English Listeners Research Articles

Understanding speech in everyday conditions—Effects of room acoustics and noise

Understanding speech in noisy, indoor settings is a considerable challenge for listeners. Room acoustics, such as reverberation and heating, ventilation, and air conditioning (HVAC) noise, all contribute to such challenges. In the current study, the effects of different backgrounds were examined in anechoic versus reverberant conditions at two different signal-to-noise ratios (SNRs) were examined. 60 monolingual American English listeners transcribed spoken English sentences under different noisy conditions. Overall, we tested two acoustic conditions: anechoic and reverberant, reverberation time = 1.5 s, with four speech conditions (target speech only, or combined with HVAC noise, two-talker babble, or both) and two different SNRs (0 dB and -3 dB). Both the target and babble speech were female voices speaking in English. Complex backgrounds resulted in decreased performance, and the detrimental effects of such masking were accentuated by reverberation and lower SNR. This study serves as a basis for future research exploring the effects of room acoustics on speech intelligibility to gain a more wholistic understanding of speech perception under realistic conditions.

Quantitative assessment of English accent variation: Levenshtein distance and perceptual identification

This is a perceptual study with a basis in acoustic phonetics. It investigates the extent to which accents of English can be quantified by the Levenshtein distance (LD), a string metric. LD defines string similarity by the minimum number of character edits required to transform one string into another. Narrowly transcribed speech samples of English provide the phonetic strings required for LD comparison. For the purposes of the current experiment, 24 speakers representing four English accent varieties are included: Midland American (control), British/Australian, Hindi-influenced, and Mandarin-influenced. High- and low-pass filters are used to augment or attenuate perceptual contribution of high- versus low-frequency information in the speech signal. Predictions stem from published correlations between LD and listeners’ perceptional ratings of intelligibility and native-likeness. For a group of monolingual American English listeners, confusion is expected between Midland American and British/Australian accents due to similarly low LDs, further intensified by low-pass filtering. However, fewer confusions are predicted between Hindi- and Mandarin-influenced English, due to varying L1 influences and the absence/presence of tonal information. This research seeks to explore accent perception, investigating the relationship between manipulated aspects of the speech signal and listener identification.

The Effect of Pitch Accent on the Perception of English Lexical Stress: Evidence from English and Mandarin Chinese Listeners

The relative weighting of f0 and vowel reduction in English spoken word recognition at the sentence level were investigated in one two-alternative forced-choice word identification experiment. In the experiment, an H* pitch-accented or a deaccented word fragment (e.g., AR- in the word archive) was presented at the end of a carrier sentence for identification. The results of the experiment revealed differences in the cue weighting of English lexical stress perception between native and non-native listeners. For native English listeners, vowel quality was a more prominent cue than f0, while native Mandarin Chinese listeners employed both vowel quality and f0 in a comparable fashion. These results suggested that (a) vowel reduction is superior to f0 in signaling initial stress in the words and (b) f0 facilitates the recognition of word initial stress, which is modulated by first language.

Listening effort for L2 accents: Adults’ adaptation to Mandarin accented english

Using task-evoked pupillary response (TEPR) and dual-task procedures, prior research has shown that adults recruit additional cognitive resources when processing fully intelligible L2-accented speech compared to speech in their own L1 accent (McLaughlin and Van Engen, 2020; Brown et al., 2020). In this study, we investigated the relationship between three prosodic measures (relative word duration variance, pitch stability, and pitch range) and changes in listening effort for L1 and L2 speakers over the course of an experiment. Using growth curve analysis, we found significant interactions between each of the three measures and trial number, indicating that prosodic factors influence speech adaptation. However, these interactions were in opposite directions for the L1 and L2 speakers. Thus, adult monolingual English listeners rely on prosodic information to adapt to a particular speaker’s voice, but use that information differently based on speaker identity.

Variability in cross-language and cross-dialect perception. How Irish and Chinese migrants process Australian English vowels.

We investigate how three adult groups- experienced L2 English listeners; experienced D2 (second dialect) listeners; and native L1/D1 listeners- categorise Australian English (AusE) lax front vowels /ɪ e æ/ in /hVt/, /hVl/ and /mVl/ environments in a forced-choice categorisation task of synthesised continua. In study 1, AusE listeners show predictable categorisations, with an effect of coarticulation raising the vowel in perception for nasal onset stimuli, and a following lateral lowering the vowel in perception. In study 2, Irish (D2) and Chinese listeners (L2) have different categorisations than AusE listeners, likely guided by their D1/L1. Coarticulation influences the D1/D2 groups in similar ways, but results in more difficulty and less agreement for the Chinese. We also investigate the role of extralinguistic factors. For the Chinese listeners, higher proficiency in English does not correlate with more Australian-like categorisation behaviour. However, having fewer Chinese in their social network results in more Australian-like categorisation for some stimuli. These findings lend partial support to the role of experience and exposure in L2/D2 contexts, whereby categorisation is likely still driven by native categories, with increased exposure leading to better mapping, but not to a restructuring of underlying phonetic categories.

The relation between perceived non-native features in the L1 speech of English migrants to Austria and their phonetic manifestation in L1 productions

Aims: The purpose of this research was to investigate to what extent the most commonly identified non-native features in the L1 speech of late consecutive bilinguals are reflected in differences in the bilinguals’ productions of these features compared with those of monolingual speakers of the L1. Design: We investigated the L1 accent of English migrants to Austria and monolingual English speakers in the United Kingdom in two inter-related studies. Data: In Study 1, an accent-perception experiment, native English listeners rated the nativeness of the monolinguals’ and bilinguals’ L1 English productions of read sentences, using a 6-point scale, and subsequently commented on the accentual features associated with perceived non-nativeness. In Study 2, the monolinguals’ and bilinguals’ productions of the most commonly identified non-native features from Study 1 were compared acoustically and auditorily. Findings: The accent-perception experiment revealed significantly higher non-nativeness ratings for the bilinguals than the monolinguals. These were associated with a wide range of segmental and prosodic features ( n = 836 feature tokens), with /iː/ and /ɪ/ the most commonly identified for segments, and articulation rate and intonation for prosody. The phonetic analysis of these features in Study 2, in turn, revealed that the bilinguals produced /ɪ/ closer to /iː/ than the monolinguals and had more final rises in questions, with the articulation rate differences between them just failing to reach significance. Originality: This is the first study to document a direct link between the features perceived as non-native in bilinguals’ L1 speech and their measurable manifestation in non-native L1 speech productions. Significance: This research contributes significantly to our understanding of the relationship between the perception and production of attrition in L1 speech, and shows that where listeners are able to provide consistent, detailed descriptions of the features associated with non-nativeness, they truthfully reflect measurable patterns in the speech signal.

The perception of vowelless words in Tashlhiyt

This study examines the perceptual mechanisms involved in the processing of words without vowels, a lexical form that is common in Tashlhiyt but highly dispreferred cross-linguistically. In Experiment 1, native Tashlhiyt and non-native (English-speaking) listeners completed a paired discrimination task where the middle segment of the different-pair was either a vowel (e.g., fan vs. fin), consonant (e.g., ʁbr vs. ʁdr), or vowelless vs. voweled contrast (e.g., tlf vs. tuf). Experiment 2 was a word-likeness ratings task of tri-segmental nonwords constructed to vary in the sonority of the middle segment. We find that vowelless words containing different types of sonority profiles were generally highly discriminable by both native and non-native listeners. This can be explained by the phonetic and acoustic properties of vowelless words: Since Tashlhiyt exhibits low consonant-to-consonant coarticulation, the presence of robust consonantal cues in the speech signal means that the internal phonological structure of vowelless words is recoverable by listeners. At the same time, word-likeness ratings of nonwords indicated that listeners relied on their native-language experience to judge the wellformedness of new words: Tashlhiyt listeners were most likely to accept obstruent-centered vowelless words; meanwhile, English listeners’ preferences increased with higher sonority values of the word center. Across both experiments, speech style variation provided further evidence as to how the phonetic implementation of vowelless words makes them perceptually stable. Thus, our findings provide an overview of the low-level acoustic-phonetic and higher-level phonological processing mechanisms involved in the perception of vowelless words. Our results can inform understandings of the relationship between language-specific phonetic variation and phonotactic patterns, as well as how auditory processing mechanisms shape phonological typology.

From first encounters to longitudinal exposure: a repeated exposure-test paradigm for monitoring speech adaptation.

Perceptual difficulty with an unfamiliar accent can dissipate within short time scales (e.g., within minutes), reflecting rapid adaptation effects. At the same time, long-term familiarity with an accent is also known to yield stable perceptual benefits. However, whether the long-term effects reflect sustained, cumulative progression from shorter-term adaptation remains unknown. To fill this gap, we developed a web-based, repeated exposure-test paradigm. In this paradigm, short test blocks alternate with exposure blocks, and this exposure-test sequence is repeated multiple times. This design allows for the testing of adaptive speech perception both (a) within the first moments of encountering an unfamiliar accent and (b) over longer time scales such as days and weeks. In addition, we used a Bayesian ideal observer approach to select natural speech stimuli that increase the statistical power to detect adaptation. The current report presents results from a first application of this paradigm, investigating changes in the recognition accuracy of Mandarin-accented speech by native English listeners over five sessions spanning 3 weeks. We found that the recognition of an accent feature (a syllable-final /d/, as in feed, sounding/t/-like) improved steadily over the three-week period. Unexpectedly, however, the improvement was seen with or without exposure to the accent. We discuss possible reasons for this result and implications for conducting future longitudinal studies with repeated exposure and testing.

Alignment in ASR and L1 Listeners’ Recognition of L2 Learner Speech: French EFL Learners & Dictation.Io

This study is an extension of Inceoglu et al.’s (2023) study on Google Voice Typing as a pronunciation learning tool. We used the Automatic Speech Recognition (ASR) tool on the dictation.io website (Agarwal, 2022), and our participants were L2 English learners of a different L1, but similar proficiency level. Twelve L1 English listeners assessed the L2 English from four L1 French speakers in terms of intelligibility and comprehensibility, measured by word transcription and Likert scale ratings respectively. Their scores were compared to ASR output. The goal was to determine how accurate the tool is, and to what extent its accuracy correlates with human listeners. The results were generally consistent with those of Inceoglu et al. (2023), with few exceptions which we discuss in the current study.

Performance of single-channel speech enhancement algorithms on Mandarin listeners with different immersion conditions in New Zealand English

Speech enhancement (SE) is a widely used technology to improve the quality and intelligibility of noisy speech. So far, SE algorithms were designed and evaluated on native listeners only, but not on non-native listeners who are known to be more disadvantaged when listening in noisy environments. This paper investigates the performance of five widely used single-channel SE algorithms on early-immersed New Zealand English (NZE) listeners and native Mandarin listeners with different immersion conditions in NZE under negative input signal-to-noise ratio (SNR) by conducting a subjective listening test in NZE sentences. The performance of the SE algorithms in terms of speech intelligibility in the three participant groups was investigated. The result showed that the early-immersed group always achieved the highest intelligibility. The late-immersed group outperformed the non-immersed group for higher input SNR conditions, possibly due to the increasing familiarity with the NZE accent, whereas this advantage disappeared at the lowest tested input SNR conditions. The SE algorithms tested in this study failed to improve and rather degraded the speech intelligibility, indicating that these SE algorithms may not be able to reduce the perception gap between early-, late- and non-immersed listeners, nor able to improve the speech intelligibility under negative input SNR in general. These findings have implications for the future development of SE algorithms tailored to Mandarin listeners, and for understanding the impact of language immersion on speech perception in noise.

Exploring effects of social information on talker-independent accent adaptation.

The present study examined whether race information about speakers can promote rapid and generalizable perceptual adaptation to second-language accent. First-language English listeners were presented with Cantonese-accented English sentences in speech-shaped noise during a training session with three intermixed talkers, followed by a test session with a novel (i.e., fourth) talker. Participants were assigned to view either three East Asian or three White faces during training, corresponding to each speaker. Results indicated no effect of the social priming manipulation on the training or test sessions, although both groups performed better at test than a control group.

Listener beliefs and perceptual learning: Differences between device and human guises

Abstract: Listeners have a remarkable ability to adapt to novel speech patterns, such as a new accent or an idiosyncratic pronunciation. In almost all of the previous studies examining this phenomenon, the participating listeners had reason to believe that the speech signal was produced by a human being. However, people are increasingly interacting with voice-activated artificially intelligent (voice-AI) devices that produce speech using text-to-speech (TTS) synthesis. Will listeners also adapt to novel speech input when they believe it is produced by a device? Across three experiments, we investigate this question by exposing American English listeners to shifted pronunciations accompanied by either a ‘human’ or a ‘device’ guise and testing how this exposure affects their subsequent categorization of vowels. Our results show that listeners exhibit perceptual learning even when they believe the speaker is a device. Furthermore, listeners generalize these adjustments to new talkers, and do so particularly strongly when they believe that both old and new talkers are devices. These results have implications for models of speech perception, theories of human-computer interaction, and the interface between social cognition and linguistic theory.

Performance of speech enhancement algorithms on the speech intelligibility of native Mandarin listeners immersed in English-speaking environment

Speech enhancement algorithms have been developed to improve speech intelligibility for listeners under noisy conditions. However, all existing algorithms were evaluated only by native listeners, the performance of such algorithms on non-native listeners was rarely investigated. This study conducts a subjective listening test on native New Zealand English listeners and native Mandarin listeners who have been immersed in New Zealand English-speaking environment for more than one year. The participants were asked to transcribe noisy English sentences processed by five widely used single-channel speech enhancement algorithms. The speech intelligibility of the two groups was quantified and compared to investigate the effectiveness of the speech enhancement algorithms on non-native listeners who are familiar with the target language.

Effect of room acoustics on speech intelligibility under noise between native and non-native listeners

The current study investigates the effect of room acoustics on the perception of speech between native and non-native listeners of English. The study particularly focuses on how the benefit of spatial release from masking (SRM) is affected by room acoustics for the different listener groups. The result of subjective listening test shows that SRM is observed both from native and non-native listeners, however the drop in the speech intelligibility is more significant in the non-native listeners when the sound sources are located farther away from the listener.

Insights into L2 connected speech segmentation: A gating experiment with listeners of different English proficiency levels

The relative contribution of bottom-up (i.e. acoustic-phonetic) and top-down (i.e. contextual) cues for successful L2 online segmentation is still a matter of debate. This study used the gating paradigm to investigate the segmentation processes of adult L2 English listeners with different proficiency levels, by looking at the type of cues they exploit and how they revise their hypotheses as connected speech is progressively revealed. Twenty-one French and Tunisian undergraduates were selected from a larger pool (n=226) and identified as skilled (n=11) and unskilled (n=10) listeners based on their scores on standardized English listening and vocabulary tests. Descriptive statistics, analysis of variance and qualitative analysis were performed on the obtained data. Overall, this study provides supporting L2 evidence for the hierarchical nature of the multiple speech segmentation cues (Mattys et al., 2005). The results indicated an early effect of context on segmentation that was independent of L2 proficiency when the context is constraining. In non-constraining contexts, successful segmentation is delayed for both groups with L2 unskilled listeners needing far more bottom-up information to process input and revise their segmentation hypotheses. We conclude that, in online L2 speech segmentation, what distinguishes proficient from non-proficient listeners is their efficient processing of bottom-up cues. Pedagogical implications are provided hoping to help L2 English teachers (and materials developers) focus on bottom-up training to improve their learners’ real-time comprehension competence.

Listener beliefs and perceptual learning: Differences between device and human guises

Listeners have a remarkable ability to adapt to novel speech patterns, such as a new accent or an idiosyncratic pronunciation. In almost all of the previous studies examining this phenomenon, the participating listeners had reason to believe that the speech signal was produced by a human being. However, people are increasingly interacting with voice-activated artificially intelligent (voice-AI) devices that produce speech using text-to-speech (TTS) synthesis. Will listeners also adapt to novel speech input when they believe it is produced by a device? Across three experiments, we investigate this question by exposing American English listeners to shifted pronunciations accompanied by either a ‘human’ or a ‘device’ guise and testing how this exposure affects their subsequent categorization of vowels. Our results show that listeners exhibit perceptual learning even when they believe the speaker is a device. Furthermore, listeners generalize these adjustments to new talkers, and do so particularly strongly when they believe that both old and new talkers are devices. These results have implications for models of speech perception, theories of human-computer interaction, and the interface between social cognition and linguistic theory

Australian English listeners' perception of Japanese vowel length reveals underlying phonological knowledge.

Speech perception patterns are strongly influenced by one's native phonology. It is generally accepted that native English listeners rely primarily on spectral cues when perceiving vowels, making limited use of duration cues because English lacks phonemic vowel length. However, the literature on vowel perception by English listeners shows a marked bias toward American English, with the phonological diversity among different varieties of English largely overlooked. The current study investigates the perception of Japanese vowel length contrasts by native listeners of Australian English, which is reported to use length to distinguish vowels unlike most other varieties of English. Twenty monolingual Australian English listeners participated in a forced-choice experiment, where they categorized Japanese long and short vowels as most similar to their native vowel categories. The results showed a general tendency for Japanese long and short vowels (e.g., /ii, i/) to be categorized as Australian English long and short vowels (e.g., /i:, ɪ/ as in "heed," "hid"), respectively, which contrasts with American English listeners' categorization of all Japanese vowels as tense regardless of length (e.g., /ii, i/ as both "heed") as reported previously. Moreover, this duration-based categorization was found not only for Australian English categories that contrast in duration alone (e.g., /ɐ:, ɐ/ as in "hard," "hud") but also for those that contrast in both duration and spectra (e.g., /o:, ɔ/ as in "hoard," "hod"), despite their spectral mismatch from the corresponding Japanese vowels (e.g., /aa, a/ and /oo, o/). The results, therefore, suggest that duration cues play a prominent role across all vowel categories-even nonnative-for Australian English listeners. The finding supports a feature-based framework of speech perception, where phonological features like length are shared across multiple categories, rather than the segment-based framework that is currently dominant, which regards acoustic cues like duration as being tied to a specific native segmental category. Implications for second and foreign language learning are discussed.

Sociophonetic Variation in Vowel Categorization of Australian English.

This study involves a perceptual categorization task for Australian English, designed to investigate regional and social variation in category boundaries between close-front vowel contrasts. Data are from four locations in southeast Australia. A total of 81 listeners from two listener groups took part: (a) so-called mainstream Australian English listeners from all four locations, and (b) L1 Aboriginal English listeners from one of the locations. Listeners heard front vowels /ɪ e æ/ arranged in 7-step continua presented at random. Varied phonetic contexts were analyzed, with a focus on coda /l/ because of a well-known prelateral merger of /e æ/ through mid-vowel lowering (e.g., celery-salary) reported to occur in some communities in this part of Australia. The results indicate that regional variation in Australian English is evident in perception. In particular, merging of /el/-/æl/ is shown to occur in the southernmost regions analyzed, but rarely in the northern regions of the geographical area under investigation. Aside from regional variation observed, age was also a factor in how participants responded to the task: older speakers had more merger than younger speakers in many locations, which is a new finding-previously, the merger was thought to be increasing in frequency over time, yet here we see this in only one location. Aboriginal English listeners also responded differently when compared with mainstream Australian English listeners. By analyzing the perception results across a variety of regional locations, with data from two different Australian social groups in the same location, this study adds a new dimension to our understanding of regional and social variations in Australian English.

The role of pragmatic markers in perceptions of L2 fluency in dialogue

This matched-guise study helps identify the features of interaction that contribute to perceptions of fluency in L2 dialogue. We focus on pragmatic markers (PM) that occur frequently in L1 interaction but relatively less so in L2 English learners' speech (Fung & Carter, 2007), namely, interpersonal (e.g., I see, okay) and cognitive PMs (e.g., I mean, well, I think). Despite evidence of their functional value in interaction (McCarthy, 2009), they are often considered markers of disfluency and rarely introduced in L2 classrooms. Speech samples of highly proficient L2 interaction were digitally manipulated to control for interpersonal and cognitive PMs simulating expert speaker-like PM use. L1 and L2 English listeners (N = 231) rated and commented upon interlocutor's fluency. The quantitative results suggest that listeners generally perceived speakers as more fluent when using PMs, although the magnitude of the effect was relatively small and driven by L1 raters. Raters' qualitative comments suggest covert influence of PM use on fluency perceptions and uncovered a bias favoring English L1 speaker PM use. Results support the potential value of explicitly addressing PMs in teaching L2 fluency and help researchers in understanding the construct of spoken fluency in dialogue to guide in its measurement and assessment.

Categorization and discrimination of Mandarin lexical tones by naïve English listeners

Unlike tone languages such as Mandarin, English lacks tones at the sub-lexical level. Accordingly, English listeners have difficulty perceptually assimilating tones as categorized or uncategorized native segments (Perceptual Assimilation Model, PAM: Best, 1995). While English listeners can categorize the four lexical tones of Mandarin, i.e., level contour, rising, dipping, and falling, when given question, statement, exclamation, and uncertainty intonations as category choices (So & Best, 2011, 2014), this does not address tone assimilation at the segmental level. We reasoned that they might assimilate tones as non-assimilable nonspeech patterns if given visual icons as tone category choices (flat, rising, dipping, and falling lines, respectively), with no reference being made to English intonation categories. Accordingly, 76 monolingual English listeners (M age = 24.85 years, 50 females) were set two tasks: to use visual icons to categorize Mandarin tones in naturally produced tone-words (/ga, ti, tu, pu/ × 4 tones) and to discriminate all six pairwise tone contrasts. All tone pairs showed ceiling-level discrimination, and listeners split their categorizations of falling and level stimuli between the falling and flat icons, suggesting that when given visual icons, tone-naïve English listeners perceive Mandarin tones as nonspeech acoustic patterns, which is consistent with PAM’s non-assimilation predictions.