Language and Speech最新文献

This study investigates how speakers adapt their use of disfluencies in public versus private speech settings. Existing studies suggest systematic differences in disfluency rates, depending on who we are communicating with, how interactive the communication is, how difficult the topic is, whether the interaction is broadcast or not, and whether the speech is pre-scripted or not. We aim to improve this understanding through analysis in the Slovenian language, using data from the Training Corpus of Spoken Slovenian ROG-Artur. We investigate whether quantitative differences in the use of disfluency exist between private and public speech, and aim to explain these differences by investigating the relationship between disfluency functions and the physical, social, cognitive, and other factors influencing communication behavior. Our results revealed significant differences in disfluency patterns: disfluencies, in general, are more frequent in private speech, whereas filled pauses, unrepaired pronunciations and blocks, are more common in public speech. We group disfluency functions into two general categories. In contextual analysis, we interpret that speakers reduce disfluencies in public speech due to its high relevance, formal expectations, partial pre-scripting, time constraints and advanced speaker skills, while the higher frequency of filled pauses, unrepaired pronunciations and blocks in public speech reflect the impact of longer dialog turns, time constraints and emotional stress. The findings of this study should be interpreted with caution, given the interpretative nature of qualitative analysis and the potential confounding effect of the involvement of different speakers in the public and private speech samples.

Numerous studies have established that prosody plays an important role in expressing meanings and functions. However, it remains unknown whether prosody is employed to convey the distinction between subjective causality (CLAIM-ARGUMENT) and objective causality (CONSEQUENCE-CAUSE). This study aimed to address this issue in English, where both types of causality are typically expressed using the same connective. Two production experiments were conducted, focusing on causality in backward order (Q "because" P) and in forward order (P "so" Q), respectively. The results show that subjective causality exhibited a larger F0 range, less integrated prosody, and a distinctive F0 contour shape compared with objective causality. These findings highlight the role of prosody in expressing subjective and objective causality in the absence of explicit lexical markers in English.

This study investigates young school-aged children's knowledge (at 4-7 years) of accurate English word-initial onset clusters. By this age, we expect children to be mostly accurate in producing #CC clusters (rather than repairing them with deletion or epenthesis). We ask how well can they recognize and reject cluster repair errors, in both real and nonce word tasks. The results suggest that these learners' cluster judgment skills lag behind their cluster production abilities, and that asymmetries in error types do not overall align between the two domains. Perceptual errors are made most often when comparing clusters with epenthesis repairs, not deletion, and the cluster's sonority profile does not directly influence error rates. After comparing these findings with similar results from adult L2 English speakers as well, we discuss the ways in which issues like recoverability, salience, and contiguity can account for our findings. We also suggest that more work on phonological knowledge and judgments in older children will provide a broader understanding of sound pattern acquisition across development.

To segment words in unfamiliar speech, listeners are known to exploit both native prosodic cues and statistical cues available in the speech signal. However, how and when these cues are combined remains a matter of debate. Here, we studied how transitional probabilities (TPs) and prosodic phrasal boundaries are combined by French speakers to segment words. Since French does not have lexical stress, prosodic phrasal boundaries unambiguously signal word boundaries, providing a unique possibility to test whether prosodic cues can overcome statistical ones, and constrain further statistically based segmentation. We tested French adults in an artificial speech segmentation task, manipulating the consistency between prosodic and TP cues, signaling either the same or different word boundaries. Results showed that participants favored prosodic phrasal boundaries over TPs, regardless of exposure time to the speech stream (Experiment 1: 3.5 minutes; Experiment 2: 7 min), supporting a prosodically driven statistical segmentation of the speech stream.

Separate traditions of research have examined the impact of linguistic factors and social factors on the intelligibility, comprehensibility, and accentedness of second language (L2) speech, but studies that simultaneously investigate social and linguistic factors are rarely conducted on L2 languages other than English and outside of Western social and cultural environments. This study explores the effects of utterance-level prosody and speaker ethnicity on perception of L2 Mandarin Chinese speech. First language (L1) Mandarin listeners (n = 292) were asked to select the correct transcriptions of each of six sentences spoken by two male L2 Mandarin speakers who differed in their prosodic accuracy. While listening to each set of sentences, a picture of an Asian face or a White face was displayed on the listener's screen. Results indicate that participants were significantly more likely to select the correct transcription of each sentence both when they heard the speaker with high prosodic accuracy and when they believed that the speaker was ethnically Chinese. Listeners also rated speakers' comprehensibility, accentedness, and perceived personal characteristics; listeners rated a speaker with higher prosodic accuracy or believed to be ethnically Chinese as more comprehensible, less accented, and higher on perceived personal characteristics. This study demonstrates that a link between linguistic and social factors exists in processing L2 speech, even outside of the social, cultural, and linguistic environments typically used as a setting for investigation of L2 speech perception, and it explores implications for L2 Mandarin pronunciation teaching.

In this study, we tested whether the perception of pitch contours within a lexical pitch accent can be better understood through tonal targets in the Autosegmental-Metrical (AM) model or as an entire tonal configuration identification. Specifically, a categorization experiment was conducted to see how South Kyungsang Korean (SKK) listeners perceive their high (H) and rising (LH) lexical pitch accents. Auditory stimuli were manipulated depending on H peak alignment (earlier vs. later), rise shape (domed or "convex" vs. scooped or "concave"), or segmental duration (shorter vs. longer). Results showed that F0 rise shape and segmental duration influenced SKK listeners' categorization, while no effect of peak alignment was observed. Specifically, they responded to more scooped shapes as an LH, while more domed shapes were mainly assigned to H responses. Moreover, shorter duration induced a H categorization, while longer duration was related to an LH. Results suggest that SKK listeners use both F0 shape and segmental duration as important cues for tonal contrast, though F0 shape shows stronger categorical effect than duration. Thus, F0 shape information is important to determine phonological representation of lexical pitch accents, as opposed to strict tonal alignment defined in Autosegmental-Metrical theory.

This study investigates the sound system of heritage speakers (henceforth, HSs) as they shift dominance from their heritage language to their majority language. Specifically, it analyzes the production of intrusive vowels in Spanish consonant clusters across the lifespan of HSs, focusing on tautosyllabic clusters (i.e., /Cɾ/) and heterosyllabic clusters (i.e., /ɾ.C/). Semi-spontaneous speech was elicited from three age groups of Spanish HSs-younger children, older children, and adults-for whom American English is the majority language, as well as from three age-matched groups of non-heritage Spanish speakers raised in Mexico. The presence and duration of intrusive vowels were examined, with the latter calculated as a ratio of intrusive vowel duration to tap duration. Overall, the results indicate that all speaker groups more frequently produce consonant clusters with intrusive vowels than without. However, Spanish HSs produced fewer and shorter intrusive vowels compared with their non-heritage counterparts. In addition, heritage and non-heritage speakers exhibited a lower rate of intrusive vowel production in the 5-to-8 -year-old group compared with adults. Despite this, the absence of an interaction between age group and speaker type suggests a parallel pattern of change in intrusive vowel production across all ages. Voicing emerged as the most consistent phonetic factor, predicting a higher rate of intrusive vowels with a longer duration relative to the tap.

Sound correspondences (SCs) have been found to be learnable phonological patterns in second dialect acquisition. Cross-linguistically, SCs consist of similar as well as distinct variants. However, in the study of SC learning, the effect of the similarity between the corresponding variants remains understudied. The salience hypothesis proposes that distinct dialect variants are more salient and learnable, while the learning bias hypothesis in phonological learning predicts that SC patterns with similar variants are preferred by learners. We conducted an artificial language learning experiment to test how sound similarity affects SC learning. Specifically, the degrees of similarity between variants were evaluated from multidimensional metrics, including phonetic and phonological measures, which are cross-validated with typological evidence. While there was no effect of variant similarity in learning simple one-to-one SCs, a preference for the most distinct dialect variant was found in the learning of SCs exhibiting more complex mapping structures (i.e., two-to-one and one-to-two). Our results confirm a preference for distinct variants in SC learning, although this effect relies on two conditions. First, the preference for distinction emerges only in the presence of complex mapping structures. Second, this preference requires an activation threshold, in that the distance of the SC must be sufficiently large to trigger the effect.

There is a long-standing debate on the relevant articulatory dimensions for describing vowel production. In the absence of a theoretical or methodological consensus, different articulatory studies of vowels rely on different measures, which leads to lack of comparability between different sets of results. This paper addresses the problem of how to parametrise the tongue measurements relevant to vowels, obtained from midsagittal articulatory imaging. We focus on the lax vowels subsystem in Northern Anglo-English. A range of measures quantifying tongue position, height, and shape are extracted from an ultrasound dataset representing 40 speakers. These measures are compared, based on how well they capture the lingual contrast between different vowels, how stable they are across different speakers, and how intercorrelated they are. The results suggest that different measures are preferred for different vowels, which supports a multi-dimensional approach in quantifying vowel articulation.

The present study examines the perceptual salience of tonal speech errors compared with segmental errors (consonant and vowel). Tonal errors are observed less often than segmental errors. We thus hypothesize that tone errors are more easily ignored during transcription tasks because tones may have lower perceptual salience relative to segments. We test this hypothesis in Mandarin, via a number reconstruction task. Sixty-nine Mandarin native listeners heard sequences of numbers in which one number was altered by substituting its vowel, consonant, or tone. They were asked to identify which number that was. We found that Mandarin listeners identified the original number most accurately when consonants were substituted. They were the least accurate when vowels were substituted. For tone substitution, the accuracy was lower than for consonant substitution, but not significantly different from vowel substitution. Reaction times to identify a number with tone substitution were comparable to those for other types of substitutions. The results show that, contrary to our hypothesis, tone errors are not perceptually less salient than segmental errors. Specifically, tone errors are as salient as vowel errors and more salient than consonant errors, suggesting a similar phonological status shared by tone, vowel, and consonant in constraining word selection.