Longer VOT Research Articles

Acoustic characteristics of English/n, l, t, k/ produced by native speakers of American English and Japanese at and near the word boundary

The acoustic characteristics of English /n, l, t, k/ produced by native speakers of American English and Japanese at and near the word boundary are compared. The phonemes are produced in two-word sequences in sentence contexts, /n/ or /l/ appears at the end of the first word (e.g., an earth, you'll earn) or at the beginning of the second word (e.g., a nurse, you learn), while /t/ or /k/ appears at the beginning of the second word (e.g., stops talking, ice cream) or as the second segment of the initial cluster /sk/ of the second word (e.g., stop stalking, I screamed). Acoustic analysis revealed that English speakers' /n/ in coda position is shorter and creakier, while Japanese speakers' /n/ is as long as in onset. In addition, English speakers' vowels in /Vl#/ context are more velarized than vowels in /V#l/ context, whereas Japanese speakers' vowels in /Vl#/ context are less /l/-colored. English speakers produced initial /t/ and /k/ with much longer VOT than in /st/, /sk/ conditions, but because Japanese voiceless stops are weakly aspirated, Japanese speakers /t/ and /k/ not significantly different in VOT between the two contexts, leading potentially to syllable misparsing by English listeners

Speaking rate normalization with and without segregation of simultaneous context sentences

Speech perception is shaped by acoustic properties of earlier sounds influencing recognition of later speech sounds. For example, when a context sentence is spoken at a faster rate, the following target word (varying from “deer” to “tier”) is perceived as “tier” (longer VOT) more often; when a slower context sentence is spoken, the following target word is perceived as “deer” (shorter VOT) more often. This is known as a temporal contrast effect (TCE, a.k.a. speaking rate normalization). Recently, Bosker, Sjerps, and Reinisch (2020 Scientific Reports) concluded that selective attention (to one of two simultaneous talkers) had no impact on TCEs. However, their paradigm was not an ideal test of this question; the voices heard were different talkers presented to each ear, and thus relatively easy to perceptually separate. Here, on each trial, the same talker spoke one sentence to both ears (no segregation), two sentences simultaneously to both ears (poor segregation), or a different sentence to each ear (easier segregation). Fast or slow context sentences preceded target words varying from “deer” to “tier.” TCE magnitudes were similar across all presentation modes. Results are consistent with the claims set forth by Bosker et al. – TCEs are automatic and low-level, not modulated by selective attention.

Enhancing secondary voicing cues in words with minimal pairs

Lexical competition conditions phonetic variation. For example, words with more phonological neighbors are produced with more hyperarticulated vowels than words with fewer neighbors, and /p,t,k/-initial words with voiced stop minimal pairs have longer VOT. Such effects enhance contrastive features and may be aimed at improving perceptibility of more potentially confusable words. Secondary features may also help distinguish words, as with lengthened vowels that cue coda consonant voicing in English. Our study investigates whether secondary features are enhanced in the production of lexically confusable words, comparing vowel duration in words with final voiced consonants between words with and without minimal pair competitors (e.g., bag with minimal pair back vs. gag with no pair gack). Fifteen American English speakers produced 30 monosyllabic English words with voiced final consonants, half with minimal pair competitors and half without (matched for frequency and neighborhood density). Pre-voiced vowels were found to be longer in minimal pair words than in non-minimal pair words; interactions indicate that the effect is stronger for some vowels. These findings show that it is not just contrastive features, but secondary features too that are subject to enhancement in lexical competition contexts, since they too can contribute to successful lexical perception.

Comparison of Voice Onset Time in People with Spastic Dysarthria and Healthy Group

Objectives: Given the role of voice onset time in speech production and its value in the identification of speech disorders, the present paper aimed at comparing VOT in people with multiple sclerosis and healthy group and investigating the factors affecting VOT. Methods: In this cross-sectional analysis study, 36 patients with MS with spastic dysarthria and 36 healthy subjects were investigated. After placing the subjects in an acoustic environment, the acoustic signal of the voiced and voiceless stop words /p/, /t/, /k/, /b/, /d/, /g/ with the vowel /a/, in the tissue cvc was recorded using the Sure-beta54 microphone. The spectrogram of the words was checked with the Praat version 6.0.36. Data were analyzed by the Shapiro-Wilk test, Independent t-test, and two-way analysis of variance. Results: Patients with MS have a longer VOT than healthy people, although the difference is not statistically significant (P > 0.05). The study of the effect of place of articulation on VOT revealed that the place of articulation was effective on the VOT in the healthy controls, however, there was no significant difference in the patient group. There was a significant difference in the effect of voiced-voiceless words on VOT (P 0.05). Conclusions: Patients with MS differ in their motor coordination between articulator structures and speech structure compared to the healthy group. Deficient speech production timing causes problems in speech motor control and ultimately changes the speech of the affected people.

Acoustic correlates of Apurímac Quechua ejective stops

Apurimac Quechua ejective and pulmonic stops are compared with respect to VOT, rise time (the following vowel’s peak amplitude – onset amplitude), and F0 difference (following vowel’s midpoint amplitude – onset amplitude). These three properties distinguish ejective from pulmonic stops in other languages (e.g., Ingush) but were previously unanalyzed in Quechuan languages. Results show that only rise time and VOT distinguish Apurimac Quechua ejective stops. All syllables (stressed and unstressed) of CV.CV.CV words were analyzed. The vowel /a/ followed each stop onset. 60 pulmonic plus 40 ejective stops were compared (all places of articulation) (α = 0.05). Rise time is the strongest acoustic cue for ejective stops: vowels following ejective stops had significantly longer rise times (meanejective= 7.74 dB, SD = 1.27) than pulmonic (meanpulmonic = 4.85 dB, SD = 2.94). Ejective stops therefore rise to peak amplitude more slowly than pulmonic. F0 difference did not significantly differ between vowels following each stop type (meanejective = -0.32 Hz, SD = 0.05; meanpulmonic = 0.40 Hz, SD = 0.03). Ejective stops had significantly longer VOTs (meanejective = 27.82 m/s, SD = 2.01) than pulmonic (meanpulmonic = 23.98 m/s, SD = 4.09). This difference is small, indicating that VOT is a less-robust acoustic cue for ejective stops.Apurimac Quechua ejective and pulmonic stops are compared with respect to VOT, rise time (the following vowel’s peak amplitude – onset amplitude), and F0 difference (following vowel’s midpoint amplitude – onset amplitude). These three properties distinguish ejective from pulmonic stops in other languages (e.g., Ingush) but were previously unanalyzed in Quechuan languages. Results show that only rise time and VOT distinguish Apurimac Quechua ejective stops. All syllables (stressed and unstressed) of CV.CV.CV words were analyzed. The vowel /a/ followed each stop onset. 60 pulmonic plus 40 ejective stops were compared (all places of articulation) (α = 0.05). Rise time is the strongest acoustic cue for ejective stops: vowels following ejective stops had significantly longer rise times (meanejective= 7.74 dB, SD = 1.27) than pulmonic (meanpulmonic = 4.85 dB, SD = 2.94). Ejective stops therefore rise to peak amplitude more slowly than pulmonic. F0 difference did not significantly differ between vowels following...

VOT in Tibetan is conditioned by tone

Several South Asian languages exhibit interactions among tone and consonant phonation, including Tamang (Mazoudon 2014) and Gurung (Ronkos 2015). In Central Tibetan, oral stops present a four-way contrast between two voicing/aspiration categories split across two lexical tones, high and low/rising. Across tones, previous analyses propose either a two-way contrast between unaspirated and aspirated stops (e.g., Dawson 1980), or a three-way contrast between voiced, unaspirated, and aspirated series (e.g., Denwood 1999), though acoustic measurements are lacking. This study recorded 19 native speakers of Tibetan in Kathmandu, Nepal, producing words exemplifying the phonological contrasts in stops and high/low tones. VOT measurements showed similar values for the unaspirated stops in both tonal environments, but a longer VOT in high-tone aspirated stops (46 ms) than low-tone aspirated stops (30 ms), or three different positive VOT values. That this three-way surface contrast cross-cuts tonal categories suggests an underlying two-way aspiration contrast. The apparent contrast between middle-VOT and long-VOT series results from the presence of a phonologically-specified high tone, which causes a lengthened VOT. This provides evidence for the conditioning of apparently-segmental characteristics by tone.Several South Asian languages exhibit interactions among tone and consonant phonation, including Tamang (Mazoudon 2014) and Gurung (Ronkos 2015). In Central Tibetan, oral stops present a four-way contrast between two voicing/aspiration categories split across two lexical tones, high and low/rising. Across tones, previous analyses propose either a two-way contrast between unaspirated and aspirated stops (e.g., Dawson 1980), or a three-way contrast between voiced, unaspirated, and aspirated series (e.g., Denwood 1999), though acoustic measurements are lacking. This study recorded 19 native speakers of Tibetan in Kathmandu, Nepal, producing words exemplifying the phonological contrasts in stops and high/low tones. VOT measurements showed similar values for the unaspirated stops in both tonal environments, but a longer VOT in high-tone aspirated stops (46 ms) than low-tone aspirated stops (30 ms), or three different positive VOT values. That this three-way surface contrast cross-cuts tonal categories suggests...

The effect of domain-initial strengthening on voice onset time and vowel length on English oronyms

Oronyms are sets of words and phrases that contain the same phonemes but differ in word boundary (atop, a top). Past research established that domain-initial strengthening (DIS) affects segments at the beginning of prosodic domains. This research examines voice onset time and vowel length in English oronyms with the hope of finding a systematic strengthening of the target sounds word-initially. The word list consists of oronym pairs with the same phonemes and the same [vowel, voiceless stop, vowel] at the beginning of the oronym (e.g. atop/a top, attack/a tack). To elicit more natural-sounding tokens, participants created sentences with the target oronym at the beginning, or read already-prepared sentences. It is hypothesized that there will be a systematic strengthening of the initial segments (in the form of either a longer VOT or vowel length). This could mean speakers are fully aware of word boundaries and use DIS to help avoid lexical ambiguity when speaking.

Individual differences in phonetic drift by English-speaking learners of Spanish

Second language learning affects L1 pronunciation even in early stages of adult L2 acquisition (e.g., Guion 2003, Chang 2012), an effect sometimes called “phonetic drift.” English voiceless stops assimilate to Korean stops in novice L1 English-L2 Korean learners in Korea, arguably due to English voiceless stops and Korean aspirated stops being linked to the same phonological category (Chang 2012), and the “novelty” of the Korean language input with its longer VOTs (Chang 2014). We tested whether novice L1 English-L2 Spanish learners in the United States also show assimilatory L2-to-L1 effects. Stronger typological and orthographic similarities might lead speakers to equate English and Spanish stops, but phonetic differences—short-lag VOT on voiceless stops and pre-voicing of voiced stops in Spanish—might impede L2-to-L1 assimilation. Monolingual English college-students were recorded every other week for the first 12 weeks of an introductory Spanish course. VOT data show considerable individual differences—some subjects show assimilation, some dissimilation, between English and Spanish stops. This suggests that while orthography and typological similarity may play a role, how a learner’s first and second language interact in novice adult L2 acquisition is also affected by individual factors such as perceptual sensitivity to phonetic differences, sociolinguistics considerations, and possibly speaker strategies for maintaining contrast (e.g., Nielsen 2011).

Voice onset time and speakers’ age: Data from Hungarian

The aim of this paper was to investigate the VOT of voiceless plosives (/p, t, k/) in the speech of Hungarian-speaking elderly. Read speech of 25 old (70 to 90 years) and 25 young (21 to 32 years) was analyzed. In each recording, the VOT of phonologically short [p, t, k] was measured. Our data show that VOT values of all three types of voiceless plosives would exhibit significant differences both in old and in young Hungarians’ speech. Bilabial and alveolar plosives had significantly longer VOT in old subjects’ speech than in that of young subjects, while old subjects produced significantly shorter VOTs in pronouncing [k] than their younger peers. We argue that these results are attributable to (1) significantly slower rate of articulation (yielding longer speech sounds in general), and (2) the articulatory and aerodynamic background of the production of plosives.

Phonetic detail that distinguishes prefixed from pseudo-prefixed words

Most English prefixes are syllables that can also begin words in which they do not function as a productive prefix. The literature notes a pronunciation difference such that true-prefixes, e.g. /dɪs/ in discolour, have a heavier rhythmic beat than pseudo-prefixes, e.g. /dɪs/ in discover. When the syllable following dis- or mis- begins with a voiceless stop, there is a clear difference in its VOT, but differences in dis/mis itself are more subtle and have not been systematically measured. Five speakers of Southern British English engaged in 40 scripted dialogues which contained such words in controlled phonetic contexts. Prefixed words were longer up to voicing onset in syllable 2 and had longer and more peripheral [ɪ], longer VOT, and shorter [s] than pseudo-prefixed words. These differences produced distinctive acoustic patterns consistent with the difference in perceived beat. Effects due to nuclear/postnuclear accent and word frequency were observed, but appear to be secondary to effects of morphological status. We conclude that the morphological status of these syllables is the primary cause of their characteristic acoustic patterns, and that their segmental composition dictates further reduction processes they may undergo due to weaker prosodic contexts, higher word frequency, casual register, and other influences.

A three-way VOT contrast in final position: data from Armenian

Standard Eastern Armenian has a three-way VOT contrast among its oral stop series. While it is rare for such a 3-way contrast to be preserved in final position, Standard Eastern Armenian is claimed to do just this (Khachaturian 1988; Ladefoged & Maddieson 1996; Vaux 1998). This phenomenon provides an ideal opportunity to explore how prosodic structure influences the realization of a complex and delicate system of contrast maintained by temporal and saliency distinctions. The cues to this contrast, including VOT, closure duration, and burst amplitude, are examined in a variety of segmental and prosodic environments. The experiment evaluates effects of intonation phrase final, intermediate phrase final, word final, and syllable final positions. We find that the 3-way VOT contrast is maintained in those prosodic domains in which the stop consonants are final. Speakers make significant VOT distinctions among the target consonants within the same boundary condition, and across prosodic conditions, larger prosodic domains exhibit longer VOT values.

Speech Perception, Lexicality, and Reading Skill

This study examined the interaction between speech perception and lexical information among a group of 7-year-old children, of which 26 were poor readers and 36 were good readers. The children's performance was examined on tasks assessing reading skill, phonological awareness, pseudoword repetition, and phoneme identification. Although good readers showed clearly defined categorical perception in the phoneme identification task for both the /bif/-/pif/ and the /bis/-/pis/ continua, the category boundary for /bif/-/pif/ was at longer VOTs than the boundary for /bis/-/pis/, which characterizes the classic lexicality effect. Poor readers showed less sharply defined categorical perception on both continua. Although poor readers did not show the classic lexicality effect, lexicality did affect the overall rate with which phonemes were identified as /b/ or /p/ at each VOT. These findings suggest that the lexicon may operate as a compensatory mechanism for resolving ambiguities in speech perception. Furthermore, statistical correction for group differences in phoneme identification made group differences in phoneme deletion disappear, suggesting that deficits in speech perception may play a causal role in the phonological core deficit associated with reading failure.

Post obstruent tensing in Korean: Its status and domain of application

The criteria that distinguish the phonological rules from the phonetic rules based on categoricality versus gradiency have drawn considerable attention (cf. Keating, 1990; Pierrehumbert, 1990). Korean Post Obstruent Tensing (POT) rule has been claimed, based on acoustic data, to be a phonological rule and that its domain of application is an accentual phrase (AP) (e.g., Jun, 1998). This paper questions the status of POT and its domain of application based on articulatory and acoustic data. By POT, a syllable initial lenis consonant (e.g., /t/, /s/) becomes tense (e.g., /t*/, /s*/, respectively) after an obstruent coda. It is known that lenis obstruents have a longer VOT, shorter closure durations, and shorter linguopalatal contact areas than tense obstruents. To investigate if POT is really a categorical process and if its application is definable in terms of an AP, acoustic and articulatory (i.e., EPG) data of derived and underlying tense obstruents were examined in four different prosodic positions: utterance-initial, intonational phrase-initial, AP-initial, AP-medial. Sentences containing a target sequence, /VpCV/, where C=/t/, /t*/, /s/, /s*/, were produced by two Korean speakers, at normal and fast speech rates. The results will shed light on the status of POT in the grammar of Korean.

Vowel-dependent VOT variation: An experimental study

That the VOT of voiceless stops is typically longer before high, close vowels and shorter before low, open ones is well documented. The hypothesis was tested that high vowels engender longer VOT because they offer greater resistance to the air escaping from the mouth (thereby delaying the transglottal pressure differential required for voicing). A correlation was sought between oral pressure decay (OPD) and VOT. Acoustic and aerodynamic data were collected from a male native speaker of American English uttering 12 nonsense words of the type atí, atà in isolation ten times. Pressure data was sampled by inserting a tube into the subject’s pharyngeal cavity via the nose. The VOT was measured from the release of the burst to the onset of the first periodic cycle. The OPD was defined as the time required for oral pressure to decrease from the peak to a fixed low threshold. The OPD and VOT measurements displayed a significant, linear correlation, indicating that VOT and vowel height are indeed mechanically linked. Moreover, tokens with voiceless stops preceding sonorants, e.g., atwà, akjà, yielded the longest VOTs since sonorants have narrower constrictions, resulting in longer OPDs.

Acoustic correlates of articulatory prosodic strengthening in Korean

Recent articulatory studies [e.g., Keating, Cho, Fougeron, and Hsu (1998), forthcoming] using electropalatography have shown that consonants are generally produced with greater linguopalatal contact in domain-initial positions. This effect, known as articulatory prosodic strengthening, is generally cumulative: the higher the prosodic position, the more linguopalatal contact. The goal of this paper is to test for acoustic correlates of any such articulatory variations in Korean consonants /n,t,th,t*/ in prosodic positions (Utterance-, IP-, AP-, Word- and Syllable-initial), so that eventually the perceptual significance of such correlates can be tested. Results show that the higher domain-initial positions are associated with longer VOTs; longer ‘‘total’’ voiceless duration in closure; lower rms burst energy in stops but with substantial inter-speaker/consonantal variations; lower % of voicing during closure; and lower acoustic energy for nasal /n/. In addition, vowels in domain-initial CV were not found to be necessarily longer in a higher position whereas vowels in domain-final positions were. Linear correlations between each of the acoustic parameters and the articulatory magnitude (linguopalatal contact) will also be discussed. All in all, this study shows that articulatory strengthening in Korean does produce reliable acoustic correlates, which mark off prosodic positions in the speech signal. [Work supported by NSF.]

Age‐related changes in the categorical perception of stop consonants.

The study focused on changes in the categorical perception of intial stop consonants as a function of age and hearing status. The identification of intial stop consonants differing in voicing and place of articulation relies on an integration of multiple acoustic cues (VOT, character of the noise burst, onset frequency and direction of the second and third formant transitions). Four hundred subjects, divided into four age groups (20–85 years) and into three hearing groups, participated in experiments that required the identification of syllables [ba] or [pa] and [ba], [da], or [ga] along synthesized continua. Stimuli were presented to subjects via headphones at 75 dB SPL. Results demonstrated that the phonemic boundary of the [ba–pa] continuum was shifted towards a longer VOT as a function of age and hearing status. Moreover, the phonemic boundaries between [ba] and [da], and between [da] and [ga], demonstrated by older and hearing impaired subjects were shifted toward [d] and [g] categories as compared to young subjects with normal hearing. The observed changes may be due to the age‐related deterioration of ability to analyze brief signals and ability to integrate multiple acoustical cues. [Work supported by Health and Welfare of Canada.]

Voice onsets and segment durations of normal speakers and beginning stutterers

Longer VOT and segment durations apparently characterize the contrived responses of school-age stutterers examined under laboratory conditions. Thus, these variables may reflect basic disparities between the speech production abilities of stutterers and normal speakers. To better explore this possibility, VOT and segment durations were assessed in five nonstutterers who were closely matched with five preschool stutterers who had been diagnosed and tested in the experiment within six months of parental concern about fluency development. Target words were measured in isolation, in elicited sentences, and conversation. Results revealed significantly slower VOT and longer segment durations in the young stuttering members of the pairs of speakers. The stutterers were also more variable in their management of these same two aspects of motor speech timing control.