Chao-yu Su

Comparison of English narrow focus production by L1 English, Beijing and Taiwan Mandarin speakers

L1 English and two varieties of L1 Mandarin English speech data were extracted from the Taiwan AESOP corpus (Asian English Speech cOrpus Project) for the purpose of investigating differences in the realization of English narrow focus by L1 speakers of North American English, Taiwan Mandarin and Beijing Mandarin. Results show the combined effect of two patterns of L2 focus production: general underdifferentiation of on-focus and post-focus contrasts, which was exhibited by both L2 speaker groups, and transfer of L1-specific prosodic features, which can be argued to represent the source of difference between the two L2 groups. Overall, on-focus/post-focus contrasts in mean F0, amplitude and pitch range were realized most robustly by L1 English speakers. L1 Taiwan Mandarin speakers produced a smaller increase in mean F0 and amplitude for on-focus constituents and much smaller decrease in mean F0 and amplitude on post-focus constituents than L1 English speakers did, whereas Beijing Mandarin speakers produced no increase in mean F0 in on-focus constituents, and the smallest decrease in mean F0 on post-focus constituents, but a 35% higher post-focus compression of intensity than Taiwan Mandarin speakers did. Notably, both L2 speaker groups failed to produce post-focus compression of pitch range, which has been shown to be a highly salient cue to the presence of focus in English.

L1/L2 difference in phonological sensitivity and information planning — Evidence from F0 patterns

Assuming that linguistic specifications and information planning contribute to different levels of prosodic organization that cumulatively constitute output prosody, quantitative analysis of respective contributions can be derived through normalization procedures that remove levels of interactions involved. The current study attempts to account for how Τ2 prosody departs from the Τ1 norm in the two levels mentioned and whether an account can be offered. F0 patterns of word English stress categories (primary, secondary and tertiary) and emphases in controlled conditions (narrow-, broad- and non-focus) are compared using speech data from English Τ1 and Mandarin Τ2 speakers. Τ1 speech exhibits similar F0 patterns of binary high-low contrasts in both stress/non-stress as well as focus/non-focus categories, suggesting comparable planning are used to express phonological and information planning. However, Lis primary stress and emphasis exhibited less degree of F0 high-low contrast, coupled with reversed F0 patterns in both the secondary and tertiary categories as well as non-emphases conditions. The results demonstrate that being less sensitive to phonological categories may also affect information planning in similar ways. We believe the results explain how stress and focus interact to cause L2 accent and unintelligibility, help understand stress and focus composition of L1-and-L2 speech, and are readily applicable to CALL.

Global F0 features of Mandarin L2 English — Reflection of higher level planning difficulties from discourse association and information structure

It has been reported that one major feature of global prosody in continuous speech is to express cross-phrase association and cohesion through adjustment of individual phrase intonations. Another major feature of global prosody that also requires phrase intonation to adjust is to express information structure. Both features involve multi-level large scale speech planning; their interactions multifaceted. These higher level prosodic expressions are systematic and predictable in LI speech, but these expressions are realized in L2 speech that can be attributed to foreign accent remains unknown. The current acoustic study thus attempts to address these two issues through corpus analysis with focus on normalization procedures that remove respective interactions in order to derive pattern that better reflect each feature involved. The global FO constitution of LI vs. Mandarin L2 English is compared to see the difference. As expected, respective positive correlations and patterns can be derived from LI speech whereas divergent patterns found in L2 speech provide explanations of L2 accent. Overall cumulative effects of interaction further account for how both features collectively contribute to accent specific to Mandarin L2 English. These findings are readily applicable to CALL that targets global prosody training.

Melody of Mandarin L2 English—when L1 transfer and L2 planning come together

It is always more difficult for L2 speakers to produce the melody and tempo of continuous speech because it requires simultaneous planning of L2 linguistic specifications, higher level discourse associations and information placements. We assume that higher level planning requires within-phrase chunking and cross-phrase paragraph phrasing while information arrangements through emphasis weighting assignment and allocation. The above involved planning is most notably delivered through distinct global melodic modulations and patterns. The present compares the onset features of extracted Phrase Commands and their consistency with tagged discourse units and perceived emphases using speech data of L1 English, Taiwan (TW) L2 English and TW L1 Mandarin. Explicitly, we study F0 contour features to compare L1/L2 chunking units and their global patterns to pinpoint L2 features. Results of distinct TW L2 features compared with L1 English are (1) less consistent discourse chunking, (2) fewer distinct contours by prosodic words, (3) less degree of emphasis contrast in prosodic phrase and (4) more distinct contours in non-emphases. While (1) and (2) may reflect general L2 planning difficulties, our results show that (3) and (4) namely, flatter overall contour, are Mandarin inherent transferred to L2. We believe our proposed methods of extracted Phrase Command more accurately and better represent global melodic features that could be applied to other L1/L1 comparison in general; the findings could also be directly applied to CALL development of L2 prosody enhancement to improve overall intelligibility.

Information allocation and prosodic expressiveness in continuous speech: A Mandarin cross-genre analysis

In addition to discourse association and assuming that allocation of key information is an important feature of prosodic expressiveness of continuous speech, the common accentuation patterns across 3 Mandarin speech genres through 4 degrees of perceived emphases are derived. Using frequency count as another control, it is found that only 6 types of emphasis patterns are needed account for 70% of the speech data regardless of genre. The 6 emphasis types are further compared for the distribution of (1) discourse units and emphasis tokens by speech genre, (2) emphasis pattern by phrase and (3) with respect to discourse positions to see if genre-specific features could be found. Results reveal that genre-dependent features can also be accounted for. In addition, individual genre properties are found to also be correlated with phrase length and specific emphasis patterns.

A hierarchical linguistic information-based model of English prosody: L2 data analysis and implications for computer-assisted language learning

Abstract The paper presents a prosody model of native English (L1) continuous speech as corrective prosodic feedback for non-native learners. The model incorporates both hierarchical discourse association and information structure to (1) pinpoint the prosodic features of multi-phrase continuous speech, and (2) simulate native-like expressive speech using corpus of North American and Taiwan L2 English. The bottom-up, additive, data-driven model aims to generate L1-like expressive continuous speech with built-in phonetic and phonological specifications at the lexical level, syntactic/semantic specifications at the next higher phrase and sentence levels, and completed with patterned paragraph associations and prosodic projections of information allocation at higher levels. The hierarchical model successfully allows us to identify L1-L2 differences by prosodic modules/patterns as novel additional features “discourse structure” and “information density” reliably nail down L1-L2 prosodic differences related to phrase association as well as information placement. Our L1 prosodic model with the proposed predictors and optimized model trained from L1 speech corpus showed increase of prediction over existing methods. As a corrective feedback for L2 learners, these predicted L1 prosodic features were compared with a baseline model by objective evaluation (RMS error and correlation) then superimposed onto the L2 speech tokens. Resynthesized L2 tokens were subsequently compared with the original L2 tokens for degrees of perceived accent using subjective evaluation (native-listener perception test). We believe the proposed model can be an effective alternative for implementing computer-assisted language learning (CALL) systems that helps generate L1-like prosody from text, and at the same time serves as corrective feedback for L2 learners.

Where and how to make an emphasis? - L2 distinct prosody and why

It has been reported in the literature that L2 English prosody differs from L1 at the lexical, syntactic and discourse levels; characterized by under-differentiated word stress and narrow focus as well as smaller discourse units, respectively. Using continuous speech data of L1 TW Mandarin, L1 English and TW L2 English, the present study compares informationstructure related L2 prosody through emphasis placement in phrases and in multi-phrase paragraphs. Results show that TW L2 English is marked by fewer and less varied emphasis patterns while emphasis placement is only related to one specific genre of L1 Mandarin. Acoustic analyses also reveal that the contrast strength of L2 produced emphases is again less robust and therefore under-differentiated than L1. These findings suggest that sources of L2 prosody are multi-fold, contributions from constraints of speech planning at lexical, sentential and discourse and information levels may all take part. We believe our findings can be applied to L2 English teaching as well as CALL technology development.