How does first language influence second language rhythm

Overview • Background: Ø Speech rhythm and “rhythm classes”. Ø New speech rhythm metrics.

Background: Rhythm • Rhythm (music, speech, etc. ) arises from the repetition of related

Stress-timing vs syllable-timing • Stress-timed languages: isochrony? sleep. ing. soundly vs sleep. il. y.

Basis of rhythmic analysis • Separate signal into vocalic and intervocalic sections the standards

New rhythm metrics • Variance-based metrics (Ramus, Nespor & Mehler, 1999): Ø ΔV: standard

Second language recordings • Comparison between “rhythm classes”: Ø Spanish (“syllable-timed”) vs English (“stress-timed”).

Spanish & English L 1/L 2 results: Variance-based metrics Key Lang. spoken Native lang.

All speakers: Effect of speech rate on ΔV • Varco. V – ΔV normalised

Spanish & English L 1/L 2 results: Rate normalised ΔV and %V Key Eng

Spanish L 1/L 2: %V differences continued… • A mí no me gustaba su

Second language recordings • What is the effect of speaking a second language, when

Cross-linguistic results: Dutch vs English Key Lang. spoken Native lang. Dut. Eng Dut Eng.

First and second language rhythm: summary • Influence of L 1 on L 2:

Interpretation of rhythm metrics • Rhythmic distinctions can be seen as arising from specific

Credits Biotechnology and Biological Sciences Research Council, U. K. Juan Toro, Barcelona. Elizabeth Johnson,

Slides: 17

Download presentation

How does first language influence second language rhythm? Laurence White and Sven Mattys Experimental Psychology Bristol University

Overview • Background: Ø Speech rhythm and “rhythm classes”. Ø New speech rhythm metrics. • Speech production study: Ø Do new rhythm metrics serve to illustrate the influence of first language on second language rhythm?

Background: Rhythm • Rhythm (music, speech, etc. ) arises from the repetition of related sounds. • In English speech, rhythm arises from repetition of stressed syllables. I did not have sexual relations with that woman – Miss Lewinsky. • In Spanish, there is much less contrast between stressed and unstressed syllables. A pocos pasos de mi casa está una tienda bonita. • Traditional distinction: Ø Syllable-timed languages (e. g. Romance). Ø Stress-timed languages (e. g. Germanic).

Stress-timing vs syllable-timing • Stress-timed languages: isochrony? sleep. ing. soundly vs sleep. il. y. re. searches • Stress-timing is alternation rather than regularity: Ø Greater differences between stressed and unstressed syllables… including: ØVowel reduction. ØMore complex consonant clusters, particularly in stressed syllable onsets and codas. e. g. (s)(t)(r)a(n)(d)(s) ØFewer open syllables.

Basis of rhythmic analysis • Separate signal into vocalic and intervocalic sections the standards committee and examine variations in length of each.

New rhythm metrics • Variance-based metrics (Ramus, Nespor & Mehler, 1999): Ø ΔV: standard deviation of vocalic intervals. Ø ΔC: standard deviation of intervocalic intervals. Ø %V: vocalic proportion of total utterance. • Also examined PVI metrics (Low, Grabe & Nolan, 2000; Grabe & Low, 2002). Ramus, Nespor & Mehler (1999)

Second language recordings • Comparison between “rhythm classes”: Ø Spanish (“syllable-timed”) vs English (“stress-timed”). Native language Language spoken Spanish Sp. Sp Sp. English Eng. Sp English • Six speakers per condition. • Five sentences per language. • Other materials also recorded for most speakers: Ø Map task to elicit non-read speech.

Spanish & English L 1/L 2 results: Variance-based metrics Key Lang. spoken Native lang. Speech rate (syls/sec) Eng. Sp 4. 8 Sp. Sp 8. 0 Eng 5. 2 Sp. Eng 6. 6

All speakers: Effect of speech rate on ΔV • Varco. V – ΔV normalised for speech rate: ΔV / Mean V (Dellwo & Wagner, 2003)

Spanish & English L 1/L 2 results: Rate normalised ΔV and %V Key Eng Lang. spoken Native lang. Eng. Sp Sp. Eng Sp. Sp

Spanish L 1/L 2: %V differences continued… • A mí no me gustaba su coche pequeño y viejo. Sp Sp. Eng Vowel duration (ms) 40, 41 vs 27, 24 Vowel duration (ms) 65, 67 vs 67, 66 47, 32 vs 71, 56

Second language recordings • What is the effect of speaking a second language, when languages are rhythmically similar? • Comparison within “rhythm classes”: Ø Dutch (“stress-timed”) vs English (“stress-timed”). Native language Language spoken Dutch English Dutch Dut. English Eng. Dut Eng • Six speakers per condition. • Five sentences per language.

Cross-linguistic results: Dutch vs English Key Lang. spoken Native lang. Dut. Eng Dut Eng. Sp Sp. Eng Sp. Sp

First and second language rhythm: summary • Influence of L 1 on L 2: Ø Consonantal metrics show little influence of L 1. Ø Vocalic metrics: ØNormalisation for speech rate necessary (Varco. V). ØClearly show influence of L 1 on L 2. ØBut… ØNot necessarily intermediate between L 1 and L 2. ØLittle accommodation to L 2 when languages are rhythmically similar.

Interpretation of rhythm metrics • Rhythmic distinctions can be seen as arising from specific segmental and prosodic processes: Ø Rhythm as emergent property rather than product of top-down timing control. • Variation within “rhythm classes” sometimes as large as between. Ø Suggests distinction is not simply bimodal. • Further research: How do rhythm metrics relate to the subjective experience of linguistic rhythm?

Credits Biotechnology and Biological Sciences Research Council, U. K. Juan Toro, Barcelona. Elizabeth Johnson, Nijmegen. Eric-Jan Wagenmakers & Atie Vogelenzang de Jong, Amsterdam. Ineke Mennen, Edinburgh. Reinier Salverda, London.