, kad kalbos yra valdomos taisyklių. Optimalumo teorija atmeta taisykles, o vietoje jų pateikia universalius apribojimus ir nustato leistiną šių apribojimų pažeidžiamumą kalbose. Universalūs apribojimai pateikiami apribojimų hierarchijoje, kuri skirtingose kalbose taip pat yra skirtinga. Apribojimo pažeidžiamumas priklauso nuo jo vietos apribojimų hierarchijoje: žemiau esantis apribojimas gali būti pažeidžiamas, jeigu jis tenkina hierarchijoje aukščiau esantį apribojimą. Tyrimo metu nustatyta, kad lietuvių kalba pažeidžia universalų onseto principą mažiau nei anglų kalba (321 atvejis anglų kalboje ir 276 atvejai lietuvių kalboje). Šia prasme lietuvių kalba yra mažiau žymėta ir demonstruoja stiprias universalumo tendencijas. Anglų kalboje šis apribojimas užima žemą poziciją apribojimų hierarchijoje, todėl yra toleruojamas. Kadangi lietuvių kalboje onseto apribojimo pažeidimo atvejų yra mažiau nei anglų kalboje, universalus onseto apribojimas gali pasirodyti esąs aukštesnėje apribojimų hierarchijos pozicijoje lietuvių kalboje nei anglų kalboje. Minėta hierarchija lietuvių kalboje nėra nustatyta, todėl šis gretinamasis tyrimas gali būti vertinamas kaip indėlis į jos sudarymą. Esminiai žodžiai: Optimalumo Teorija, onsetas, skiemuo, suvaržymų hierarchija, sonoriškumas. Pateikta / Submitted 2015 10 28 Priimta / Accepted 2015 11 11 65