Heinrich – Heine-Université de Düsseldorf (Allemagne) – Spoken Morpholgy: Morpho-phonology revisited.

 Vos rendez-vous sont prévus les 7 septembre (16h-18h),  14 septembre (16h-18h), 5 octobre (16h-18h) et 12 octobre (16h-18h)

Lieu : Université Paris Diderot – Bâtiment Olympe de Gouges – room 203 – 8, place Paul Ricoeur 75013 Paris (at the end of the road Albert Einstein)

This seminar is about the pronunciation of morphologically complex words. Traditional approaches to the role of sound structure in complex words have focused on phonologically conditioned allomorphy or morphologically conditioned segmental or prosodic alternations (such as compound stress, stress shift, stress preservation, truncation, degemination, or syllabification). Such studies have detected interesting generalizations across sets of words but also exceptions to the observed regularities, and these findings have given rise to highly influential theoretical models like Lexical Morphology and Phonology, to far-reaching assumptions about the modular organization of grammar, and have informed models of speech production. However, recent empirical work on the variability of morpho-phonological alternations and on the phonetic implementation of complex words seriously challenges established theories of morphology-phonology interaction, the common distinction between lexical and post-lexical phonology, and current theories of speech production. In this seminar we will look in detail at some pertinent such studies from my lab that deal with compound stress, homophonous suffixes, (de-)gemination of prefixes and phonetic reduction in tri-constituent compounds.


Lecture 1:  Compound stress, informativity and analogy     Date : September 7th, 2015 (16:00-18:00)

After an introduction to the major theme of the seminar we turn to two studies in which we have investigated the variability of compound stress assignment in English. Based on the analysis of thousands of compounds it is shown that compound stress is best accounted for by models in which stress is assigned based on distributional properties of lexical items in the lexicon instead of by categorical rules.

Lecture 2:  Gemination and Degemination in English Prefixaton    Date : September 14th, 2015 (16:00-18:00)

In English, affixation may lead to the adjacency of two identical consonants across a morpheme boundary (e.g.un#necessary). The standard view of what happens in the case of un- and in- is that un- geminates (i.e. is pronounced with a long nasal), while in- displays degemination (i.e. is pronounced as if it had only one nasal, not two). I will present the first systematic empirical study of this phenomenon based on conversational speech. It turns out thatin- may also geminate, contra to the unanimously expressed opinion in the literature. This finding has important theoretical and methodological implications, which I will discuss.

Lecture 3: Homophony in morphology: the acoustic properties of morphemic and non-morphemic word-final S in English    Date : October 5th, 2015 (16:00-18:00)

In this lecture I will present an investigation that challenges the traditional assumption that homophonous suffixes are indeed phonetically non-distinct. The analysis of more than 600 tokens of non-morphemic /s/ and /z/, and of six different English /s/ and /z/ morphemes (plural, genitive, genitive-plural and 3rd person singular, as well as cliticized forms of has and is) shows that there are significant differences in acoustic duration between the seven categories. These findings challenge standard assumptions in morphological theory, Lexical Phonology and models of speech production.

Lecture 4: Morpho-phonology and hierarchical morphological structure   Date : October 12th, 2015 (16:00-18:00)

In the final lecture I will investigate the hypothesis that the internal bracketing of a word with more than two morphemes is reflected in the phonetic implementation in such a way that embedded forms show more phonetic reduction than forms at higher structural levels. I will present a study that tests the predictions derived from this hypothesis with tri-constituent compounds. The analysis of the durational properties of almost 500 compound tokens shows that certain effects can be found that indeed are indicative of the hierarchical structure and of the strength of the internal boundary of tri-constituent compounds. This means that morphological structure can be directly read off from the speech signal, an unexpected result for most, if not all, extant models morpho-phonology and speech production.