ABX Discrimination Research Articles

Distributional learning of non-native tone contrasts by older adults after training and overnight consolidation

Despite decline in psychoacoustic and statistical learning (SL) abilities, older adults demonstrate remarkably intact perceptual learning in both L2 (tone-word learning) and L1 settings (perceptual adaptation to accented/noise-vocoded speech) but show limited transfer of learning to untrained stimuli. This study tests whether perceptual learning is maintained in an implicit statistical learning task where older adults learn L2 tonal contrasts through exposure to probability distributions of tonal tokens, which may pose higher requirements on both psychoacoustic and SL abilities, and whether sleep-dependent consolidation helps the generalization of perceptual knowledge. L1-Cantonese older adults learned to discriminate a perceptually difficult level-falling tone contrast following a pre-test, training, post- training overnight interval. Training stimuli were synthesized by interpolating naturally produced Mandarin level and high-falling tones into six equidistant steps. Participants either heard a bimodal (two-peak resembling level-falling categories) or unimodal distribution (single-peak) consisting of 256 tokens. ABX discrimination task was administered for testing, with tokens by two genders and on two pseudo-syllables to test generalization. Pilot data of 14 participants showed a trend of group effect with the bimodal group outperforming the unimodal group after training and sleep-dependent consolidation, showing that perceptual learning is maintained in a paradigm that relies heavily on psychoacoustic and SL abilities.

Bilingual phonological contrast perception: The influence of Quanzhou Southern Min on Mandarin non-sibilant fricative discrimination.

This study explores the discrimination of Mandarin non-sibilant fricatives by bilingual speakers (N = 40) of Quanzhou Southern Min (L1) and Mandarin (L2) in different phonological contexts, including rounded vowels and the glide [w]. The results of the ABX discrimination task indicate significant contextual effects of the following sound, in line with predictions based on the Perceptual Assimilation model (PAM) [Best (1995). J. Phon. 20(3), 305-330]. Additionally, the observed result could not be fully explained by the acoustic distance between stimuli, and discrimination ability was better for speakers with more exposure to and use of Mandarin.

Perception of English vowel length modulation due to vowel identity and the identity of the following consonant by Korean learners of English

In English, consonant voicing has large effects on the previous vowel duration, as does the status of the vowel as tense or lax. Our aim is to examine the relationship between discrimination and identification of these contrasts by Korean L2 listeners. An ABX discrimination task shows little correlation between the two contrasts, so that listeners who distinguish English consonant voicing contrasts do not tend to do better with the English tense/lax contrasts. An identification task, however, indicates a strong correlation across the two different contrasts such that listeners who identify the correct underlying final consonant are able to identify the correct underlying vowel identity in their English perception. Further, comparing the participants’ performance on the two tasks shows a different relationship between discrimination and identification. For voicing contrasts, discrimination performance needs to reach a level of accuracy before the listeners improve in identification, suggesting that auditory discrimination is a foundation for identification. However, for the tense-lax contrast, the listeners’ performance in the two tasks is entirely uncorrelated, suggesting that part of the difficulty for the listeners acquiring the tense-lax contrast is that they auditorily attend to aspects of the signal which are inappropriate for identification.

Discrimination of L2 British English Monophthong Contrasts: The Case of L2 Thai Learners of English

This article reports on the second language (L2) perception of contrasts among British English monophthongs. This study has two aims: 1) to explore the discriminability of contrasts in L2 British English monophthongs; and 2) to test the perceptual assimilation model-L2 (PAM-L2) towards the ability to discriminate British English contrasts. The contrasts considered were: /iː/-/ɪ/, /æ/-/ʌ/, /ɜː/-/ʌ/, /uː/-/ʊ/, /e/-/æ/, /ʌ/-/ɒ/, /ʊ/-/ɔː/, /ɑː/-/ʌ/, /ɜː/-/ɔː/. Fifty-two native speakers of Thai who were learning English as a foreign language in Thailand participated in a forced-choice ABX discrimination task. The participants were divided between two groups – those high-experienced and those low-experienced. The results evidence how both groups performed well on most contrasts (over 80% correct), except for /ʌ/-/ɒ/. Although the discriminability of the contrast /ʌ/-/ɒ/ was lower than with the other contrasts, the discrimination scores among both groups remained in a middle range (over 70%). No effect of L2 experience was found, thus suggesting that the abilities of both groups did not differ. The PAM-L2 was accurate in predicting that neither group of L2 Thai learners would have difficulty in discriminating the considered L2 sound contrasts. These results imply that the results gained from a perceptual assimilation task are useful in predicting the discriminability of L2 sound contrasts, as suggested by the PAM-L2.

Perception and imitation of prevoicing across language backgrounds

The present study investigates languages users' ability to perceive and imitate prevoicing, testing speakers of languages differing its contrastive status: (1) those where prevoicing serves as a primary cue to the laryngeal contrast (e.g., Portuguese); (2) those where prevoicing never occurs, with aspiration being the primary cue to the contrast (e.g., Cantonese); and (3) English, where prevoicing occurs in free variation. After hearing pairs of words differing in presence/absence of prevoicing, participants were asked to imitate them, followed by an ABX discrimination task. Preliminary findings show above-chance discrimination accuracy as well as significant imitation of the difference, in participants all language backgrounds, with place of articulation, consonantal, and vocalic cues all affecting the extent of prevoicing in production. As expected, speakers of languages where prevoicing is absent showed less overall prevoicing than other groups; however, there were not clear differences across language groups in the faithfulness of imitation, or in discrimination ability. While faithful imitation was usually associated with accurate discrimination, good discrimination was also found on many tokens that were not faithfully imitated, indicating that factors other than perception are necessary to account for variability in imitative ability.

Modified Input Training and Cue Reweighting in Second Language Vowel Perception

Many studies have shown that the High Variability Phonetic Training method can help L2 learners acquire nonnative phoneme contrasts. However, few studies have explored how this paradigm can work using modified input. We investigate L1 Mandarin learners of English as a foreign language and their perception of the tense/lax [i]/[ɪ] contrast. Mandarin learners rely on durational (rather than spectral) cues to distinguish these L2 vowels. In an ABX discrimination task, the experimental group was trained using both natural and synthesized stimuli whose vowels had been systematically lengthened or shortened thus removing the reliability of the durational cue. The control group was trained using natural stimuli only. The pre- and post-tests were identification tasks. An ANCOVA comparing the pre- and post-test identification accuracy revealed that the experimental group improved significantly more than the control group. The modified durational input allows participants to re-weight their processing cues and attend to spectral differences.

Perception and imitation of prevoicing across language backgrounds

The present study investigates speakers’ ability to perceive and imitate prevoicing, testing speakers of languages differing its contrastive status: those where prevoicing serves as a primary cue to the laryngeal contrast (Indo-Aryan languages; “true voicing” languages, e.g., Tagalog), those where prevoicing never occurs (“aspirating” languages, e.g., Cantonese), and those where prevoicing occurs in free variation (other aspirating languages, e.g., English). After hearing pairs of Hindi words differing in presence/absence of prevoicing, participants were asked to imitate them, followed by an ABX discrimination task. Preliminary findings show above-chance discrimination accuracy as well as significant imitation of the difference in participants all language backgrounds, with place of articulation, consonantal, and vocalic cues all affecting the extent of prevoicing in production. As expected, speakers of languages where prevoicing is absent showed less overall prevoicing than other groups; however, there were not clear differences across language groups in discrimination ability. While faithful imitation was usually associated with accurate discrimination, good discrimination was also found on many tokens that were not faithfully imitated, indicating that other (production-based) factors are necessary to account for variability in imitative ability.

On the relationship between perception and production of L2 sounds: Evidence from Anglophones’ processing of the French /u/–/y/ contrast

Previous studies have yielded contradictory results on the relationship between perception and production in second language (L2) phonological processing. We re-examine the relationship between the two modalities both within and across processing levels, addressing several issues regarding methodology and statistical analyses. We focus on the perception and production of the French contrast /u/–/y/ by proficient English-speaking late learners of French. In an experiment with a prelexical perception task (ABX discrimination) and both a prelexical and a lexical production task (pseudoword reading and picture naming), we observe a robust link between perception and production within but not across levels. Moreover, using a clustering analysis we provide evidence that good perception is a prerequisite for good production.

Evaluation of wind farm noise amplitude modulation synthesis quality

Wind farm noise amplitude modulation (WFNAM) is a major contributor to annoyance and could cause sleep disturbance. In laboratory listening experiments assessing its annoyance and sleep disturbance potential, WFNAM stimuli are commonly synthesised and can thus suffer from a lack of ecological validity. Here, five stimuli synthesis methods were compared with measured noise in terms of their perceived similarity. An ABX discrimination listening test and one-third octave band spectra were used for evaluation of the aural and visual similarity, respectively, between the synthesised and measured noise spectra. The results showed that synthesising WFNAM using a simple method can be ecologically valid as listeners could not accurately differentiate between measured and synthesised WFNAM. However, time varying features of WFNAM do play a small but significant role in human perception and therefore hearing test evaluation of synthesis is recommended for obtaining the most ecologically valid synthesised WFNAM.

CAN CAPTIONED VIDEO BENEFIT SECOND LANGUAGE PRONUNCIATION?

AbstractThe present study investigated the potential benefits of extended exposure to captioned videos for second language pronunciation. We tested 90 L2 adult learners of English on speech processing skills (segmentation, speed of lexical access, and sentence processing) and phonological accuracy in perception (ABX discrimination) and production (accentedness ratings) before and after an 8-week treatment consisting of regular exposure to audiovisual materials. Participants were randomly assigned to four experimental conditions involving two viewing modes (captioned or uncaptioned) and two task focus conditions (focus on phonetic form or focus on meaning). Results showed benefits in speech segmentation and speech processing skills irrespective of viewing mode. No significant benefits were found for phonological accuracy in perception. In production, a focus on phonetic form improved pronunciation only in the absence of captions, whereas captioned viewing led to pronunciation gains as long as there was no focus on phonetic form. These findings suggest that pronunciation improvement can take place with the help of captions or, in the absence of captions, when learners’ attention is directed to pronunciation. Cognitive overload might explain why no benefits were obtained when attention was directed to pronunciation in a captioned viewing mode.

Production and perception of English vowels preceding voiced and voiceless consonants by Korean learners of English

In English, consonant voicing has large effects on the previous vowel duration, as does the status of the vowel as tense or lax. Our aim is to examine whether there is an interplay between L1 interference and L2 experience on L2 English vowel productions and perceptions. Korean L2 speakers were chosen as participants; Korean has no contrasts in tense versus lax vowels, and no coda consonant voicing contrast in a monosyllabic structure, but exhibits post-vocalic voicing contrasts in disyllabic structures, and in these cases, voiced consonants exhibit the lengthening of the preceding vowel. The participants were asked to read the list of English nonce words consisting of the English high and mid vowels and sets of plosives contrasting in voicing as a coda in the two different structures. In an ABX discrimination task, the subjects were asked to identify the different words from among minimal triples. None of the speakers exhibited different patterns in both structures in productions or perceptions. All speakers exhibited durational correlates to the voicing contrast, and to the tense-lax distinction. The effect of L2 English experience was not shown in perception task. Overall, there was no correlation between vowel length differences in production and perception scores.

SECOND AND THIRD LANGUAGE LEARNERS’ SENSITIVITY TO JAPANESE PITCH ACCENT IS ADDITIVE

AbstractThis study examines second (L2) and third (L3) language learners’ pitch perception. We test the hypothesis that a listener’s discrimination of and sensitivity (d’) to Japanese pitch accent reflects how pitch cues inform all words a listener knows in an additive, nonselective manner rather than how pitch cues inform words in a selective, Japanese-only manner. Six groups of listeners performed a speeded ABX discrimination task in Japanese. Groups were defined by their L1, L2, and L3 experience with the target language’s pitch cues (Japanese), a language with less informative pitch cues (English), or a language with more informative pitch cues (Mandarin Chinese). Results indicate that sensitivity to pitch is better modeled as a function of pitch’s informativeness across all languages a listener speaks. These findings support cue-centric views of perception and transfer, demonstrate potential advantageous transfer of tonal-L1/L2 speakers, and highlight the cumulative role that pitch plays in language learning.

Development of tonal discrimination in young heritage speakers of Cantonese

This study uses the Perceptual Assimilation Model for Suprasegmentals (PAM-S) (So & Best, 2008, 2010), supported by the assumptions of the L2 Intonation Learning theory (LILt, Mennen, 2015), to investigate how young heritage speakers of Cantonese living in the United States acquired Cantonese tones. Sixty-seven heritage speakers, aged 5–11, were tested on their perception of Cantonese tonal contrasts using an ABX discrimination task. They were compared to 64 peers aged 5–12 in Hong Kong, where Cantonese is spoken as the majority language but English is also acquired from a young age. Two pairs of tones were tested: Tones 2 (mid rising) and 5 (low rising), which have similar pitch heights and contours, and Tones 1 (high level) and 4 (low falling), which have a larger phonetic contrast. As predicted, the heritage speakers were more accurate in discriminating between the more distinct pair of tones than between the more similar pair. They also scored lower than their peers from Hong Kong in both contrast conditions. Age of testing predicted accuracy for both groups, and Chinese literacy also had a significant effect for the heritage speakers. The potential lack of the Tone 2–5 contrast in the heritage speakers’ input is discussed as an explanation for these findings. This study illustrates the divergence in heritage speakers’ phonological development compared to majority language speakers, and shows the relevance of the PAM-S and LILt to the heritage language context.

Effect of back wood choice on the perceived quality of steel-string acoustic guitars.

Some of the most prized woods used for the backs and sides of acoustic guitars are expensive, rare, and from unsustainable sources. It is unclear to what extent back woods contribute to the sound and playability qualities of acoustic guitars. Six steel-string acoustic guitars were built for this study to the same design and material specifications except for the back/side plates which were made of woods varying widely in availability and price (Brazilian rosewood, Indian rosewood, mahogany, maple, sapele, and walnut). Bridge-admittance measurements revealed small differences between the modal properties of the guitars which could be largely attributed to residual manufacturing variability rather than to the back/side plates. Overall sound quality ratings, given by 52 guitarists in a dimly lit room while wearing welder's goggles to prevent visual identification, were very similar between the six guitars. The results of a blinded ABX discrimination test, performed by another subset of 31 guitarists, indicate that guitarists could not easily distinguish the guitars by their sound or feel. Overall, the results suggest that the species of wood used for the back and sides of a steel-string acoustic guitar has only a marginal impact on its body mode properties and perceived sound.

Newly learned categories induce pre-attentive categorical perception of faces

Face perception is modulated by categorical information in faces, which is known as categorical perception (CP) of faces. However, it remains unknown whether CP of faces is humans’ inborn capability or the result of acquired categories. Here, we examined whether and when newly learned categories affect face perception. A short-term training method was employed in which participants learned new categories of face stimuli. Behaviorally, using an AB-X discrimination task, we found that the discrimination accuracy of face pairs from different learned categories was significantly higher than that of faces from the same category. Neurally, using a visual oddball task, we found that deviant stimuli whose category differed from standard stimuli evoked a larger N170. Importantly, the visual mismatch negativity (vMMN), starting from 140 ms after stimuli onset, was stronger with the between-category deviants than with the within-category deviants under the unattended condition. Altogether, our study provides empirical evidence indicating that CP of faces could be induced by newly learned categories, and this effect occurs automatically during an early stage of processing.

Guitarists' evaluation and discrimination of steel-string acoustic guitars built with back/side woods of varying price, prestige, and sustainability

The different woods used for the back plates of acoustic guitars are often compared by guitarists for their sound qualities, but these comparisons are rarely done under blinded conditions. For this experiment, six steel-string acoustic guitars were built to be as similar as possible except for the woods used for the backs and sides. Bridge admittance measurements and spectral analyses of acoustic recordings revealed small differences between the guitars in their low-frequency modes. Fifty-two experienced guitar players rated the guitars for sound quality, playability, and other perceptual attributes while wearing welder’s goggles to prevent visual identification. The guitars received on average similar ratings for sound quality and playability. A factor analysis showed that the other perceptual attributes clustered around the dimensions of “clarity,” “warmth,” and “loudness,” which were all positively related to perceived sound quality, and did not differ significantly between the guitars. An ABX discrimination test performed by a subset of 31 guitarists showed that guitarists could not easily distinguish the guitars by their sound. Overall, the results suggest that the woods used for the back plate of a steel-string acoustic guitar have only a minimal influence on its perceived sound, despite varying greatly in price, prestige, and sustainability.

Perceptual thresholds for non-ideal diffuse field reverberation.

The objective of this study is to understand listeners' sensitivity to directional variations in non-ideal diffuse field reverberation. An ABX discrimination test was conducted using a semi-spherical 28-loudspeaker array; perceptual thresholds were estimated by systematically varying the level of a segment of loudspeakers for lateral, height, and frontal conditions. The overall energy was held constant using a gain compensation scheme. When compared to an ideal diffuse field, the perceptual threshold for detection is -2.5 dB for the lateral condition, -6.8 dB for the height condition, and -3.2 dB for the frontal condition. Measurements of the experimental stimuli were analyzed using a Head and Torso Simulator as well as with opposing cardioid microphones aligned on the three Cartesian axes. Additionally, opposing cardioid measurements made in an acoustic space demonstrate that level differences corresponding to the perceptual thresholds can be found in practice. These results suggest that non-ideal diffuse field reverberation may be a previously unrecognized component of spatial impression.

The perceptual effects of tonewood choice for the back/side plates of the steel-string acoustic guitar

The sound qualities of different tonewoods used for the back/side plates of acoustic guitars are the subject of intense debate among guitarists. However, comparisons between different guitars are rarely performed under blinded conditions. We asked a sample of professional, semi-professional, and amateur guitarists (n = 52) to rate the sound quality of six steel-string acoustic guitars, built to be as similar as possible except for the woods used for the back/side plates, which were: Brazilian rosewood, Indian rosewood, mahogany, maple, sapele, and walnut. The guitarists played the guitars in a dimly lit room while wearing welder’s goggles to prevent visual identification. The sound quality ratings did not differ significantly between guitars. Test-retest correlations, measured for a subset of guitarists (n = 34), were low, suggesting that the lack of significant rating differences was not due to averaging out contrasting individual preferences. The results of a blinded ABX discrimination test, performed by a subset of guitarists (n = 31), indicate that guitarists could not easily distinguish the guitars by their sound or feel. Overall, the results suggest that the species of wood used for the back/side plates of a steel-string acoustic guitar may play a lesser role in perceived sound quality than is commonly believed.

The role of experience in children's discrimination of unfamiliar languages.

Five- and six-year-old children (n = 160) participated in three studies designed to explore language discrimination. After an initial exposure period (during which children heard either an unfamiliar language, a familiar language, or music), children performed an ABX discrimination task involving two unfamiliar languages that were either similar (Spanish vs. Italian) or different (Spanish vs. Mandarin). On each trial, participants heard two sentences spoken by two individuals, each spoken in an unfamiliar language. The pair was followed by a third sentence spoken in one of the two languages. Participants were asked to judge whether the third sentence was spoken by the first speaker or the second speaker. Across studies, both the difficulty of the discrimination contrast and the relation between exposure and test materials affected children’s performance. In particular, language discrimination performance was facilitated by an initial exposure to a different unfamiliar language, suggesting that experience can help tune children’s attention to the relevant features of novel languages.

Experimental approaches to the production and perception of prosody.

This Special Issue brings together a selection of papers orally presented at the conference Phonetics and Phonology in Iberia (PaPI), held in Lisbon in 2013 (labfon.letras.ulisboa.pt/papi2013). For 10 years the PaPI conferences have been a highly productive setting for the sharing and discussion of investigation within phonology and phonetics and their interactions, and have undoubtedly played their part in the development of experimental work in the field of prosody (see in particular Frota, Vigario, & Freitas, 2005; Prieto, Mascaro, & Sole, 2007; Romero & Riera, in press; Vigario, Frota, & Freitas, 2009).Prosodic research has witnessed remarkable progress in the last few decades. The limits of the field are pushed farther constantly as new investigations appear, powered by more elaborate experimental methods, new technology and resources, innovative bridges promoting knowledge transfer across disciplines, increasingly sophisticated theoretical models and a continuously evolving set of questions.This volume also reflects the productivity of experimental approaches to prosody. It compiles work addressing a range of questions that are of current interest in prosody, exploring topics such as the phonetic exponents of rhythm and information structure in language development, languageparticular factors in the perception of word stress, social and situational factors underlying speakers' pitch accent selection and the phonology of fine grain phonetic variation in accent realization. All of the papers follow a laboratory-based approach to phonology and a variety of experimental methods are employed to elicit controlled production data as well as to investigate perception, including tasks involving picture-matching and other types of interactive games, discourse completion, ABX discrimination, sequence recall and context matching acceptability judgements.Two papers investigate prosodic development, one contrasting child and adult Dutch monolinguals and the other comparing English and Spanish monolingual and simultaneous English- Spanish bilingual children.In the paper 'Quiet is the New Loud: Pausing and Focus in Child and Adult Dutch', Romoren and Chen report on an experiment that tested the use of pauses to signal focused material in Dutch, both in adults and in 4 to 5 year-olds. Adults tend to use accent and post-focal deaccentuation to mark focus. However, previous research has shown that adult speakers of Dutch also use pauses before focus and that children at this age do not fully master adult focus prosody (Chen, 2007, 2011). The experiment elicited sentences with focus structure by means of a picture-matching game and combining controlled speech contexts with wh-questions. Results showed a significant difference in pause durations before focal and non-focal targets, the pause being longer in the former, both in adults and in children's productions. Nevertheless, the use of pauses was found to be more systematic in children. The authors interpret this difference as a consequence of the child's stage in language development. Research on the development of information structure signalling, using controlled and efficient elicitation methods, is fairly rare, and to our knowledge this is the first investigation on pausing as a means to cue focus in child speech.In the paper by Schmidt and Post ('The Development of Prosodic Features and their Contribution to Rhythm Production in Simultaneous Bilinguals'), the impact of bilingualism in terms of facilitory and inhibitory effects are investigated in relation to the acquisition of prosodic features. Focusing on phrase-final lengthening and accentuation in English and Spanish, two areas where the rhythmic differences between the two languages manifest, the authors demonstrate that the two languages can be rhythmically separated at 4 years, and clearly separable by 6 years of age, and that bilinguals have an advantage over monolinguals in the prosodic acquisition of the structurally more complex language (English). …