Manner assimilation in Uyghur Suyeon Yun (suyeon@mit.edu) 10th Workshop on Altaic Formal Linguistics (1) Possible patterns of manner assimilation in nasal-liquid sequences (a) Regressive assimilation lateralization: /NL/ [LL] (b) Progressive assimilation nasalization: /NL/ [NN] (2) This talk (a) New data on Uyghur, which show both regressive assimilation in (1a) and progressive assimilation (1b) (b) Formal analysis within Optimality Theory (OT; Prince and Smolensky, 2004), adopting the P-map hypothesis (Steriade, 2008) and partial ranking (Anttila, 1997) (c) Alternative analyses: (i) SyllableContact (Baertsch and Davis, 2001), (ii) Articulatory Phonology (Browman and Goldstein, 1989, 1992) Data (3) Nasal-liquid clusters in Uyghur: nasal-final stem + /l/-initial suffix (e.g., -lar pl., -liri your (sg.)) cf. It seems that there is no stem-internal nasal-liquid sequence. (4) Previous description in Hahn (1991): /nl/ [ll] The alveo-dental nasal /n/ assimilates to an immediately following /l/. (p. 87) qalganlar [... ll... ] (the) remaining ones kælgænligniz [... ll... ] your having come, your coming (5) Recording Speaker: a male native speaker of Uyghur from Shinjang Target items: 31 nasal-final nouns with the plural suffix -lar Procedure (a) Word in isolation: How do you say XXX in Uyghur? (b) Plural -lar form in normal speed speech: How do you say XXXs (plural) in Uyghur? (c) Plural -lar form in slow speech: Could you say it slowly? (d) Plural -lar form in fast speech: Could you say it very fast? Recording took place in a quiet room using Marantz PMD660 and a head-mounted microphone Shure SM10A (6) Result 1: No assimilation in the sequence of a non-nasal consonant and /l/ final C r /bor-lar/ [borlar] chalk (pl.) b /kitab-lar/ [kitablar] book (pl.) t /karat-lar/ [karatlar] bed (pl.) k /isik-lar/ [isiklar] door (pl.) q /qulaq-lar/ [qulaqlar] ear (pl.) S /qus-lar/ [quslar] bird (pl.) ts /atsquts-lar/ [atsqutslar] key (pl.) z /qiz-lar/ [qizlar] girl (pl.) s /dærs-lar/ [dærslar] class (pl.) G /bag-lar/ [baglar] orchard (pl.) x /dæræx-lar/ [dæræxlær] tree (pl.) My biggest thanks should go to my consultant, Mettursun Beydulla, who kindly shared his native language with me. This work benefited from discussion with Michelle Fullwood, Michael Kenstowicz, Donca Steriade, and especially Adam Albright and Edward Flemming, who I am really grateful to. 1
(7) Result 2: Optional manner assimilation in nasal-liquid sequences (a) /nl/: Optional lateralization, e.g., /dukan-lar/ [dukanlar] [dukallar] (b) /ml/: Optional nasalization, e.g., /Sam-lAr/ [Samlar] [Samnar] (c) /Nl/: No deletion but shortening of the /l/, e.g., /tson-lar/ [tsonlar] adults (2 items only; won t be considered) (8) Lateralization or nasalization occurs most in fast speech, least in slow speech candles shops final C # of items speed normal slow fast n 18 nl: 14 (78%) ll: 4 (22%) nl: 15 (83%) ll: 3 (17%) nl: 6 (33%) ll: 12 (67%) m 11 ml: 8 (73%) mn: 3 (27%) ml: 10 (91%) mn: 1 (9%) ml: 6 (55%) mn: 5 (45%) 0 20 40 60 80 100 /nl/ normal slow fast speech rate Figure 1: dark: [nl] / light: [ll] 0 20 40 60 80 100 /ml/ normal slow fast speech rate Figure 2: dark: [ml] / light: [mn] Lateralization in /nl/ sequences is not obligatory but optional, unlike the previous description There is also optional nasalization in /ml/ Both types of assimilation are more likely to occur in fast speech than in normal or slow speech (9) Younger generation Korean speakers pronunciation of English (loan) words: show the same lateralization/nasalization patterns as in Uyghur (a) /nl/ [nl] [ll] green light [k1rin lait h 1] [k1ril lait h 1] listen, listen, listen [lis1n lis1n lis1n] [lis1l lis1l lis1n] (b) /ml/ [ml] [mn] dream live (concert) [t1rim laip1] [t1rim naip1] SM life [esem laip h 1] [esem naip h 1] (10) Cross-linguistically, there are more languages that undergo lateralization in /nl/ than languages that undergo nasalization In the languages reported in Seo (2003): (a) /nl/ [ll]: Klamath, Ponapean, Toba Batak, Moroccan Arabic, Leti, Uyghur (b) /nl/ [nn]: Tatar, Yakut (11) What to explain (a) Asymmetry in place of articulation Coronal nasal /n/ is the target of regressive manner assimilation, i.e., lateralization Labial nasal /m/ is the trigger of progressive manner assimilation, i.e., nasalization (b) Optionality Both types of the manner assimilation are not obligatory, but the clusters in question may remain intact WAFL 10 May 2, 2014 2
Analysis (12) Markedness (a) Agree(lateral): *[+αlateral][ αlateral] Penalizes [lateral] disagreement in homorganic nasal-liquid sequences cf. ICC [lat]: A sequence of consonants must be identical in laterality (Pulleyblank, 1997) (b) Agree(nasal): *[+αnasal][ αnasal] Penalizes [nasal] disagreement in homorganic/heterorganic nasal-liquid sequences cf. ICC [nas]: A sequence of consonants must be identical in nasality (Pulleyblank, 1997) (c) Feature specification [lateral] [nasal] [sonorant] (13) Faithfulness l + + n + + m 0 + + t Max(feature) constraints (a) Max(+lateral): No deletion of [+lateral] (b) Max(+nasal): No deletion of [+nasal] Subdivision of Max(+lateral): Max(+lateral)/ C and Max(+lateral)/ V Assumption: The duration of [l] is longer postvocalically (... V C(V)... ) than prevocalically (... C V... ) Losing a postvocalic & preconsonantal [l] will cause a perceptually greater change than losing a prevocalic & postconsonantal [l] (cf. the P-map hypothesis; Steriade, 2008) Max(+lateral)/ C Max(+lateral)/ V (a) Max(+lateral)/ C: No deletion of [+lateral] preceding a consonant (b) Max(+lateral)/ V: No deletion of [+lateral] preceding a consonant More on prevocalic & postconsonantal [l]: Max(+lateral)/n V and Max(+lateral)/m V Assumption: The duration of [l] is longer when following [n] than when following [m] Losing a post-[n] [l] will cause a perceptually greater change than losing a post-[m] [l] Max(+lateral)/n V Max(+lateral)/m V (a) Max(+lateral)/n V: No deletion of [+lateral] in n V (b) Max(+lateral)/m V: No deletion of [+lateral] in m V Ident(place): No featural change in place of articulation Ident(sonorant): No featural change in [sonorant] blocks changes from an obstruent to a sonorant, and vice versa (14) Crucial rankings: Id(sonorant), Id(place) Max(+lateral)/n V Max(+lateral)/m V, Max(+nasal), Agree(lateral), Agree(nasal) Highly ranked Max(+lateral)/n V penalizes /nl/ [nn] but not /ml/ to [mn] Partial ranking (Anttila, 1997): Some constraints rankings are fixed, but others vary Id(son), Id(place) Max(+lat)/n V { Max(+lat)/m V, Max(+nas), Agree(lateral), Agree(nasal) } fixed varying (15) Optional lateralization in /nl/ Input: /nl/ Max(+lat)/n V Agree(lat) Agree(nas) Max(+lat)/m V Max(+nas) a. nl c. nn! WAFL 10 May 2, 2014 3
(16) Optional nasalization in /ml/ Input: /ml/ Id(place) Max(+lat)/n V Agree(lat) Agree(nas) Max(+lat)/m V Max(+nas) a. ml b. ll! c. mn (17) Prediction: /ln/ [ln] [ll] Uyghur: No /ln/ sequences Younger Koreans English: /ln/ [ln] [ll] Input: /ln/ Max(+lat)/ C Max(+lat)/n V Agree(lat) Agree(nas) Max(+nas) a. ln c. nn! (18) Prediction: /lm/ [lm] Uyghur: /lm/ [lm] Younger Koreans English: /lm/ [lm] Input: /lm/ Id(place) Max(+lat)/ C Agree(nas) Max(+lat)/m V Max(+nas) a. lm b. ll! c. nm! (19) Summary (a) Preference for lateralization in homorganic /nl/ sequences: Max(+lateral)/n V Max(+lateral)/m V (b) Nasalization in heterorganic /ml/ sequences: Preservation of the place of articulation of the labial nasal (c) Optionality: Partial ranking between Max(+lat)/m V, Max(+nas), Agree(lateral), Agree(nasal) Alternative 1: SyllableContact analysis (20) Traditional analysis of the lateralization and nasalization using SyllableContact (SyllCon) The Syllable Contact Law 1 (Davis and Shin, 1999): A syllable contact A.B is the more preferred, the greater the sonority of the offset A and the less the sonority of the onset B. = Nasal-liquid sequences are dispreferred because the sonority of the onset liquid is higher than the coda nasal in those sequences SyllCon: Avoid rising sonority over a syllable boundary. = Penalizes nasal-liquid sequences on surface (21) Constraint ranking for /nl/ [ll] in Uyghur (Baertsch and Davis, 2001): FaithOnset SyllCon FaithStem (22) Applying Baertsch and Davis s (2001) constraints to the current data /nl/ [nl] [ll]: requires free ranking between SyllCon and FaithStem Input: /nl/ FaithOnset SyllCon FaithStem a. nl c. nn! 1 This is Davis and Shin s (1999) translation of the definition of the Syllable Contact Law in Vennemann (1988). WAFL 10 May 2, 2014 4
/ml/ [ml] [mn]: requires free ranking between FaithOnset and SyllCon Input: /ml/ Idplace FaithOnset SyllCon FaithStem a. ml b. mn c. ll! (23) Unwelcome results: FaithOnset SyllCon FaithStem equally ranked, and FaithOnset penalizes both /nl/ [nn] and /ml/ [mn] Expected: /nl/ [nl] [ll] *[nn] Input: /nl/ FaithOnset SyllCon FaithStem a. nl c. nn (24) No evidence for syllabification that may support the SyllableContact analysis Changes in nasal-liquid clusters in Uyghur, with shallow sonority rise, are triggered by SyllableContact? Obstruent-liquid clusters in Uyghur in (6), with steep sonority rise, remain intact Possible argument for SyllableContact: Syllabification differs /VNLV/ [VN.LV] (violating SyllCon) vs. /VTLV/ [V.TLV] (not violating SyllCon)... there is no evidence for different syllabifications in Uyghur (25) Need to assume another markedness constraint to explain manner assimilation in /ln/ sequences Alternative 2: A gestural approach (26) Articulatory Phonology (Browman and Goldstein, 1989, 1992) The basic units of production, perception, and mental representation of speech are one and the same These units are coordinated articulatory gestures (27) Nasals: labial ([m]) or tongue tip ([n]) closure + velum widening (28) Laterals: tongue tip narrowing + tongue body widening (cf. Proctor, 2009) (29) Lateralization in /nl/: Extension of the TB gesture /nl/ [nl] [ll] (30) Nasalization in /ml/: Extension of the VEL gesture /ml/ [ml] [mn] WAFL 10 May 2, 2014 5
(31) May explain what is happening, e.g., /nl/ [ll], but not what is not happening, e.g., /nl/ [nn] Why no velum gesture extension in /nl/? Conclusion (32) Summary In Uyghur, homorganic /nl/ sequences undergo regressive lateralization and heterorganic /ml/ sequences undergo progressive nasalization... and these occur optionally Perceptually-driven constraint ranking + partial ranking can account for the current data References Anttila, A. (1997). Variation in Finnish phonology and morphology. PhD thesis, Stanford University. Baertsch, K. and Davis, S. (2001). Turkic C+/l/(uster) phonology. In CLS 37: The Main Session. Papers from the 37th Meeting of the Chicago Linguistic Society, pages 29 43. Browman, C. P. and Goldstein, L. M. (1989). Articulatory gestures as phonological units. Phonology, 6:201 251. Browman, C. P. and Goldstein, L. M. (1992). Articulatory phonology: An overview. Phonetica, 49:155 180. Davis, S. and Shin, S.-H. (1999). The syllable contact constraint in Korean: An optimality-theoretic analysis. Journal of East Asian Linguistics, 8:285 312. Hahn, R. F. (1991). Spoken Uyghur. University of Washington Press, Seattle. Prince, A. and Smolensky, P. (2004). Optimality Theory: Constraint interaction in generative grammar. Blackwell, Malden, Mass. & Oxford. Proctor, M. I. (2009). Gestural Characterization of a Phonological Class: the Liquids. PhD thesis, Yale University. Pulleyblank, D. (1997). Optimality theory and features. In Archangeli, D. and Langendoen, D. T., editors, Optimality Theory: The Overview, pages 59 101. Blackwell, Oxford. Seo, M. (2003). A segment contact account of the patterning of sonorants in consonant clusters. PhD thesis, The Ohio State University. Steriade, D. (2008). The phonology of perceptibility effects: the P-Map and its consequences for constraint organization. In Hanson, K. and Inkelas, S., editors, The Nature of the Word: Studies in Honor of Paul Kiparsky, pages 151 179. MIT Press, Cambridge. Vennemann, T. (1988). Preference Laws for Syllable Structure. Mouton de Gruyter, Berlin/New York. WAFL 10 May 2, 2014 6