Loading wiki pages...

Wiki Version:
*Introduction*. Long-distance dependencies like vowel harmony are especially complex when blockers are present (Heinz, 2010), with justifications that suggest variable or optional long-distance dependencies should be particularly difficult to acquire. This is consistent with claims by Poliquin (2006), who doubts that Laurentian French high-vowel (HV) laxing harmony is learnable because the target and iterativity parameters of harmony cannot be determined in disyllabic words and those words comprise the vast majority of the potentially harmonizing words in the French lexicon. The Laurentian French case is further complicated by laxing disharmony in a subset of disyllabic words and by opacity from retensing (e.g. Bosworth, 2011; Dumas, 1987; Fast, 2008; and Poliquin, 2006). The following illustrates the processes affecting the tenseness (tense/lax quality) of HV that are of interest for the current analysis: (1) Closed-syllable laxing of HV (final syllables, obligatory) [vi] *vie* ‘life’ [vɪt] *vite* ‘quick’ (2) Retensing of HV before /v(ʁ) z ʒ/ (final syllables, obligatory) [viːz] *vise* ‘target.3sg’ [viːv(ʁ)] *vivre* ‘live.inf’ (3) Closed-syllable laxing of HV (non-final syllables, optional) [mitɛn] *mitaine* ‘mitten’ [mistɛʁ]~[mɪstɛʁ] *m**ystère* ‘mystery’ (4) Retensing of HV from coda resyllabification (word-internal resyllabification, obligatory) [vɪt] *vite * ‘quick’ [vitɛs] *vite * ‘speed’ (5) HV laxing disharmony in disyllables (C0[V +high]C0[V +high] words, optional) [midi]~[mɪdi] *midi* ‘noon’ (6) HV laxing harmony, with speakers having different harmony patterns (Dumas, 1987; Poliquin, 2006) (non-final syllables, all optional) a. Initial-targeting non-local harmony [vizɪt]~[vɪzɪt] *visite* ‘visit’ [difisɪl]~[dɪfisɪl] *difficile* ‘difficult’ [inedɪt]~[ɪnedɪt] *in**édite* ‘unpublished.fem’ b. Local non-iterative harmony [vizɪt]~[vɪzɪt] *visite * ‘visit’ [difisɪl]~[difɪsɪl] *difficile* ‘difficult’ [inedɪt] *inédite* ‘unpublished.fem’ c. Local iterative harmony (opaque non-high) [vizɪt]~[vɪzɪt] *visite* ‘visit’ [difisɪl]~[dɪfɪsɪl] *difficile* ‘difficult’ [inedɪt] *inedited* ‘unpublished.fem’ d. Local iterative harmony (transparent non-high) [vizɪt]~[vɪzɪt] *visite* ‘visit’ [difisɪl]~[dɪfɪsɪl] *difficile* ‘difficult’ [inedɪt]~[ɪnedɪt] *in**édite* ‘unpublished.fem’ Learners must navigate complex interactions between (often optional) processes and are consequently faced with unclear motivations for the tenseness of a HV in forms they encounter. The current study leverages automated classification using forced alignment to offer the first large-scale study of HV tenseness in Laurentian French drawing on production data. We first test the community-level grammar that a learner is expected to acquire (using mixed-effects logistic regression to simulate learning with adaptation to individual speakers). We then probe individual speakers’ grammars to test for the harmony patterns that speakers acquire given that community-wide patterns may obscure important individual differences (Blaxter et al., 2019). Our results demonstrate that learners who are faced with variability and optionality in long-distance dependencies generate distinct grammars, and certain speakers generate grammars that resolve the uncertainty by ascribing tenseness to non-(dis)harmonic sources. *Methods*. A forced aligner (Milne, 2014) was trained to distinguish tense and lax realizations of high vowels in final syllables (where tenseness is categorically predictable), and then classified the tenseness of high vowels in non-final syllables (Milne & Lamontagne, 2016). The conversational corpus consists of 131 speakers producing over 24000 high vowels in non-final syllables. We generated both community and individual grammars using decision trees and mixed-effects logistic regression, which approximates Maximum Entropy Optimality Theory (Goldwater & Johnson, 2003) in interpretation. *Results*. We find suggestive evidence consistent with community-wide preferences for closed-syllable laxing (3). Our results additionally confirm that speakers would be expected to acquire laxing harmony, both initial-targeting non-local harmony (6a) and local iterative harmony that treats neutral vowels as transparent (6d). Laxing disharmony (5) and retensing from resyllabification (4), however, are not supported at the community level. Probing individual speakers’ grammars where sufficient tokens are present, we find that they are distinct and reflect wider variation that those assumed by Poliquin (2006) to emerge from innate assumptions due to the poverty of the stimulus. Not only do some speakers appear to have more than one type of harmony (6a-6d) simultaneously (and not necessarily triggered solely by final syllables), but some speakers produce tense/lax variation in non-final syllables motivated by other factors, notably (a) a preference to lax initial syllables independent of harmony, (b) a preference to lax syllables based on distance from word edges (odd-even asymmetry), and (c) long-distance laxing disharmony. *Discussion*. Speakers of Laurentian French infer different grammars from highly variable input of long-distance dependencies, even when there are detectable patterns in that input across speakers. Many speakers do generate grammars with HV harmony (including grammars with derivational opacity), but some speakers resort to other mechanisms to condition the tenseness of high vowels. Crucially, speakers may generate alternative grammars that treat laxing as centralization or reduction in prosodically weak syllables, as initial-syllable laxing without a harmonic trigger, or as (long-distance) disharmony. The presence of centralization and non-disharmonic laxing suggests that speakers could treat tenseness in high vowels as parallel to height in mid-vowels (/e-ɛ ø-œ o-ɔ/) given similar processes (e.g. Dumas, 1987), indicating the potential for a change in the representation of phonological contrasts. The generation of long-distance disharmony is of particular interest because it reflects the conclusion by Heinz (2010) that long-distance dissimilation may be a form of long-distance phonotactics with blocking based on assumptions about the computational complexity of phonology. Speakers also ascribe unexpected tenseness to lexical factors: many words exhibit idiosyncratic behavior. This suggests that these words are learned with specific tenseness values independent of the generalizations found, which is consistent with the possibility of ongoing phonologization of high-vowel tenseness in the variety (e.g. Lamontagne, 2019) or with the generation of detailed phonetic representations for frequent words (as in Exemplar Theory; e.g. Johnson, 1997). *References*. *Blaxter, T., K. Beeching, R. Coates, J. Murphy & E. Robinson. (2019). *Each p[ɚ]son does it th[εː] way: Rhoticity variation and the community grammar. *Language Variation and Change* 31(1), 91-117. *Bosworth, Y. (2011).* *Weight and feet in Québécois*. PhD thesis, University of Texas at Austin. *Dumas, D. (1987).* *Nos façons de parler: Les prononciations en français québécois*. Québec: Presses de l’Université de Québec. *Fast, A. (2008).* Optimality and opacity in Canadian French vowel harmony: A variationist account. *Revue des étudiants en linguistique du Québec* 3(1). *Goldwater, S. & M. Johnson. (2003).* Learning OT constraint rankingsusing a maximum entropy model. In J. Spenader, A. Eriksson, and O. Dahl, (eds.), *Proceedings of the Stockholm Workshop on Variation within Optimality Theory*. Stockholm University, 111–120. *Heinz, J. (2010).* Learning Long-Distance Phonotactics. *Linguistic Inquiry* 41 (4): 623–661. *Johnson, K. (1997).* Speech perception without speaker normalization: an exemplar model. In K. Johnson and J.W. Mullennix (eds.) Talker Variability in Speech Processing. San Diego: Academic Press (pp. 145-166). *Lamontagne, J. (2019).* “Acoustic evidence of phonemicization: Lax high vowels in Quebec French”. Linguistic Society of America Annual Meeting in New York City, January 3rd – 6th, New York City (USA). *Milne, P. (2014).* *The variable pronunciations of word-final consonant clusters in a force aligned corpus of spoken French*. PhD thesis, University of Ottawa. *Poliquin, G. (2006).* *Canadian french vowel harmony*. PhD thesis, Harvard University.
OSF does not support the use of Internet Explorer. For optimal performance, please switch to another browser.
This website relies on cookies to help provide a better user experience. By clicking Accept or continuing to use the site, you agree. For more information, see our Privacy Policy and information on cookie use.

Start managing your projects on the OSF today.

Free and easy to use, the Open Science Framework supports the entire research lifecycle: planning, execution, reporting, archiving, and discovery.