Amandine Michelas

When syntax meets prosody: Tonal and duration variability in French Accentual Phrases

Though two levels of phrasing are generally accepted for French, a large degree of intra-speaker variability in the amount of preboundary lengthening is generally found within the lowest level (the Accentual Phrase or AP). A question that still remains to be answered is whether this source of variability is merely due to speech rate fluctuations or to the existence of an additional level of phrasing ranked between the AP and the Intonation Phrase. This paper examines the effect of syntactic boundary strength on two phonetic correlates of boundary marking in French, for the AP level. Specifically, preboundary lengthening and tonal cues associated to four boundary levels are investigated through duration and f0 measures. Results show that the alignment of the AP boundary with a major syntactic break significantly modifies the degree of preboundary lengthening associated with the boundary. More precisely, AP-final syllables aligned with a NP/VP break are longer than AP-final syllables contained within a complex NP, though they are not marked by a stronger tonal boundary. Effects of speech rate on the tonal composition of the AP are also reported. We discuss the implications in the light of current models of French prosodic structure and the syntax/prosody interface.

Pitch scaling and the internal structuring of the Intonation Phrase in French

Within the Autosegmental-Metrical approach to French intonation, two levels of phrasing are generally accepted, the Accentual Phrase (AP) and the Intonation Phrase (IP), while the existence of an intermediate level, i.e., the intermediate phrase or ip, is controversial. Our data provide strong evidence for the existence of the ip by uncovering systematic pitch scaling effects associated with the right edge of this constituent. We first show that the presence of an ip-break is responsible for blocking recursive downstep of subsequent AP-final LH* rises in declarative utterances, causing the return of the H target to the pitch level set by the first accentual peak of the phrase (i.e., complete reset). This mechanism is claimed to result from control over the reference pitch level for the entire ip, which can alternatively be modeled through secondary association of the last pitch accent of the domain. We also report on duration evidence pointing to the existence of preboundary lengthening associated to the last syllable of the ip. Finally, we provide additional evidence for the internal structuring of the Intonation Phrase by reporting data on partial reset occurring on the first H peak following the ip boundary.

How deaf are French speakers to stress

This event-related potential study examined whether French listeners use stress at a phonological level when discriminating between stressed and unstressed words in their language. Participants heard five words and made same/different decisions about the final word (male voice) with respect to the four preceding words (different female voices). Compared to the first four context words, the target word was (i) phonemically and prosodically identical (/ʃu/-/ʃu/; control condition), (ii) phonemically identical but differing in the presence of a primary stress (/ʃu/-/ʃu/), (iii) prosodically identical but phonemically different (/ʃo/-/ʃu/), or (iv) both phonemically and prosodically different (/ʃo/-/ʃu/). Crucially, differences on the P200 and the following N200 components were observed for the /ʃu/-/ʃu/ and the /ʃo/-/ʃu/ conditions compared to the /ʃu/-/ʃu/ control condition. Moreover, on the N200 component more negativity was observed for the /ʃo/-/ʃu/ condition compared to the /ʃu/-/ʃu/ conditions, while no difference emerged between these two conditions on the earlier P200 component. Crucially, the results suggest that French listeners are capable of creating an abstract representation of stress. However, as they receive more input, participants react more strongly to phonemic than to stress information.

Do patients with schizophrenia use prosody to encode contrastive discourse status

Patients with schizophrenia (SZ) often display social cognition disorders, including Theory of Mind (ToM) impairments and communication disruptions. Thought language disorders appear to be primarily a disruption of pragmatics, SZ can also experience difficulties at other linguistic levels including the prosodic one. Here, using an interactive paradigm, we showed that SZ individuals did not use prosodic phrasing to encode the contrastive status of discourse referents in French. We used a semi-spontaneous task to elicit noun-adjective pairs in which the noun in the second noun-adjective fragment was identical to the noun in the first fragment (e.g., BONBONS marron “brown candies” vs. BONBONS violets “purple candies”) or could contrast with it (e.g., BOUGIES violettes “purple candles” vs. BONBONS violets “purple candies”). We found that healthy controls parsed the target noun in the second noun-adjective fragment separately from the color adjective, to warn their interlocutor that this noun constituted a contrastive entity (e.g., BOUGIES violettes followed by [BONBONS] [violets]) compared to when it referred to the same object as in the first fragment (e.g., BONBONS marron followed by [BONBONS violets]). On the contrary, SZ individuals did not use prosodic phrasing to encode contrastive status of target nouns. In addition, SZs difficulties to use prosody of contrast were correlated to their score in a classical ToM task (i.e., the hinting task). Taken together, our data provide evidence that SZ patients exhibit difficulties to prosodically encode discourse statuses and sketch a potential relationship between ToM and the use of linguistic prosody.

When pitch Accents Encode Speaker Commitment: Evidence from French Intonation

Recent studies on a variety of languages have shown that a speaker’s commitment to the propositional content of his or her utterance can be encoded, among other strategies, by pitch accent types. Since prior research mainly relied on lexical-stress languages, our understanding of how speakers of a non-lexical-stress language encode speaker commitment is limited. This paper explores the contribution of the last pitch accent of an intonation phrase to convey speaker commitment in French, a language that has stress at the phrasal level as well as a restricted set of pitch accents. In a production experiment, participants had to produce sentences in two pragmatic contexts: unbiased questions (the speaker had no particular belief with respect to the expected answer) and negatively biased questions (the speaker believed the proposition to be false). Results revealed that negatively biased questions consistently exhibited an additional unaccented F0 peak in the preaccentual syllable (an H+!H* pitch accent) while unbiased questions were often realized with a rising pattern across the accented syllable (an H* pitch accent). These results provide evidence that pitch accent types in French can signal the speaker’s belief about the certainty of the proposition expressed in French. It also has implications for the phonological model of French intonation.

On French listeners’ ability to use stress during spoken word processing

ABSTRACT Previous studies have suggested that French listeners experience difficulties when they have to discriminate between words that differ in stress. A limitation is that these studies used stress patterns that do not respect the rules of stress placement in French. In this study, three stress patterns were tested on bisyllabic words (1) the legal stress pattern in French, namely words that were unstressed compared to words that bore primary stress on their last syllable (/ʒuʁi/-/ʒu’ʁi/), (2) an illegal stress location pattern, namely words that bore primary stress on their first syllable compared to words that bore primary stress on their last syllable (/’ʒuʁi/-/ʒu’ʁi/) and (3) an illegal pattern that involves an unstressed word, namely words that were unstressed compared to words that bore primary stress on their first syllable (/ʒuʁi/-/’ʒuʁi/). In an ABX task, participants heard three items produced by three different speakers and had to indicate whether X was identical to A or B. The stimuli A and B varied in stress (/ʒu’ʁi/-/ʒuʁi/-/ʒu’ʁi/), in one phoneme (/ʒu’ʁi/-/ʒu’ʁɔ˜/-/ʒu’ʁi/) or in both stress and one phoneme (/ʒu’ʁi/-/ʒuʁɔ˜/-/ʒu’ʁi/). The results showed that French listeners are fully able to discriminate between two words differing in stress provided that the stress pattern included an unstressed word. More importantly, they suggest that the French listeners’ difficulties mainly reside in locating stress within words.

Poster #M83 DO PATIENTS WITH SCHIZOPHRENIA USE PROSODIC FOCUS MARKING TO ATTRIBUTE MENTAL STATES IN A CONVERSATION SITUATION?

Many previous studies have established that individuals with schizophrenia (SZ) experienced deficits in social cognition including Theory of Mind (ToM) disorders (Brune 2005, Green et al. 2008). Very few studies have investigated these impairments in schizophrenia during natural communication situations (McCabe et al. 2004, Champagne-Lavau et al. 2009). However, to fully characterize ToM ability in schizophrenia, we need to understand several components of this ability that specifically appear during social interactions. In the present study, we were interesting in focus marking as a linguistic marker of ToM. We know that speakers use prosody to encode the difference between the given/contrastive statuses (i.e. focus marking) of discourse referents in accordance with their beliefs about the hearers knowledge state. Specifically, in French, the contrastive information shows a tendency to be produced in a separate prosodic phrase from the rest of the utterance. In this study we tested whether the prosodic encoding of contrast is altered in SZ speech during social interaction and whether this alteration reflects ToM impairment.