Coarticulatory Vowel Nasality Research Articles

Comparing human and machine's use of coarticulatory vowel nasalization for linguistic classificationa).

Anticipatory coarticulation is a highly informative cue to upcoming linguistic information: listeners can identify that the word is ben and not bed by hearing the vowel alone. The present study compares the relative performances of human listeners and a self-supervised pre-trained speech model (wav2vec 2.0) in the use of nasal coarticulation to classify vowels. Stimuli consisted of nasalized (from CVN words) and non-nasalized (from CVCs) American English vowels produced by 60 humans and generated in 36 TTS voices. wav2vec 2.0 performance is similar to human listener performance, in aggregate. Broken down by vowel type: both wav2vec 2.0 and listeners perform higher for non-nasalized vowels produced naturally by humans. However, wav2vec 2.0 shows higher correct classification performance for nasalized vowels, than for non-nasalized vowels, for TTS voices. Speaker-level patterns reveal that listeners' use of coarticulation is highly variable across talkers. wav2vec 2.0 also shows cross-talker variability in performance. Analyses also reveal differences in the use of multiple acoustic cues in nasalized vowel classifications across listeners and the wav2vec 2.0. Findings have implications for understanding how coarticulatory variation is used in speech perception. Results also can provide insight into how neural systems learn to attend to the unique acoustic features of coarticulation.

The perception of nasal coarticulatory variation in face-masked speech.

This study investigates the impact of wearing a face mask on the production and perception of coarticulatory vowel nasalization. Speakers produced monosyllabic American English words with oral and nasal codas (i.e., CVC and CVN) in face-masked and un-face-masked conditions to a real human interlocutor. The vowel was either tense or lax. Acoustic analyses indicate that speakers produced greater coarticulatory vowel nasality in CVN items when wearing a face mask, particularly, when the vowel is lax, suggesting targeted enhancement of the oral-nasalized contrast in this condition. This enhancement is not observed for tense vowels. In a perception study, participants heard CV syllables excised from the recorded words and performed coda identifications. For lax vowels, listeners were more accurate at identifying the coda in the face-masked condition, indicating that they benefited from the speakers' production adjustments. Overall, the results indicate that speakers adapt their speech in specific contexts when wearing a face mask, and these speaker adjustments have an influence on listeners' abilities to identify words in the speech signal.

Prosodic Structural Effects on Non-Contrastive Coarticulatory Vowel Nasalization in L2 English by Korean Learners.

This acoustic study explores how Korean learners produce coarticulatory vowel nasalization in English that varies with prosodic structural factors of focus-induced prominence and boundary. N-duration and A1-P0 (degree of V-nasalization) are measured in consonant-vowel-nasal (CVN) and nasal-vowel-consonant (NVC) words in various prosodic structural conditions (phrase-final vs. phrase-medial; focused vs. unfocused). Korean learners show a systematic fine-tuning of the non-contrastive V-nasalization in second language (L2) English in relation to prosodic structure, although it does not pertain to learning new L2 sound categories (i.e., L2 English nasal consonants are directly mapped onto Korean nasal consonants). The prosodic structurally conditioned phonetic detail in English appears to be accessible in most part to Korean learners and was therefore reflected in their production of L2 English. Their L2 production, however, is also found to be constrained by their first language (L1-Korean) to some extent, resulting in some phonetic effects that deviate from both L1 and L2. The results suggest that the seemingly low-level coarticulatory process is indeed under the speaker's control in L2, which reflects interactions of the specificities of the phonetics-prosody interface in L1 and L2. The results are also discussed in terms of their implications for theories of L2 phonetics.

Producing and perceiving socially structured coarticulation: Coarticulatory nasalization in Afrikaans

Most theories of phonetics assume a tight relation between production and perception, and recent years have also seen increasing evidence for such a relation at the level of the individual. For the most part, however, this evidence comes from socially homogeneous speech communities where the targeted pattern of variation is mostly socially neutral. What implications might socially structured phonetic variation in the speech community have for the perception-production link? If listeners can predict the phonetic patterns of a talker based on the talker’s actual or assumed identity, would they adjust their perceptual strategies accordingly, possibly weakening the link between their own production and perception patterns? This study reports the results of a pair of experiments that investigate the production and perception of coarticulatory vowel nasalization in Afrikaans, a language for which variation in coarticulatory nasalization is socially structured. Relying on nasal airflow measures, the production experiment showed that speakers of White Afrikaans produce more extensive coarticulatory nasalization than speakers of Kleurling Afrikaans. The perception experiment used an eye-tracking paradigm to assess listeners’ perceptual reliance on coarticulatory nasalization, and found (i) that Afrikaans speakers’ use of coarticulatory nasalization in production predicts their perceptual reliance on this information, (ii) that they rapidly adjust to the coarticulatory timing patterns in the speech of other speakers, but also (iii) that they do not adjust their perceptual reliance on coarticulation in response to the assumed identity of the speaker. The link between perception and production therefore persists, even in this situation of socially structured variation in coarticulatory timing.

The degree and time course of nasal coarticulation across communicative contexts: A study of the LUCID corpus

We analyzed the degree and time course of coarticulatory vowel nasalization in hyperarticulated clear speech produced in real and imagined communicatively challenging conditions from the LUCID corpus (Baker & Hazan, 2011). Southern British English speakers completed an interactive spot-the-difference task in pairs when there was no overt communication barrier (NB), when the partner was non-native (L2), and when one speaker’s speech was vocoded (VOC) or mixed with talker babble (BABBLE). They also read sentences conversationally (READ-CO) and clearly (READ-CL). The results showed significantly greater overall keyword vowel nasalization in the BABBLE and L2 conditions than in NB, READ-CO, and READ-CL, as well as in VOC than in NB and READ-CO. READ-CL and READ-CO did not differ significantly. Examining vowel nasalization over time revealed that the significant differences emerged, on average, at 7.4% into the vowel. Speakers increased coarticulatory nasalization early in the vowel when producing hyperarticulated speech in response to a real communicative barrier, consistent with the idea that nasal coarticulation facilitates speech processing by cueing upcoming segments. It remains to be determined why nasal coarticulation is increased while coarticulation for other consonant-vowel sequences is decreased during the production of hyperarticulated listener-oriented speaking styles (Guo and Smiljanic, 2021).

Partial compensation for coarticulatory vowel nasalization across concatenative and neural text-to-speech.

This study investigates the perception of coarticulatory vowel nasality generated using different text-to-speech (TTS) methods in American English. Experiment 1 compared concatenative and neural TTS using a 4IAX task, where listeners discriminated between a word pair containing either both oral or nasalized vowels and a word pair containing one oral and one nasalized vowel. Vowels occurred either in identical or alternating consonant contexts across pairs to reveal perceptual sensitivity and compensatory behavior, respectively. For identical contexts, listeners were better at discriminating between oral and nasalized vowels in neural than in concatenative TTS for nasalized same-vowel trials, but better discrimination for concatenative TTS was observed for oral same-vowel trials. Meanwhile, listeners displayed less compensation for coarticulation in neural than in concatenative TTS. To determine whether apparent roboticity of the TTS voice shapes vowel discrimination and compensation patterns, a "roboticized" version of neural TTS was generated (monotonized f0 and addition of an echo), holding phonetic nasality constant; a ratings study (experiment 2) confirmed that the manipulation resulted in different apparent roboticity. Experiment 3 compared the discrimination of unmodified neural TTS and roboticized neural TTS: listeners displayed lower accuracy in identical contexts for roboticized relative to unmodified neural TTS, yet the performances in alternating contexts were similar.

Prosodic structurally conditioned variation of coarticulatory vowel nasalization in Mandarin Chinese: Its language specificity and cross-linguistic generalizability.

This study compares prosodic structural effects on nasal (N) duration and coarticulatory vowel (V) nasalization in NV (Nasal-Vowel) and CVN (Consonant-Vowel-Nasal) sequences in Mandarin Chinese with those found in English and Korean. Focus-induced prominence effects show cross-linguistically applicable coarticulatory resistance that enhances the vowel's phonological features. Boundary effects on the initial NV reduced N's nasality without having a robust effect on V-nasalization, whose direction is comparable to that in English and Korean. Boundary effects on the final CVN showed language specificity of V-nasalization, which could be partly attributable to the ongoing sound change of coda nasal lenition in Mandarin.

Focus and boundary effects on coarticulatory vowel nasalization in Korean with implications for cross-linguistic similarities and differences.

This study investigates focus and boundary effects on Korean nasal consonants and vowel nasalization. Under focus, nasal consonants lengthen in CVN# but shorten in #NVC, enhancing [nasal] vs [oral]. Vowels resist nasalization under focus, enhancing [oral]. Domain-initial nasal consonants denasalize, exercising no coarticulatory influence. Domain-final nasal consonants shorten counter to expectation, although vowel nasalization increases. Comparison with English data reveals similarities (focus-induced coarticulatory resistance) despite cross-linguistic differences in marking prominence, but it also suggests that prosodic-structural conditioning of non-contrastive vowel nasalization, albeit based on phonetic underpinnings of coarticulatory process, is fine-tuned in language-specific ways, resulting in cross-linguistic variation.

Is the target of imitation directly acoustic or a pattern within a speaker’s phonetic system?

Speakers may converge phonetically with another talker to whose speech they are exposed, resulting in reduced acoustic distance between interlocutors. However, it is unclear whether the target of imitation is raw acoustics or a linguistic pattern. Zellou et al. (2016) present a case that distinguishes these possibilities: after listening to a model talker whose speech was manipulated to reduce coarticulatory vowel nasality (measured acoustically in A1-P0), participants’ nasality also decreased in post-test productions, relative to their baseline. However, the model speaker had naturally low A1-P0 (correlated with high nasality), and even after the manipulation to raise A1-P0, it was lower than all participants’. If the target of imitation is acoustic, participants diverged from the model (raised A1-P0). But if the target of imitation is a linguistic pattern (decreased coarticulatory nasality), participants converged. The current study uses an AXB perceptual task to determine whether it is acoustically more similar items (more nasal; speakers’ baseline) or linguistically more similar items (less nasal; post-test) that listeners judge as “more similar” to the model talker’s production. A generalized linear mixed model showed a significant preference for post-test items (55.4%, z = 5.19), indicating that linguistic similarity is the basis for listeners’ similarity judgments.

Specificity for coarticulatory vowel nasality in lexical representations.

Surface-level phonetic details are used during word recognition. Yet, questions remain about how these details are encoded in lexical representations and the role of memory and attention during this process. The current study utilizes lexical repetition priming to examine the effect of a delay between hearing a word repeated with either the same or different coarticulatory patterns on lexical recognition. Listeners were faster to recognize repeated words with the same patterns of coarticulatory nasality, confirming that subphonemic information is encoded in the lexicon. Furthermore, when listeners had to adapt to more than one talker, greater coarticulatory specificity in delayed priming was observed suggesting that word-specific encoding of subphonemic details is an active cognitive process.

Imitation of coarticulatory vowel nasality across words and time

ABSTRACTWe investigated phonetic imitation of coarticulatory vowel nasality using an adapted shadowing paradigm in which participants produced a printed word (target) after hearing a different word (prime). Two versions of primes with nasal codas were used: primes with a natural degree of vowel nasality and hypernasalised primes. The version of the prime participants heard varied, whether consistent with their past experience with nasality from the talker or inconsistent, as well as the duration of delay between prime and target. People spontaneously modify coarticulatory nasality to resemble that demonstrated in the prime they were exposed to. Furthermore, this imitation also reflects the degree of nasality demonstrated by overall experience with the speaker’s vowels. The influence of past experience on imitation increases with increased delay between prime and target. Imitation of another speaker appears to involve tracking general articulatory properties about the speaker, and not solely what was specific to the most recent experience.

Phonetic imitation of coarticulatory vowel nasalization.

This study investigates the spontaneous phonetic imitation of coarticulatory vowel nasalization. Speakers produced monosyllabic words with a vowel-nasal sequence either from dense or sparse phonological neighborhoods in shadowing and word-naming tasks. During shadowing, they were exposed to target words that were modified to have either an artificially increased or decreased degree of coarticulatory vowel nasality. Increased nasality, which is communicatively more facilitative in that it provides robust predictive information about the upcoming nasal segment, was imitated more strongly during shadowing than decreased nasality. An effect of neighborhood density was also observed only in the increased nasality condition, where high neighborhood density words were imitated more robustly in early shadowing repetition. An effect of exposure to decreased nasality was observed during post-shadowing word-naming only. The observed imitation of coarticulatory nasality provides evidence that speakers and listeners are sensitive to the details of coarticulatory realization, and that imitation need not be mediated by abstract phonological representations. Neither a communicative account nor a representational account could single-handedly predict these observed patterns of imitation. As such, it is argued that these findings support both communicative and representational accounts of phonetic imitation.

Normalizing nasality? Across-speaker variation in acoustical nasality measures

Although vowel nasality has traditionally been investigated using articulatory methods, acoustical methods have been increasingly used in the linguistic literature. However, while these acoustical measures have been shown to be correlated with nasality, their variability across-speakers is less-well investigated. To this end, we examined cross-speaker variation in coarticulatory vowel nasality in a large collection of multi-speaker English data, comparing measurements in CVC and NVN words at two points per vowel. Two known correlates were analyzed: A1-P0, where A1 is the amplitude of the harmonic under F1, and P0 is the amplitude of a low-frequency nasal peak (Chen 1997), and the bandwidth of F1 (Hawkins and Stevens 1985). Speakers varied both in terms of baseline measurements (e.g., the mean measurements in oral versus nasal vowels), and in the oral-nasal range (the measured difference between oral and nasal). This suggests that although analyzing centered within-speaker differences is unproblematic, the comparison of raw nasality measurements across speakers is unlikely to yield meaningful data. Moreover, this suggests that the perception of vowel nasality may require some process of speaker normalization, much like other aspects of vowel perception. Some suggestions for normalizing nasality measurements in research will be offered.

Phonetic and phonological patterns of nasality in Lakota vowels

Lakota (Siouan) has both contrastive and coarticulatory vowel nasality, and both nasal and oral vowels can occur before or after a nasal consonant. This study examines the timing and degree patterns of acoustic vowel nasality across contrastive and coarticulatory contexts in Lakota, based on data from six Lakota native speakers. There is clear evidence of both anticipatory and carryover nasal coarticulation across oral and nasal vowels, with a greater degree of carryover than anticipatory nasalization. Nasality in carryover contexts is nonetheless restricted: the oral–nasal contrast is neutralized for high back vowels in this context and realized for three of the six speakers in low vowels. In the absence of nasal consonant context, contrastive vowel nasalization is generally greatest late in the vowel. Low nasal vowels in carryover contexts parallel this pattern (despite the location of the nasal consonantbeforethe vowel), and low nasal vowels in anticipatory contexts are most nasal at the start of the vowel. We relate the synchronic patterns of coarticulation in Lakota to both its system of contrast and diachronic processes in the evolution of nasality in Lakota. These data reflect that coarticulatory patterns, as well as contrastive patterns, are grammatical and controlled by speakers.

Nasal coarticulation changes over time in Philadelphia English

This study examines change over time in coarticulatory vowel nasality in both real and apparent time in Philadelphia English. We measure nasal-adjacent vowels in words from a corpus of conversational speech and find systematic, community-level changes in degree of nasal coarticulation over time in Philadelphia. Specifically, in all speakers who were under the age of 25 when interviewed, there is an overall trend of increasing nasality in people born between 1950 and 1965, yet people born after 1965 move towards less nasality than speakers born earlier; finally, those born after 1980 reverse this change, moving again toward greater nasal coarticulation. This finding adds nasality to the set of phonetic dimensions that are demonstrably susceptible to diachronic change in a speech community. The observation that the degree of nasal coarticulation changes towards increased coarticulation at one time period and decreased coarticulation at a different time period adds to the growing body of evidence that subphonemic variation is not universally determined, suggesting instead that it is learned and encoded. Furthermore, the changes in nasality are independent from an observed frequency effect. These empirical patterns suggest that language-internal factors, such as lexical frequency, are independent from language external factors, such as community-level phonetic change over time.

Vowel nasality as a perceptual cue to pharyngeal consonants in moroccan arabic

Pharyngeal consonants have been shown cross-linguistically to be co-articulated with velopharyngeal port opening (Bladon and Al-Bamerni, 1982, Elgendy, 2001). We examined whether coarticulatory vowel nasality is a perceptual cue to an adjacent pharyngeal, and/or nasal, consonant in Moroccan Arabic (MA). Monosyllabic MA words spoken by a native speaker containing a pharyngeal, voiced /ʕ/ or voiceless /ℏ/, or a nasal /n,m/ were cross-spliced with vowels from other contexts creating two conditions per word: “Oral” stimuli were pharyngeals cross-spliced with pharyngealized vowels (i.e., adjacent to /d ʕ/) or nasals cross-spliced with oral vowels (i.e., adjacent to /d/). “Nasal” stimuli were cross-spliced with nasal vowels. The stimuli were presented to nine native MA speakers in a timed lexical repetition task. An RM ANOVA of repetition times showed a significant interaction between nasality, consonant, and directionality [F(7,324) = 3.755, p=0.046]. For anticipatory coarticulation, vowel nasality resulted in faster RTs, indicating facilitated perception for voiced /ʕ/ (t(52) = 1.67, p =.001) but not voiceless /ℏ/. For carryover coarticulation, vowel nasality resulted in faster RTs for /ℏ/ (t(52)=2.05,p = .02), but not /ʕ/. In both directions, vowel nasality resulted in faster RTs for nasals. These results indicate that MA listeners use vowel nasality as a perceptual cue to a word with a pharyngeal segment.

The effects of coarticulatory timing and lexical frequency on vowel nasalization in English: An aerodynamic study.

Recently it has been suggested that a certain degree of variability in coarticulatory vowel nasalization is due to variation in the temporal alignment of nasal and oral gestures for N. In particular, the extent of vowel nasalization in VNC sequences is shown to be inversely related to the duration of the oral gesture for the nasal. Beddor [nasals and nasalization: the relation between segmental and coarticulatory timing, ICPhS (2007) present acoustic data which suggests this to be the case in English; in environments where nasals are shorter, such as VNC[−voice] versus VNC[+voice], anticipatory nasalization is longer. The present study examines vowel nasalization in such environments and attempts to corroborate the aforementioned acoustic findings with aerodynamic data. Additionally, other possible sources of variation in the extent of anticipatory nasalization are explored, namely, the effects of lexical frequency, which has been claimed to correlate negatively rather than positively with coarticulation [carborough (2004).]

Temporal characteristics of coarticulatory vowel nasalization in English

Previous research has shown that, in English /VNC/ sequences, vowels and nasal consonants are shorter when the final oral consonant is voiceless than when voiced [O. Fujimura, Symp. on Articulatory Modeling, Grenoble (1977)]. This study investigates whether consonant voicing affects the temporal characteristics of coarticulatory vowel nasalization, as well as V and N duration. Specifically, is vowel nasalization proportionately longer in voiceless contexts, compensating for the shorter N? Speech materials are multiple instances of real-word and nonsense /bVNC/ sequences (V=/i I e ε æ/, N=/m n ■/, C=/p t k s/ and /b d g z/) produced by four American English speakers. Temporal measures include duration of V, N, C and vowel nasalization. Onset of vowel nasalization was identified as onset of the nasal formant and/or widening of formant bandwidth in the low-frequency region of FFT spectra sampled across the vowel in 10-ms increments. Vowel duration was on average 20% shorter and N duration 40% shorter in voiceless than in voiced contexts. However, vowel nasalization duration was not compensatorily longer in voiceless contexts. Regardless of N duration, coarticulatory nasalization remained a constant proportion (about 80%) of total vowel duration across all segmental contexts. Variation in this proportion across vowels and speakers is minor. [Work supported by NSF.]