Atelier de phonologie




Bienvenue ! Welcome!
L'Atelier de phonologie est un séminaire de recherche informel consacré à la phonologie : représentation, computation, acquisition et modalités d'apprentissage, interaction avec les autres modules linguistiques, etc. Y sont présentés des travaux en cours, des préparations de conférences, des revues de la littérature, ainsi que des présentations plus accomplies. Le séminaire, hébergé par le laboratoire SFL (Structures Formelles du Langage), est ouvert à la communauté de phonologues locale et internationale. Il se réunit environ toutes les deux semaines, le mercredi matin de 10h à midi, et est suivi d'un déjeuner sur place. Les archives des réunions passées (depuis l'automne 2012) est disponible ici.


The Atelier de phonologie is an informal research seminar dedicated to phonology: its representations, its computations, its acquisition and learnability, its interaction with other linguistic modules, etcetera. The seminar features informal presentations of work in progress, practice talks, literature reviews, as well as more polished talks. It is open to the local community as well as to international visitors. It is hosted at the SFL lab and it meets approximately every two weeks, always on a Wednesday morning, followed by lunch. The archive of past meetings (since fall 2012) is available here.





Où et quand / When and where


Où - Where:
59/61, rue Pouchet, Paris (cartes et informations sur les transports : here you can find a map and information concerning public transportation). 
Quand - When: 

le mercredi, de 10h à 12h. On Wednesdays, from 10am to noon.






Calendrier / Schedule 2017-2018







17 janvier
salle 124












Carlo Geraci
(
Institut Jean Nicod, Département d’études cognitives, ENS, EHESS, CNRS, PSL Research University, Paris)

Typological and Historical Relations Across Sign Languages: evidence from phonology
 
Résumé - This project develops a theoretically-informed coding schema (and online tool) to annotate articulatory features of sign languages. We apply these tools to 24 sign languages and compute over their featural properties to identify micro- and macro-language families. Thus, we provide proof of concept that quantitative methods based on phonetic/phonemic information can be used to probe typological and historical classifications of sign languages, along the lines of what has been done recently in spoken language phylogenetics (e.g., Dunn et al. 2005). These results can be combined with the (often impoverished) historical records of sign language genesis to evaluate relations across sign languages and concretely identify points of cross-linguistic variation in sign languages.

24 janvier
salle 108



















7 février
Salle 159



















21 Février
Salle 159
















7 mars
Salle 159


















21 mars
Salle 159


















4 avril
Salle 159






















18 avril
salle 159






















25 avril
salle 159













16 mai
salle 108



























30 mai
salle 159


























6 juin
salle 159























13 juin
salle 124














Edoardo Cavirani 
(Meertens)

GENDER and NUMBER spell-out on NPs. The case of Lunigiana dialects
 
Résumé - The spell-out sequence of the F and PL morphosyntactic features on the NPs of a set of Lunigiana dialects (IT), seems to point at a violation of the Mirror Principle (Baker 1985). Indeed, while the gender exponent generally precedes the exponent expressing number in Romance (e.g. Spanish lob–ROOT –oM –sPL ‘wolves’), in the dialects under scrutiny the phonological exponent for number occurs between the root and the exponent for gender (e.g. Colonnatese don–ROOT –jPL –aF ‘women’). If gender is encoded in n, a categorizing head (cf. Marantz 2001), then gender morphemes should be overtly expressed at the right of the root, as is the case of Spanish (e.g. [√ [GENDER]]). Number (henceforth #), by merging above n, should be spelled out at the right of the [√ + GENDER] complex (e.g. [# [√ + GENDER]]). I argue that the spell-out strategies of F and PL attested for F.PL nouns in Lunigiana dialects, as well as the observed microvariation, hinge upon post-syntactic, mainly phonological, requirements. This account builds on the hypothesis that |A|F and |I|PL are floating elements and that n, which has been suggested to encode gender, is spelled out by either an empty CV structure (cf. Lowenstamm 2008; Cavirani & van Oostendorp 2017; see also Bendjaballah 2014) or by a completely empty morpheme.



Noa Bassel Si Berrebi
(Tel Aviv University)

Pharyngeal minds: variation in the underlying representation of Hebrew speech (that is not reflected in production)

Résumé - We report a lexical decision task experiment performed by two groups of third generation Hebrew speakers: a group of Yemenite descent Israelis, and a group of European descent Israelis. Our participants speak similar dialects of Modern Hebrew, but were exposed to different dialects during acquisition, and our goal is to show whether and how this variation in input affected their mental phonological inventory.
We manipulated Hebrew words such that two different sounds had been switched one with the other: [ħ] and [x]. The former is a marked sound of Mizrahi dialects of Modern Hebrew, including the Yemenite dialect; The latter is its common counterpart. The two sounds originate from different phonemes, and are represented by different graphemes in the orthographic system, but in the production of most Hebrew speakers today they are phonetically indistinguishable.
All our participants do not produce the sound [ħ] in their speech, but half of them were exposed to it during acquisition. The manipulated words are incompatible with the inputs that both types of speakers were exposed to during language acquisition, therefore the significant differences we find in their reactions to the manipulated items can be attributed to their mental representation of these sounds.



Johanna Benz

(Universität Leipzig)

Phonologically conditioned affix order

Résumé - Phonologically conditioned affix order (PCAO) is a controversial phenomenon. Paster (2006, 2009) has argued that all putative cases reduce to segmental metathesis or infixation and that “truely” phonologically conditioned affix order does not exist. The theoretical significance of this claim is clear: much like phonologically conditioned allomorphy, PCAO adds to our understanding of the (un-)availability of phonological information to morphology. In this talk, I argue that Washo affix order is phonologically conditioned, thus providing a counterexample to Paster’s claim. I show that affixes in Washo are reordered to avoid a stem-final stressed syllable. In Stratal Optimality Theory (Kiparsky 2000), the pattern results from an interaction of phonological constraints such as NonFinality and *Clash and morphological alignment constraints. Alternative approaches (particularly subcategorization as proposed by Paster) as well as theoretical implications of “Stratal P>>M” (where phonological constraints outrank morphological constraints within a single module which is not strictly parallel) are discussed.
 

Matilde Accattoli
(université Paris 8)

How the unmarked emerges from analogy: Retention vs. deletion of desinential yod in Proto-Italo-Romance

It is well known that in Vulgar Latin the first vowel of a hiatus lost its syllabicity, becoming a glide. I will focus on yod, whose effects on the preceding consonant (gemination, palatalization/affrication) are regularly attested in Old Italian (as in VĪNĔA > *vinja > viɲɲa ‘vineyard’), except in verbal inflection. In verbs, yod effects are sporadic, since many verbs lost their desinential yod in a pre-documentary stage, shifting to the yod-less inflectional class (sometimes just partially, which led to heteroclitic paradigms). Comparing Classical Latin and Old Italian data, I will show that the fate of yod in verbs depended largely on the preceding syllable: yod retention was preferred after a light syllable (1SG FĂCĬO > *fakjo > fatʧo ‘I do’), whereas yod deletion (= class shift) was preferred after a heavy one (1SG PARTĬO > parto ‘I leave’, RĪDĔO > rido ‘I laugh’). Starting from a proto-Romance C$j syllabification (Pensado Ruiz 1988, Vennemann 1988), I will analyze the class shift as a phonologically grounded analogical change, exploiting the available yod-less inflectional pattern to avoid the emergence of marked outputs as (C)VCC.YV and (C)V:C.YV. Therefore, analogy (or O-O correspondences) contributed to the “emergence of the unmarked” (TETU effects, McCarthy and Prince 1994). Alternative analyses will be considered as well.


Xiaoxi Liu

(University of Essex)

Depressor effects: a recursive multi-­layered geometry

Résumé -­ Observations on depressor effects in Bantu, Khoisan and Chinese Wu show that (i) a nasality and voicing asymmetry exists in depression;; (ii) voicing is necessary but not sufficient for depression; and (iii) the depressed L-­tone in African and Chinese are phonologically different. Classical feature and constraints analyses are less able to explain the varieties of depressors. Nor is the difference between African and Chinese depressor effects phonologically studied. In this talk, I propose an alternative geometry analysis by extending the geometry structures in Clements (1985), Kula (2012) and RCVP (van der Hulst 2005), and aim to explain the problem of nasality-­voicing asymmetry through a hierarchical recursive representation by positioning voicing (|L|-­element) hierarchically above nasality
(|L|-­element). This asymmetry is also additionally supported by evidence from Beijing Mandarin final nasal loss in falling tone syllables. I show that the Laryngeal structure of the multi-­layered geometry is able to explain the diversity of depressors. The multi-­layered recursive structure will also be supported
by derived environment effects.



Discussion

Qui a le choix et quand? Allomorphie et architecture grammaticale

Cette semaine à l’atelier nous discuterons la notion d’allomorphie d’un point de vue théorique. Nous définirons l’allomorphie comme « deux représentations sous-jacentes en distribution complémentaire exprimant le même concept ou ayant la même fonction grammaticale ». Nous prendrons comme point de départ la proposition suivante : il existe des cas d’allomorphie dans lesquels le choix entre les deux allomorphes est optimisant, c’est-à-dire que l’allomorphe choisi est phonologiquement meilleur que l’allomorphe délaissé. Étant donné ce point de départ, la question se pose concernant l’étape dérivationnelle dans laquelle l’allomorphe est choisi. Est-ce au même moment, dans le même calcul, où sont évalués les autres aspects de la phonologie, comme le veut la version classique de l’OT? Est-il est possible d’envisager une phonologie à étapes, où d’abord on choisit l’allomorphe, et puis on évalue le reste de la phonologie à l’étape suivante (et si oui, pourquoi dans cette ordre, et pas l’inverse)? Dans un système d’épellation (« spell-out ») comme celui de la Morphologie Distribuée, beaucoup de cas d’allomorphie sont décrits comme ayant lieu lors de cette épellation. Il semble être architecturellement problématique de permettre à ce mécanisme de considérer l’améliorisation phonologique de la forme, car il précède la phonologie. Or, cela semble être le cas dans les effets de PCO allomorphique et l’haplologie (le français *parler de des amis, l’anglais boss’s ‘patron.gen’, bosses’ ‘patrons.gen’ *bosses’s, ou l’espagnol darselo ‘le lui donner’, *darlelo).

    Cette session prendre la forme d’une discussion. Des cas empiriques à considérer seront préparés par Noam Faust, qui animera la session. Nous écouterons également Francesc Josep Torres-Tamarit sur une proposition qu’il est en train de préparer avec Peter Jurgec, qui utilise l’allomorphie pour rendre compte de l’opacité.


Competing views on Stress

1) Francesc Torres-Tamarit (SFL/Université Paris 8)

This talk is part of a discussion session with Guillaume Enguehard (LLL/Université d'Orléans). Its main objectives are (i) to debunk myths about OT, (ii) to introduce the basics of Metrical Phonology (what is stress and how it should be represented), and (iii) to motivate the need for constituency (e.g. feet) in the light of some puzzling metrical phenomena. I will also present an OT analysis of stress assignment in Munster Irish based on internally layered ternary feet and show why OT is not only compatible but also preferable in dealing with Enguehard's (to appear) substance-free analysis of vowel reduction in Russian. 

2) Guillaume Enguehard (LLL/Université d'Orléans) 

This talk is part of a discussion session with Francesc Torres-Tamarit (CNRS/SFL). Its main objectives are (i) to debunk myths about Hjelmslev, (ii) to introduce the basics of structuralism (what is a prosodeme and how it should be represented), and (iii) to deduce the representation of prominence (e.g. extra space) from the definition of distinctive units. I will also present a structural analysis of vowel reduction in Russian and show why a structuralist approach is not only compatible but also preferable in dealing with Torres-Tamarit & Hermans (2017) stress assignment in Munster Irish.



Flore Picard

(Université Paris 3)

Quelques problématiques phonologiques dans l'étude des langues sames

Les langues sames sont une famille de neuf langues finno-ougriennes parlée en Laponie. Elles présentes quelques phénomènes phonologiques et morphophonologiques typologiquement rares et intéressants, comme la présence d'un triple contraste de quantités consonantiques. Si plusieurs de ces langues ont individuellement été le sujet de descriptions et de recherches en phonologie, il manque aujourd'hui une analyse diasystémique qui permettrait de mettre en relation des phénomènes souvent traités comme distincts dans chacune de ces langues. Nous proposons de passer en revue quelques unes des problématiques phonologiques des langues sames et leur traitement dans la littérature, pour faire un premier pas vers une analyse plus unifiée.


Benjamin Storme 

(Université Paris 8, CNRS SFL) 

Phonologically-conditioned cyclicity in Standard French

The phonology of derivatives can be regular (i.e. it obeys the language’s phonotactics) or cyclic (i.e. the derivative bears resemblance to its derivational base beyond what is predicted by the language’s phonotactics). This talk reports preliminary evidence for a pattern of phonologically-conditioned cyclicity in Standard French (SF), where the realization of a mid vowel as close-mid or open-mid in the final syllable of a derivative’s stem (e.g. ê in fêt-ard ‘partier’ from fête ‘party’) is cyclic when a schwa can or must occur between the stem and the suffix (e.g. coqu-(e)-let [kɔk(ə)le] ‘hen-DIM ’ from coq [kɔk] ‘hen’) but regular otherwise (e.g. fêt-(*e)-ard [fetar] ‘partier’ from fête [fɛt] ‘party’). The evidence is preliminary because only the data of 4 among the 10 participants who participated in the acoustic study (from Paris, age: 20-30) have been fully analyzed yet. The model with phonologically-conditioned cyclicity was found to provide the best fit to the acoustic data as compared to a fully phonotactic model, a fully cyclic model, and a model with morphologically-conditioned cyclicity (derivation vs. inflection). The vowel of the suffix is also shown to affect the realization of the stem mid vowel, in accordance with previous results on V-to-V coarticulation in Standard French (Nguyen and Fagyal 2008). All in all, the results of this study (if confirmed when all participants are considered) add to the body of evidence suggesting that preschwa syllables have a special status in French (Durand 1976, Selkirk 1978).



Alexandre Vaxman

Parameter dependencies for accent assignment in typological perspective 

Metrical stress theories tend to be excessively powerful. Assuming separation of accent (“primary stress”) from rhythm (“non-primary stress”), this talk introduces a specific set of binary parameters, augmented with parameter dependencies and ordering, and discusses how this system successfully generates all, and only, the attested phonological accent languages. The proposed parameter system forms a basic component of the Scales-and-Parameters (S&P) grammar (Vaxman 2016). Parameter space reduction in S&P is due to dependencies holding among certain parameters. In particular, the Accent Locality Dependency formally captures a broad descriptive generalization that severely restricts possible accent locations in weight-sensitive systems with final extrametricality. The generalization receives strong empirical support from tests against data in StressTyp (the largest-to-date typological database of stress patterns) and in the existing literature. The resulting parameter system, with its dependencies and a universal order of parameter setting, leads to an algorithm for generation of accentual languages (sets of accent patterns). An exhaustive parametric typology reveals that the generated parameter space accurately fits the set of attested phonological accent languages, which amounts to descriptive adequacy. Importantly, although the S&P parameter system, by itself, is limited to this class of languages, augmenting the system with a component containing novel types of weight scales allows for a uniform account of accent assignment in phonological, lexical and mixed accent languages in terms of a single accentual grammar.



Ander Egurtzegi 

(Ludwig-Maximilians University of Munich)


Contact-induced shift in the laryngeal features of Proto-Basque stops


Modern Basque shows an “true voicing” opposition with voiced vs. plain voiceless stops, and non-contrastive voiceless aspirated variants in some eastern dialects (Hualde 2003). Nevertheless, most authors (cf. Martinet 1950; Michelena 1977; Trask 1985; Hualde 1999; Lakarra 2013) reconstruct a different system for Proto-Basque. The classic reconstruction of the Proto-Basque stop system (Martinet 1950) involves two series of stops (fortis and lenis) with different phonetic realizations in different positions within the word. I present a new approach to the nature of this opposition. I reconstruct two series of stops: a voiceless aspirated series (specified for [spread glottis]) and an unaspirated series (laryngeally neutral) that is realized as unaspirated voiceless stops word initially and passively voiced in medial position, undergoing intervocalic spirantization in later times (tentatively, in the early Middle Ages). Following the analysis provided by frameworks such as laryngeal realism (Iverson & Salmons 1995; Jessen & Ringen 2002; Honeybone 2005; Beckman, Jessen & Ringen 2013, etc.), I reconstruct the Proto-Basque stop opposition as contrasting in [spread glottis] instead of [voice], which is the relevant feature in modern Basque, a true voice language, as the Romance languages historically in contact with it (Castilian Spanish, Gascon, French, etc.).



 

Markus Pöchtrager 

(university of Vienna) 



Vowel height and structure

Basically all versions of Government Phonology/Element Theory share the idea that the element A is a key player in the representation of vowel height. However, this element A has proven to be an eternal trouble maker, in that its behaviour is quite different from other elements. While that difference has been known for a long time, finding a viable explanation for it has turned out to be a more recalcitrant problem, if it has been attempted at all.

In this talk I first want to look at the element A in vowels and propose a particular kind of structure to replace it. This structure allows for a unified account of vowel reduction and also provides a link to stress. Interestingly enough, this structural approach also makes predictions about other vowel-related phenomena, viz. vowel harmony. In particular, it allows for an account of the behaviour and distribution of so-called transparent vowels (vowels that are skipped by harmony), an unexpected result for a theory of vowel height. Better still, the model can also also be extended to deal with cases of lenition of consonants. 














Archives

Les archives des réunions passées (depuis l'automne 2012) sont disponibles ici.

The archive of past meetings of the Atelier is available here.




Contacts


Pour venir présenter ou être ajouté à la liste de diffusion, merci de nous contacter à l'adresse :
If you would like to come present to the Atelier or to be added to the mailing list, please contact us at:
noam.faust@univ-paris8.fr.