Session Poster VIII:Poster VIII
|
| Poster VIII-2 |
EPG CHARACTERISTICS OF VELAR STOPS IN NORMAL ADULT ENGLISH SPEAKERS
Marko Liker, University of Zagreb, Faculty of Humanities and Social Sciences Fiona Gibbon, Queen Margaret University, Edinburgh Paper File |
| This study aimed to determine the main characteristics of normal tongue palate patterns for velar stops. EPG data from the EUR-ACCOR database were analyzed for nonsense VCV sequences containing /k/ in nine vowel contexts for seven English speaking adults. Incomplete EPG closure across the palate for /k/ occurred in 19% (range 4%-41%) of utterances with the most number of incomplete closures in /aka/ and the least in /uki/. As predicted, place of articulation was dependent on vowel context. The most fronted was the velar occlusion in /iki/ and the most retracted in /aka/. In terms of amount of contact, /k/ in /iki/ environment had almost twice as much tongue palate contact compared to /aka/ in all speakers (mean 42% for /iki/ compared to 22% for /aka/). There was considerable interspeaker variability in all variables. The implications of the results for diagnosing and treating speech disorders are discussed. | |
| Poster VIII-4 |
Phonetic variation in Anong vowels
Ela Thurgood, Department of English, California State University, Chico Paper File |
| The study presents an acoustic analysis of Anong vowels, a language on the edge of extinction, investigating how language death affects their range, distribution, and degree of variability. The three most significant processes operating on Anong vowels are then discussed: coloring, nasalization, and laryngealization. | |
| Poster VIII-6 |
BAYESIAN FRAMEWORK FOR VOICING ALTERNATION & ASSIMILATION STUDIES ON LARGE CORPORA IN FRENCH
Martine ADDA-DECKER, LIMSI/CNRS bat. 508 F-91403 Orsay cedex Pierre André Hallé, LPP/CNRS 19 rue des Bernardins, 75005 Paris Paper File Additional Files |
| The presented work aims at exploring voicing alternation and assimilation on large corpora using a Bayesian framework. A voicing feature (VF) variable has been introduced whose value is determined using statistical acoustic phoneme models (3-state gaussian mixture Hidden Markov Models). For all relevant consonants, i.e. oral plosives and fricatives, their surface form voicing feature is determined by maximising the acoustic likelihood of the competing phoneme models. A voicing alternation (VA) measure counts the number of changes between underlying and surface form voicing features. Using a corpus of 70h of French journalistic speech, an overall voicing alternation rate of 2.7% has been measured, thus calibrating the method's accuracy. VA rate remains below 2% word-internally and on word starts and raises up to 9% on lexical word endings. In assimilation contexts rates grow significantly (>20%) highlighting regressive voicing assimilation. Results exhibit a weak tendency for progressive devoicing. | |
| Poster VIII-8 |
Differentiating the Effects of Speech Tempo on CV Coarticulation
Augustine Agwuele, Texas State University, San Marcos Harvey, M Sussman, University of Texas At Austin Björn Lindblom, University of Texas at Austin Amanda Miller, Cornell University Paper File Additional Files |
| When suprasegmental and contextually induced variations interact, coarticulation of consonant and vowel undergoes more complex, and often subtle, resistances and assimilations. This study empirically documents the separate effects of increase in speech tempo on C-locus and V-midpoint F2 patterns. Two female speakers of American English each produced 810 tokens [3V1* 3tops * 10V2 contexts * 3repetitions). Modified locus equation regression metric was employed to dissociate the effects of tempo on the vowel F2midpoints relative to the F2onsets at the CV boundary. The analyses uncovered two main systematic changes to F2onsets and these were attributed to (a) rate-induced reduction of the vowel space, (b) rate-induced changes on coarticulation per se. | |
| Poster VIII-10 |
TEMPORAL EXTENT OF NASALIZATION RELATIVE TO THE TONGUE ARTICULATION IN FRENCH NASALIZED VOWELS
Julie MONTAGU, LPP UMR 7018/CNRS Paris 3 Paper File Additional Files |
| The goal of this study is to investigate the temporal extent of nasalization in French oral vowels followed by nasal consonants. The extent of this nasalization was expected to vary as a function of vowel height. Results of this study show that the temporal extent of nasalization is related not only to the height dimension but also to the front-back position of the tongue. More specifically, it is shown that mid-low [E] patterns with front high vowels [i, y], and that mid-low [O] patterns with back high vowel [u], implying that the mid-low vowels do not pattern together as would be expected were the extent of nasalization mainly related to the height dimension. Links between the articulatory gestures (velum opening, height and front-back positions of the tongue) and their timing are outlined in the discussion. | |
| Poster VIII-12 |
ARTICULATORY MODELING OF CONSONANT RELEASE GESTURE
Hosung Nam, Yale University & Haskins Laboratories Paper File |
| This study examines the dynamics of consonant release motion. The release in syllable onset is shown to exhibit consonant-like speed to reach the rest position but the rest position corresponds to that of a concomitant vowel. We attempt to model such dual status of the release motion in the task-dynamic speech production model by splitting dynamic blending parameter into stiffness and target. | |
| Poster VIII-14 |
PHONETIC REALIZATION OF CONTRASTIVELY ASPIRATED AFFRICATES IN NEPALI
G. N. Clements, LPP, CNRS/Sorbonne-nouvelle (Paris) Rajesh Khatiwada, LPP, CNRS/Sorbonne-nouvelle (Paris) Paper File |
| Minimally contrastive aspirated affricates are rare among the world's languages. We investigate the realization of these sounds in Nepali, an Indo-Aryan language spoken in Nepal. Static palatography of CVCV forms reveals no articulatory difference between plain and aspirated affricates at the supralaryngeal level. However, a study of their acoustic characteristics shows that aspiration has variant realizations according to the nature of the consonant (voiced or voiceless), the identity of the following vowel (open /aa/ vs. palatal /i/), and position within the word (first vs. second syllable). Moreover, aspiration is not realized uniquely or even primarily at the affricate release, but more reliably on the following vowel, which is partly or entirely aspirated or assibilated, according to the following vowel. It is suggested that some of this variation may be attributed to the enhancement of an auditorily weak feature -- [spread glottis] -- by more salient secondary features. | |
| Poster VIII-16 |
SECONDARY STRESS VOWELS IN AMERICAN ENGLISH: TARGET UNDERSHOOT OF F1 AND F2 FORMANT VALUES
Mariko Sugahara, Doshisha University Paper File |
| We investigated whether the distinction between primary stress and secondary stress was ever correlated with vocal tract shape difference, which is acoustically realized as F1 and F2 transition differences. Since the syllables with secondary stress are usually shorter than those with primary stress, a natural prediction is that vocal tract articulators undergo some target undershoot when the secondary stress vowels are produced, which is in turn reflected as the target undershoot of F1 and F2 values. Our investigation of the F1 and F2 values of the first syllable of the DIgest(n)-diGEST(v) pair and the IMport(n)-imPORT(v) pair indicates that both F1 and F2 formants of the secondary stress vowels do not fully achieve their target values. That is, there is a phonetic reduction (target undershoot) of F1 and F2 values of the secondary stress vowels. | |
| Poster VIII-18 |
The sensitivity of intraoral pressure in consonants and consonant clusters to following vowel context in German
Laura Koenig, Haskins Labs and Long Island University, Brooklyn Susanne Fuchs, Center for General Linguistics (ZAS) Berlin Paper File |
| Most studies of intraoral pressure (IOP) variation have focused on consonantal voicing, place, and manner, but indirect data suggest that IOP during consonants may also show coarticulatory effects due to surrounding vowels. Here, we explore how IOP varies in consonants and consonant clusters of two German speakers depending on the following vowel (/i, I/ vs. /A/). Our data suggest that IOP varies consistently with vowel context, particularly with respect to the duration of the IOP pulse. Comparison with simultaneously-recorded EPG data indicate that the IOP results reflect longer articulatory contact time in the high vowel context. | |
| Poster VIII-20 |
Talking under conditions of altered auditory feedback: Does adaptation of one vowel generalize to other vowels?
E.J.S. Pile, Queen's University, CANADA H.R. Dajani, Queen's University, CANADA D.W. Purcell, University of Western Ontario, CANADA K.G. Munhall, Queen's University, CANADA Paper File |
| Evidence of perceptual learning has been found in various sensory systems, including the auditory system, but little research has examined the specificity of such learning. In the current study, participants' auditory feedback was altered in real time such that they heard their production of "head" shifted completely to sound like "had". This feedback modification induces a compensatory change in speech production. Following a period of training with this auditory feedback, subjects were tested on the vowels in "hid" and "hayed" to determine whether learning on one vowel generalized to nearby vowels. All participants produced a reliable compensation to the altered feedback, but there was no transfer of this compensation to the other vowels. Production of the nearby vowels also had no effect on the unlearning of the trained vowel. | |
| Poster VIII-22 |
PROCESSING OF ACOUSTIC CUES FOR VOICING IN ENGLISH: A MMN STUDY
Outi Tuomainen, Centre for Developmental Language Disorders and Cognitive Neuroscience, UCL, London Heather van der Lely, Centre for Developmental Language Disorders and Cognitive Neuroscience, UCL, London Paper File |
| In speech perception, multiple acoustic cues are used to signal a specific speech sound contrast. In the present experiment the processing of those acoustic cues that are responsible for syllable final stop consonant voicing in English was studied (i.e. vocalic duration and F1 offset frequency). Altogether nine subjects participated in an identification experiment of which six subjects took part in an active and passive (event-related potential, ERP) discrimination task. The subjects were presented with four different versions of English non-words [bot] and [bod] and their corresponding non-speech analogues as a control condition for the discrimination task. Results showed that the duration cue plays the most important role in British English syllable final stop voicing as measured with both identification and discrimination tasks. However, more data will be reported in the final paper. Keywords: speech perception, cue-weighting, Event-related potentials, mismatch negativity (MMN) | |
| Poster VIII-24 |
Effects of Noise on Lexical Tone Perception by Native and Non-Native Listeners
Tasha Dees, Department of Communication Sciences and Disorders, Northwestern University Ann Bradlow, Department of Linguistics, Northwestern University Sumitrajit Dhar, Department of Communication Sciences and Disorders, Northwestern University Patrick C.M. Wong, Department of Communication Sciences and Disorders, & Northwestern University Interdepartmental Neuroscience Program, Northwestern University Paper File |
| Speech in ecological environments is often accompanied by noise, which often proposes challenges to the listeners. In this study we examined differences in native Mandarin-speaking and native English-speaking adults’ ability to perceive lexical tones and vowels in different noisy listening conditions (white noise and multi-talker babble noise) at different signal-to-noise ratios. We found that although white noise impaired both listener groups’ speech perception ability, Mandarin multi-talker babble noise impaired Mandarin listeners to a greater extent. These results provide evidence for the differential effects of energetic and informational masking on native and non-native speech perception and the perceptual consequence of the reorganization of the auditory system after native language learning. | |
| Poster VIII-26 |
Musicians outperform nonmusicians in a study with shadowing speech
Barbara Pastuszek-Lipinska, Adam Mickiewicz University Paper File |
| In order to examine whether music education may be viewed as one of the factors which improve second language acquisition there were developed a research project. Two groups of subjects : with music training and without musical background were recruited and their responses to foreign language stimuli were recorded and then examined. Research results provide an evidence that musically trained people encounter fewer difficulties in foreign language sounds discrimination. | |
| Poster VIII-28 |
THE EFFECT OF DURATION SHORTENING ON THE ACOUSTIC CHARACTERISTICS OF THE DIPHTHONGS IN HANGZHOU CHINESE
JIAN LI, Department of Chinese, Translation and Linguistics, City University of Hong Kong Paper File |
| This study investigates whether the shortening of duration caused by fast speech rate and syllable final stop has an effect on the spectral characteristics of the diphthongs in Hangzhou Chinese (HC). Diphthongs in the open and closed syllables produced in normal and fast speech are investigated. The results indicate while the duration of the diphthongs are shortened in fast speech as compared to normal speech, the spectral targets of the diphthongs are not substantially changed, which may be due to the increased articulatory effort. The diphthongs in the open and closed syllables are different in temporal and spectral characteristics, which are not likely to be a direct result of duration shortening. The relationship between vowel duration and spectral reduction is discussed in the paper. | |
| Poster VIII-30 |
AN ACOUSTIC STUDY OF PLAIN AND PALATALIZED SIBILANTS IN OCOTEPEC MIXE
Silke Hamann, Utrecht University Heriberto Avelino, Berkeley University Paper File |
| In Totontepec Mixe, the stem-initial sibilants /s \texttoptiebar{ts} ʂ/ undergo a palatalization process when the prefix /j/ is added. Earlier descriptions report that this palatalization is realized either as addition of a glide (in the case of the alveolar and retroflex sibilants) or as a change in the primary place of articulation (in the case of the affricate). The acoustic measurements in the present study indicate that all palatalized sibilants have undergone secondary palatalization (they have an additional glide), and that only the retroflex segments show a consistent change in primary place of articulation. | |
| Poster VIII-32 |
An acoustic study of vowel contrasts in North Indian English
Olga Kalashnik , Monash University Janet Fletcher, University of Melbourne Paper File |
| The preliminary findings of an acoustic phonetic analysis of the vowels produced by speakers of English as a second language from North India are presented in this paper. Citation-form monophthong productions of a group of male NIE speakers were recorded and acoustically analysed. Results confirm a number of earlier impressionistic studies including the lack of a quality distinction between many tense/lax vowel pairs. North Indian English also appears to illustrate some vowel patterns that suggest it may be a separate subvariety of IE. | |
| Poster VIII-34 |
PROSODIC PARAMETERS FOR THE DETECTION OF REGIONAL VARIETIES IN ITALIAN
Elena SARDELLI, Dept of Linguistics, Univ. of Pisa Giovanna MAROTTA, Dept of Linguistics, Univ. of Pisa Paper File |
| The paper investigates the prosodic parameters which can be considered as relevant for the detection of regional varieties in spoken Italian. On the basis of acoustic analysis of three Italian varieties (Rome, Milan and Catanzaro), a crucial role is recognized in scaling and duration. Besides these features, tonal patterns and local pitch range also came out as being significant in variety discrimination through statistical distances. The same parameters are found relevant in sentence-type discrimination within the same variety. Results allow us to consider prosody as an important source of information for developing an automatic model in the geographical recognition of the speaker. | |
| Poster VIII-36 |
THE PHONETICS AND PHONOLOGY OF WH-QUESTION INTONATION IN MALTESE
ALEXANDRA VELLA, University of Malta Paper File Additional Files |
| This paper reports on a preliminary analysis of wh-question intonation in Maltese. The study aims to provide evidence for a characterisation of one aspect of the intonation of wh-questions in Maltese. It focuses on the initial H tone reported to occur at the beginning of such questions in Maltese and tries to establish an empirical basis for an analysis of this H tone. Results show that a H tone target is indeed always present on the wh-word in Maltese wh-questions. Moreover, the H target seems to be associated with the first syllable of the wh-word, whether this syllable is accented or not. Furthermore, the H tone in question is characterised by a raising in F0 as compared to other H tone targets across a range of different sentence types. | |
| Poster VIII-38 |
Duration, pauses, and the temporal structure of Mandarin conversational speech
Li-chiung Yang, Spoken Language Research, Dept. of Foreign Languages and Literature, Tunghai University, Taichung Paper File |
| In this study, we investigate pauses and durational patterns in Mandarin spontaneous conversation, as well as investigate how reliably such elements can serve as boundary-marking predictors across different types of speech modes and how language activities are affected by their cognitive correlates. Our results show that pause duration is significantly correlated with specific boundary status. We show that the duration elements are a fundamental component of discourse organization in spontaneous speech and simultaneously reflect exigencies of both cognitive processes and interactive communicative exchange. | |
| Poster VIII-40 |
MULTIFACTOR ANALYSIS OF DISCOURSE TURN IN GREEK
Antonis Botinis, Department of Linguistics, University of Athens Aikaterini Bakakou-Orphanou, Department of Linguistics, University of Athens Charalabos Themistocleous, Department of Linguistics, University of Athens Paper File |
| The present article concerns the study of turn construction in telephone conversations in Greek news broadcasts. The article is based on the segmentation of discourse in turn constructional units (TCUs) and the analysis of TCU structure as a function of prosodic characteristics of speaker, pause, fundamental frequency (F0), as well as syntactic structure and lexical elements, focusing on turn-taking and turn-leaving. The examination of significant correlations justifies a multifactor model of linear construction of TCU. | |
| Poster VIII-42 |
DISTRIBUTION OF DISFLUENCIES AND ERRORS IN ENGLISH DISCOURSE
Nanette Veilleux, Simmons College Alejna Brugos, Boston University, MA Stefanie Shattuck-Hufnagel, Massachusetts Institute of Technology, Cambridge, MA Alicia Patterson, Massachusetts Institute of Technology, Cambridge, MA Paper File |
| this increase is due to the heavy processing requirement incurred either in planning the next chunk of discourse or in the introduction of many new or high perplexity entities. In a sample of academic lecture speech, we find that non-error disfluencies (e.g. filled pauses) occur preferentially shortly after but not right at the beginning of a discourse segment. This suggests that the processing load may not increase just at the boundary onset, i.e. that the speaker can make use of earlier planning during the first portion of the new segment. In contrast, errors of selection or serial ordering of grammatical elements do not show a boundary-related peak in their distribution across a discourse segment, supporting the hypothesis that this second kind of nonfluent event arises at a different point in the speech production planning process. | |
| Poster VIII-44 |
Abstract Phonetic Categories are Predictable from Quantitative Phonotactics
Eleonora Cavalcante Albano, State University of Campinas Paper File |
| This paper presents a new finding and discusses some of its theoretical implications. The finding is that phonetic categorization, including major class membership, is entirely predictable from phonotactic biases in 3 Brazilian Portuguese word databases. The predictor parameters are log frequencies of 'VC, 'CV and V'CV sequences consisting of the 7 stressed vowels combined with the 19 onset consonants, plus the 5 pre-stressed vowels. Correct vowel categorization arises through discriminant function analysis of 'VC and 'CV data. Correct consonant categorization arises through discriminant function analysis of V'CV data. Results are consistent across databases and, thus, suggest that statistical biases in the lexicon can be stable enough to code phonetic categories. These findings have a bearing on the issue of the relationship of phonotactics to phonetics. | |
| Poster VIII-46 |
EVALUATING PHONOLOGICAL STATUS: SIGNIFICANCE OF PARADIGM UNIFORMITY VS. PROSODIC GROUPING EFFECTS
Renate Raffelsiefen , Institut für Deutsche Sprache Caren Brinckmann , Institut für Deutsche Sprache Paper File |
| A central concern of linguistic phonetics is to define criteria to determine the phonological status of sounds or sound properties observed in phonetic surface form. Based on acoustic measurements we show that the occurrence of schwa in German is determined exclusively by segmental and prosodic structure, with no paradigm uniformity effects. We argue that these findings are consistent with a uniform representation of syllabic sonorants as schwa sonorant sequences in the lexicon. The stability of schwa in CVC-suffixes (e.g. German diminutive suffix -chen), as opposed to its phonetic absence in segmentally comparable context, is argued to be conditioned by the prosodic organisation of such suffixes external to the phonological word of the stem. | |
| Poster VIII-48 |
Introducing a Comprehensive Approach to Assessing Pronuniciation Talent
Matthias Jilka, University of Stuttgart Henrike Baumotte, University of Stuttgart Natalie Lewandowski, University of Stuttgart Susanne Reiterer, University of Tübingen Giuseppina Rota, University of Stuttgart Paper File |
| This paper introduces a comprehensive project with the objective of finding the neuronal correlates of pronunciation talent. It concentrates on the first part of this undertaking, describing the extensive tests necessary to measure phonetic talent in its various dimensions such as production and perception, the segmental and suprasegmental levels of speech or different utterance forms such as spontaneous speech, reading and imitation. The project also investigates psychological and behavioral influences on pronunciation performance, as well as correlations with general linguistic aptitude. Example tasks making use of the distinction between categorical and realizational differences in intonation are used to demonstrate the detailed analyses allowed by the chosen approach. The described measures allow a reliable classification of talent level to be used in the selection of subjects for the neuroimaging studies in the second part of the project. Additionally, they provide general insights into the interactions between the examined talent-related parameters. | |
| Poster VIII-50 |
An acoustic study of first- and second-generation Gujarati immigrants in Wembley: Evidence for accent convergence?
Bronwen Evans, University College London Ajay Mistry, University College London Caroline Moreiras, University College London Paper File |
| The present study investigated differences in the vowel production of first- and second-generation Gujarati immigrants in Wembley, north London. Subjects were recorded producing the eleven British English monophthongal vowels in the phonetic context /hVd/. F1 and F2 formant frequency values were measured and compared with existing acoustic and auditory descriptions of London English. The results demonstrated that second-generation immigrants had not acquired the foreign-accented vowels of their parents. Instead, these subjects produced vowels that were more similar to Standard Southern British English (SSBE) speakers than to those of second-generation immigrants in similar ethnic communities in London. | |
| Poster VIII-52 |
From Russian to Polish: Positive transfer in third language acquisition
Grit Mehlhorn, University of Leipzig Paper File |
| Multilingual learners possess a larger repertoire of phonetic-phonological parameters, of language awareness and phonological knowledge as well as an increased cognitive flexibility, which supports their acquisition of the phonetics of further foreign languages. This papers shows phonetic similarities between Russian (as L2) and Polish (as L3) which can be used for positive transfer by native speakers of German (L1) learning these Slavic languages. Pedagogical implications concerning the teaching of phonetics for a third language are drawn. | |
| Poster VIII-54 |
PERCEPTION OF MANDARIN TONES BY CHINESE- AND ENGLISH-SPEAKING LISTENERS
Tsan Huang , SUNY at Buffalo, Department of Linguistics Paper File Additional Files |
| This paper reports on two experiments that tested the hypothesis that native phonology may influence speech perception. Both experiments used natural speech tokens of Standard Mandarin tones and Chinese- and American English-speaking listeners. The results from both the AX discrimination and the degree of difference rating experiments show language-specific effects: the Chinese-speaking listeners’ tone perception space was warped due to tone sandhi processes that neutralize two otherwise contrastive lexical tones. On the other hand, the English-speaking listeners showed phonetic listening, paying more attention to the similarity in pitch offset and onset between a pair of tones. | |
| Poster VIII-56 |
The reassigned spectrogram as a tool for voice identification
Sean A. Fulop, California State University Fresno Sandra Ferrari Disner, no affiliation Paper File |
| A precise imaging scheme, the reassigned (or time-corrected instantaneous frequency) spectrogram, holds out considerable promise for identifying the speaker of an utterance. Unlike conventional spectrograms, reassigned spectrograms can display a few tens of milliseconds of phonation in great detail, without blurring in the time or frequency domains, and they are also impervious to many forms of noise or channel contamination. They are thus able to reveal some unique time-frequency features of an individual's phonatory process. While further testing is needed to establish evaluation criteria and confidence estimates, it is encouraging to see how readily pairs of reassigned spectrograms can be matched in the set illustrated herein. At the very least, such images can augment the techniques that are currently in use for speaker identification and verification. | |
| Poster VIII-58 |
The phonological deficit in developmental dyslexia: is there a suprasegmental component?
Catherine Dickie, University of Edinburgh Mitsuhiko Ota, University of Edinburgh Ann Clark, Queen Margaret University Paper File |
| Adult dyslexics were tested on a range of tasks which were presented in two closely matched versions: a segmental version and a suprasegmental version. The tasks targeted phonological contrasts on one hand and the metalinguistic ability to manipulate phonological units on the other hand. Results showed that dyslexics did show a deficit in suprasegmentals as well as segmentals when the tasks involved manipulation, but the representation of suprasegmental contrasts does not appear to be impaired. | |
| Poster VIII-60 |
Speech Breathing in Patients with Adductor Spasmodic Dysphonia
Ruth Huntley Bahr, University of South Florida Katie Biedess, University of South Florida Marion B. Ridley, University of South Florida Paper File |
| Inductive plethysmography was used to calculate respiratory measures related to volume, timing, thoracic displacement and respiratory efficiency in patients with adductor spasmodic dysphonia (ADSD) compared to controls. Results revealed significant differences between groups and across tasks. Those with ADSD had statistically higher ventilation rates, more breaths per minute, a higher degree of muscular inefficiency/breathlessness and labored breathing. Differences between tasks were attributed to a higher cognitive-linguistic demand required during conversational speech. These findings support the idea that individuals with ADSD may experience difficulties with respiration as the effects of their Botox injection dissipate. | |
| Poster VIII-62 |
Effects of auditory feedback on hearing-impaired subjects’ production of the rounding feature: a pilot study
Johanna-Pascale ROY, Université du Québec à Montréal Lucie MÉNARD, Université du Québec à Montréal Amélie BRISEBOIS, Université du Québec à Montréal Mark TIEDE, Haskins Laboratories & MIT R.L.E. Paper File |
| This paper describes a pilot study that investigated the effects of auditory feedback on vowel production in prelingually hearing-impaired subjects. The rounding feature in French vowels is used to study the effects of hearing state on labial configurations and acoustic patterns. Subjects were recorded in two conditions: without their hearing aid (no auditory feedback) and with their hearing aid (with auditory feedback). The results show that temporary deprivation of auditory feedback causes changes in vowel production in the labial space. However, these changes are not reflected by a variation in vowel contrast, even though subjects tend to speak more clearly without their aid. It may be hypothesized that prelingually hearing-impaired subjects produce robust rounding features minimally influenced by temporary deprivation of auditory feedback. | |
| Poster VIII-64 |
Content-Based Transformation of the Expressivity in Speech
Grégory beller, IRCAM Xavier Rodet, IRCAM Paper File |
| In this paper we describe a speech expressivity transformation system giving the opportunity to a user to modify the expressivity of a spoken utterance. Statistical model are learned on a multispeaker expressive database using a Bayesian Network. The acoustic modification of the speech signal is achieved by a phase vocoder technology. The parameters of those transformations are context dependents. They change along the sentence in respect of pragmatic information such as stressing and depending on the phonetic transcription of the text. The system is now working for several acted emotions in french and is used for an artistic purpose dealing with multimedia and cinema. | |
| Poster VIII-66 |
Testing the Ecological Validity of Repetitive Speech
Greg Kochanski, The University of Oxford Christina Orphanidou, The University of Oxford Paper File |
| Can one trust experiments conducted with repetitive speech to represent normal language behaviour? We compare the spectra of repetitive productions of sentences with the same sentences read from a randomised list. We use a data-driven spectral distortion measure that is trained to respond to linguistically relevant differences. The measure is a distance measurement which is based on a classifier that separates sounds into linguistically equivalent and linguistically nonequivalent categories. We find that repetitive speech is not distinct from individually uttered speech. The difference between these two sorts of speech is smaller than variation within each. It is substantially smaller than typical differences between utterances produced by different subjects. | |
| Poster VIII-68 |
An Acoustic Study of Real and Imagined Foreigner-Directed Speech
Rebecca Scarborough, Stanford University, Department of Linguistics Jason Brenier , University of Colorado, Department of Linguistics Yuan Zhao, Stanford University, Department of Linguistics Lauren Hall-Lew, Stanford University, Department of Linguistics Olga Dmitrieva, Stanford University, Department of Linguistics Paper File |
| The acoustic properties of foreigner-directed speech are surprisingly understudied, and many existing studies evoke imagined interlocutors to elicit foreigner-directed speech. In this study ten native English speakers described the path between landmarks on a map to two confederate listeners (one native English speaker and one native Mandarin speaker) and to two imagined listeners (described as a native English speaker and a non-native speaker). Vowel duration, rate of speech, and vowel space size were examined across native-foreigner and real-imagined conditions. Stressed vowels were longer and rate of speech was slower in the foreigner-directed and imaginary conditions than in the native-directed and real ones. Vowel space differences were not significant. So, speakers made acoustic-phonetic adjustments in foreigner-directed speech that are consistent with those seen in listener-directed clear speech, and these accommodations were greater when the listener was imagined rather than real. | |