Coarticulatory Effects Research Articles

The Historical Phonology of Intrusive Stops: A Non-Linear Description

Intrusive stops are traditionally understood as sounds emerging at the phonetic surface as the result of specific coarticulation effects, generally involving the manner of articulation features “nasal”, “continuant” and “lateral” as in French chambre (< Lat. cam(e)ra), etre (< Lat. ess(e)re) and OF moldre (< Lat. mol(e)re). In this paper I will discuss some recent proposals concerning the formal description of this phenomenon. In doing so, I will show that two types of intrusive-stop formations have to be distinguished: one type involves the appearance of an epenthetic consonant inside liquidfinal clusters; the second type concerns intrusive stops which originate inside sequences of consonants which do not have liquids as their second member. More specifically, it will be argued that these two types correlate systematically with a difference in segmental status of the intrusive stop involved: the former type requires the insertion of a consonant, whereas the intrusive element which appears in the latter type does not acquire segmental status.

The BAD-LAD split: Secondary /æ/-lengthening in Southern Standard British English

Descriptions of the Received Pronunciation (RP) and Southern Standard British English (SSBE) have previously commented on vowel lengthening in certain words such as bad and glad as opposed to shorter lad and pad. This paper comprises the first thorough description of the conditioning of /æ/ duration in twenty-one young native SSBE speakers, finding inconclusive evidence for a lexically specified split but significant general lengthening effects of postvocalic /g/ and /d/; this secondary /æ/-lengthening is discussed in reference to phonological analyses of the TRAP-BATH split (primary /æ/-lengthening) and previously established descriptions of co-articulatory segmental effects on vowel length.

Acoustic characteristics of vowels produced by Greek intelligible speakers with profound hearing impairment II: The influence of stress and context

Purpose: The present paper examines the influence of stress and context on selected acoustic characteristics of vowels produced by six adult Greek intelligible speakers with profound hearing impairment and six speakers with normal hearing (three males and three females in each group).Method: F1, F2 and F3 formant frequencies and the duration of vowels are measured in words of the form /′pVCV/ with V = /i, ɛ, ɐ, ɔ, u/ and C = /p, t, k, s/. Variation in these parameters due to context and stress is reported for the two genders. A co-articulatory measure and three measures that examine the area of the vowel space and differences along the F1 and F2 axes are calculated.Result: The results show a reduction of the vowel space in the unstressed condition for both groups, while vowel duration was found to be significantly longer in this condition for the speakers with hearing impairment. In addition, smaller C-to-V carryover co-articulatory effects were found for the speakers with hearing impairment.Conclusion: Findings are discussed within the framework of perceptual and production constraints in hearing impairment and with reference to current models of co-articulation.

Perceptual compensation in nasal and nasalized vowels in Brazilian Portuguese

Listeners of a language can factor out effects of coarticulation in the signal, consistent with the language’s phonological patterns. Such perceptual compensation should not occur for phonetic features that are inherent to a segment, not effects of coarticulation. Vowel nasality is an interesting feature in this case, because it can be coarticulatory or inherent depending on the language. In Brazilian Portuguese (BP), it can be both. Contrastive nasal vowels may be followed by a brief nasal resonance (nasal appendix), while coarticulatory nasalized vowels must be followed by a nasal consonant. This study explores whether or not BP listeners perceptually compensate for vowel nasality in the perception of both phonemic nasal and coarticulatory nasalized vowels in BP, examining. If BP listeners process nasal vowels as phonemes, they should not attribute nasality in the vowel to coarticulation effects in any context, except nasal consonant. Randomized stimuli pairs containing oral, nasal and nasalized vowels in three different contexts (zero, nasal appendix, nasal consonant) were presented to BP listeners who had to judge which vowel was more nasal. Preliminary results demonstrate that BP listeners do not perceptually compensate for vowel nasality in any context, be it nasal appendix or nasal consonant.

Vowel-dependent variation in Cantonese /s/ from an individual-difference perspective.

Individual variation is ubiquitous in the acoustic realization of human speech; however, little is known about the nature of individual differences in coarticulation. Through an in-depth case study of the temporal dynamics of vocalic influences on the acoustic realization of Cantonese /s/, this study demonstrates that coarticulatory effects may vary by the sex and self-reported autistic-like traits of the individual. These findings have significant implications for research in phonetics, phonology, and sound change.

Dynamic measures of voice stability in young and old adults

This paper examines age-related differences in voice quality between 15 young and 15 old speakers, using electroglottographic (EGG) recordings of three elicitation types: sustained vowels, hVd word lists, and readings of the Rainbow Passage. For each elicitation type, fundamental frequency (F0), contact quotient (Qx), and their standard deviations are analysed. Results found, for all three elicitations, that Qx was significantly lower for older speakers. F0 was higher for the older speakers, but this was only significant for the hVd word lists. Sustained vowels were used to examine the effect of target pitch and loudness on F0 and Qx. Both Qx and F0 were raised with increased target loudness. The hVd word lists were used to examine the phonetic effects of different vowels. Lip rounded vowels had higher Qx than non-lip rounded vowels. Readings of the Rainbow Passage encompassed prosodic and phonetic variations in F0 and Qx, in addition to voice quality. F0 and Qx contours of the hVd word lists and the vowel/ei/taken from the Rainbow Passage were analysed using the discrete cosine transform. The hVd frame produced coarticulation effects on the Qx contour, but no such effects were evident for the vowel /ei/. The study concluded that using multiple elicitation types is beneficial in the study of the effect of age on the voice, and that analysing contour tracks for F0 and Qx will allow further examination of how prosodic and phonetic effects interact with age-related changes in voice quality.

Palatalization in educated Cairene Arabic

In the speech of some educated Cairenes, the coronal stops /t d tˤ dˤ/ acquire a secondary palatal articulation before high front vowels and glides, resulting in [tj] or [dj]. Based on first-hand data, this article provides a detailed phonetic and phonological investigation of this palatalization process in an attempt to fill a descriptive gap in the literature on Cairene Arabic. By examining the full range of triggers, targets, and potential blockers, I demonstrate that this is a phonetic coarticulatory effect, and that it should be distinguished from phonological assimilation. I also explore the hypothesis that this process – characterized as a sound change in progress – has been phonologized in a different sociolect of Cairene Arabic.

A comparison between the Dutch and American-English digits-in-noise (DIN) tests in normal-hearing listeners

Objective: The Dutch digits-in-noise test (NL DIN) and the American-English version (US DIN) are speech-in-noise tests for diagnostic and clinical usage. The present study investigated differences between NL DIN and US DIN speech reception thresholds (SRTs) for a group of native Dutch-speaking listeners. Design: In experiment 1, a repeated-measures design was used to compare SRTs for the NL DIN and US DIN in steady-state noise and interrupted noise for monaural, diotic, and dichotic listening conditions. In experiment 2, a subset of these conditions with additional speech material (i.e. US DIN triplets without inter-digit coarticulation/prosody) was used. Study sample: Experiment 1 was conducted with 16 normal-hearing Dutch students. Experiment 2 was conducted with nine different students. Results: No significant differences between SRTs measured with the NL DIN and US DIN were found in steady-state noise. In interrupted noise the US DIN SRTs were significantly better in monaural and diotic listening conditions. Experiment 2 demonstrated that these better SRTs cannot be explained by the combined effect of inter-digit coarticulation and prosody in the American-English triplets. Conclusions: The NL DIN and US DIN are highly comparable and valuable tests for measuring auditory speech recognition abilities. These tests promote across-language comparisons of results.

Sign Transition Modeling and a Scalable Solution to Continuous Sign Language Recognition for Real-World Applications

We propose a new approach to modeling transition information between signs in continuous Sign Language Recognition (SLR) and address some scalability issues in designing SLR systems. In contrast to Automatic Speech Recognition (ASR) in which the transition between speech sounds is often brief and mainly addressed by the coarticulation effect, the sign transition in continuous SLR is far from being clear and usually not easily and exactly characterized. Leveraging upon hidden Markov modeling techniques from ASR, we proposed a modeling framework for continuous SLR having the following major advantages, namely: (i) the system is easy to scale up to large-vocabulary SLR; (ii) modeling of signs as well as the transitions between signs is robust even for noisy data collected in real-world SLR; and (iii) extensions to training, decoding, and adaptation are directly applicable even with new deep learning algorithms. A pair of low-cost digital gloves affordable for the deaf and hard of hearing community is used to collect a collection of training and testing data for real-world SLR interaction applications. Evaluated on 1,024 testing sentences from five signers, a word accuracy rate of 87.4% is achieved using a vocabulary of 510 words. The SLR speed is in real time, requiring an average of 0.69s per sentence. The encouraging results indicate that it is feasible to develop real-world SLR applications based on the proposed SLR framework.

Do Isolated Vowels Represent Vowel Targets in French? An Acoustic Study On Coarticulation

Coarticulatory effects of labial, dental, palato-velar and uvular places of articulation on vowel targets of ten French oral vowels /i, e, ɛ, a, u, o, ɔ, y, o, œ/ were examined. The average vowel formant frequencies F1, F2, F3 and F4 in symmetrical sequences CVCVCVC to formant values of the same vowels in isolation were compared. The results show that the direction and magnitude of coarticulation of most vowels follow, as expected, contextual assimilation or acoustic centralization. Nevertheless, vowels /a/ and /ɔ/ present unpredictable coarticulatory patterns. This can be explained by the fact that 1) /a/ has two phonetic variants depending on the environment: back vowel [ɑ] in isolation and central/front vowel [a] in continuous speech, and 2) uttering [ɔ] in isolation violates the phonotactic rules of standard French.These results suggest that coarticulatory effects on vowels /a/ and /ɔ/ are probably not to be studied from their isolated positions.

The Effect of Stress and Speech Rate on Vowel Coarticulation in Catalan Vowel-Consonant-Vowel Sequences.

The goal of this study was to ascertain the effect of changes in stress and speech rate on vowel coarticulation in vowel-consonant-vowel sequences. Data on second formant coarticulatory effects as a function of changing /i/ versus /a/ were collected for five Catalan speakers' productions of vowel-consonant-vowel sequences with the fixed vowels /i/ and /a/ and consonants: the approximant /δ/, the alveolopalatal nasal /ɲ/, and /l/, which in the Catalan language differs in darkness degree according to speaker. In agreement with predictions formulated by the degree-of-articulation-constraint model of coarticulation, the size of the vowel coarticulatory effects was inversely related to the degree of articulatory constraint for the consonant, and the direction of those effects was mostly carryover or anticipatory in vowel-consonant-vowel sequences with highly constrained consonants (/ɲ/, dark /l/) and more variable whenever the intervocalic consonant was less constrained (/δ/, clear /l/). Stress and speech-rate variations had an effect on overall vowel duration, second formant frequency, and coarticulation size but not on the consonant-specific patterns of degree and direction of vowel coarticulation. These results indicate that prosodically induced coarticulatory changes conform to the basic principles of segmental coarticulatory organization.

Deriving long-distance coarticulation from local constraints

Coarticulatory effects can extend over two or more syllables. For example, we find in a study of English nonce words of the form [bV1C1əC2V2t] that F2 of V1 is shifted toward the F2 of V2. One approach to such long-distance coarticulatory effects posits direct interactions between the segments involved. For example, coproduction models attribute coarticulatory variation to temporal overlap between segments, so coarticulatory effects of V2 on V1 imply that a V2 gesture begins two syllables earlier, during V1. An alternative account posits that long-distance coarticulation results from iterative local coarticulation. That is, V1 can show coarticulatory effects of V2 because each intervening segment can partially assimilate to the next, resulting in a chain of coarticulatory effects between the two vowels. Since the iterative coarticulation analysis posits that long-distance coarticulation is mediated by intervening segments, it predicts that (i) coarticulatory variation at V1 due to V2 should be predictable...

Statistical parametric speech synthesis using a hidden trajectory model

A novel spectral modeling method for statistical parametric speech synthesis using a hidden trajectory model (HTM) is presented in this paper. An HTM is a structured generative model with a two-stage implementation. First hidden formant trajectories are generated from time-aligned formant target sequences by a bidirectional filter. This target-filtering model could provide a correlation structure across temporal frames and describe the effect of co-articulation on speech signals efficiently. Then the observed cepstral features are constituted by a formant-related component and a residual component. The formant-related component is predicted from hidden formant trajectories using a nonlinear and analytical function, and the prediction residuals are modeled by context-dependent Gaussians. In this paper, we apply HTM-based acoustic modeling to speech synthesis and investigate the effectiveness of this method in improving the naturalness and controllability of synthetic speech. Experimental results show that this proposed method can improve the accuracy of spectral feature prediction and the naturalness of synthetic speech compared with the conventional HMM-based method, especially for the conditions where the amount of training data is limited. Furthermore, this method can achieve effective controllability on vowel quality and formant sharpness of synthetic speech by conveniently manipulating the distribution parameters for the phone-dependent targets of formant frequencies and bandwidths.

Place of articulation of anterior nasal versus oral stops in Croatian

The purpose of this investigation is to analyse the place of articulation of anterior nasal versus oral stops in Croatian. Although there is agreement that placement for /n/ and /td/ is in the anterior region, there is disagreement among different authors about the precise place of articulation for these sounds. Some authors view these targets as sharing identical placement while others view placement of /n/ as more posterior to /td/. In this paper we use electropalatography (EPG) to investigate whether placement for these sounds is the same or different. The speech of six participants was recorded for the purposes of this study. The speech material consisted of 972 VCV sequences (V = /iau/, C = /ntd/). Four EPG indices were analysed: the ACoG measure, the amount of contact at dental and alveolar articulatory zones (dentoalveolar articulation being inferred indirectly), incomplete EPG closures and the lateral contact measure. Coarticulatory effects of vowels on placement were also measured. The results showed that /ntd/ generally shared the place of articulation in the dentoalveolar region, but also that relating quantitative physiological data to specific places of articulation should be done cautiously, taking into account variability in individual productions. The analyses also showed that /n/ had more incomplete EPG closures and a significantly lower amount of lateral contact when compared with /t/ and /d/. The nasal was more variable and showed less coarticulatory resistance in different vowel contexts than /t/ and /d/. The results of this study are discussed in terms of existing descriptions of Croatian consonant system and in light of cross-linguistic findings.

Quantifying lingual coarticulation using ultrasound imaging data collected with and without head stabilisation

Previous studies reporting the use of ultrasound tongue imaging with clinical populations have generally provided qualitative information on tongue movement. Meaningful quantitative measures for use in the clinic typically require the speaker’s head to be stabilised in relation to a transducer, which may be uncomfortable, and unsuitable for young children. The objective of this study was to explore the applicability of quantitative measurements of stabilisation-free tongue movement data, by comparing ultrasound data collected from 10 adolescents, with and without head stabilisation. Several measures of tongue shape were used to quantify coarticulatory influence from two contrasting vowels on four different consonants. Only one of the measures was completely unaffected by the stabilisation condition for all the consonants. The study also reported cross-consonant differences in vowel-related coarticulatory effects. The implications of the findings for the theory of coarticulation and for potential applications of stabilisation-free tongue curve measurements in clinical studies are discussed.

Early Speech Segmentation in French-learning Infants: Monosyllabic Words versus Embedded Syllables.

Lexical acquisition relies on many mechanisms, one of which corresponds to segmentation abilities, that is, the ability to extract word forms from fluent speech. This ability is important since words are rarely produced in isolation even when talking to infants. The present study explored whether young French-learning infants segment from fluent speech the rhythmic unit of their native language, the syllable. Using the Headturn Preference Procedure and the passage word order, we explored whether these infants can segment monosyllabic words (at 6 and 8 months), syllables embedded in bisyllabic words (at 6 months) and bisyllabic words (at 6 months). Our results bring direct evidence in support of the early rhythmic segmentation hypothesis, by establishing syllabic segmentation both for monosyllabic words and embedded syllables at 6 months, while failing to find segmentation of bisyllabic words at the same age. They also indirectly extend to French previously reported effects of coarticulation, acoustic variation and infant-directed speech on segmentation found in English. Therefore, our study contributes to a better understanding of the similarities and differences in early segmentation across languages, and thus to a better understanding of the mechanisms underlying segmentation.

Potential sports concussion identification using acoustic-phonetic analysis of vowel productions

Concussions impair cognitive function and muscle motor control; however, little is known about how this impairment affects speech production. In the present study, concussed athletes speech is recorded at the initial office visit and subsequent visits. The last recording, when the brain is determined to have recovered using present criteria, becomes the baseline. The vocabulary consists of seven h-vowel-d (hVd) words (who’d, heed, hood, hid, had, hud, and heard) produced three times each for a total of 21 productions. The study is focused on vowel characteristics and the limited coarticulatory effects of the hVd vocabulary make it ideal for the study. Duration measurements are made by experimenter analysis and the formant measurements are made using the automatic speech recognition engine ELBOW. The preliminary comparisons from the subjects completing the protocol show formant drift for three or more of the seven vowels, and duration is affected for each talker and each vowel. These results were anticipated since the impairment would affect articulatory movement and timing. The results as well as a discussion of development of an automated real-time concussion identification application will be presented.

Spatial and dynamic aspects of retroflex production: An ultrasound and EMA study of Kannada geminate stops

This study investigates the production of geminate retroflex stops in Kannada using a combination of ultrasound and articulography. Data obtained from 10 native speakers of the language show that the retroflex gesture is dynamically complex and asymmetrical, involving an anticipatory retraction of the tongue tip, followed by the raising of this articulator towards the hard palate, and subsequent rapid flapping-out movement during the closure and the release. The retroflex constriction and the forward movement appear to be facilitated by the simultaneous fronting of the posterior tongue body, flattening of the anterior tongue body, and lowering of the jaw. Compared to dental and velar stops, retroflex stops exert extensive anticipatory and perseverative coarticulatory effects on adjacent vowels and inter-speech intervals. With respect to the magnitude of the tongue tip displacement, the anticipatory effects are greater than perseverative effects. The results of the study thus offer a multi-faceted view of spatial and dynamic aspects of retroflex stop production in Kannada, confirming and extending previous findings for other Dravidian languages. The results also provide support for general models of lingual consonant production and coarticulation.

The coarticulation effects of vowel on the adjacent consonants in Mongolian

The coarticulation effects of vowel on the adjacent consonants in Mongolian

PHONETIC INFLUENCES ON ENGLISH AND FRENCH LISTENERS’ ASSIMILATION OF MANDARIN TONES TO NATIVE PROSODIC CATEGORIES

This study examined how native speakers of Australian English and French, nontone languages with different lexical stress properties, perceived Mandarin tones in a sentence environment according to their native sentence intonation categories (i-Categories) in connected speech. Results showed that both English and French speakers categorized Mandarin tones primarily on the phonetic similarities of the pitch contours between the Mandarin tones and their native i-Categories. Moreover, French but not English speakers were able to detect the fine-detailed phonetic differences between Tone 3 (T3) and Tone 4 (T4; i.e., low or low-falling tone vs. high-falling tone), which suggests that the stress differences between these languages may affect nonnative tone perception: English uses lexical stress, whereas French does not. In the discrimination task, the French listeners’ performance was better than that of the English listeners. For each group, discrimination of the Tone 1 (T1)–T4 and Tone 2 (T2)–T3 pairs was consistently and significantly lower than that of the other tone pairs, and the difference between T1-T4 and T2-T3 was significant. Discrimination of the Mandarin tone pairs was not fully predicted by pairwise categorizations to native i-Categories, however. Some discrimination differences were observed among tone pairs showing the same assimilation patterns. Phonetic overlaps in native i-Category choices for the Mandarin tones, strength of categorization (So, 2012), and tonal coarticulation effects (Xu, 1994, 1997) may offer possible accounts of these discrepancies between categorization and discrimination performance. These findings support the perceptual assimilation model for suprasegmentals (So &amp; Best, 2008, 2010a, 2010b, 2011, 2013), extended to categorization of nonnative tone words within sentence contexts to native i-Categories.