Effects of dialectal origin on articulation rate in French

Mathieu Avanzi, Pauline Dubosson, Sandra SchwabInterspeech 2012

TL;DR

This paper investigates how dialectal origin and other sociolinguistic factors influence articulation rate in French. The study finds that Parisian French speakers articulate faster than Swiss French speakers, regardless of whether the Swiss speakers are native or non-native French speakers. Interestingly, advanced non-native French speakers do not necessarily articulate slower than native speakers of the same variety.

What Problem Does This Paper Address?

The pace at which we speak, specifically our articulation rate (how quickly we produce speech sounds, excluding pauses), can vary significantly. While previous research has explored factors like speech style (reading vs. conversation) and speaker demographics (age, gender) on articulation rate, there's been a lack of consensus on how dialectal origin affects it, especially in French. Furthermore, the question of whether non-native speakers of a language articulate differently from native speakers of the same variety, particularly when they have achieved a high level of proficiency, remains underexplored. This paper aims to bridge this gap by systematically comparing articulation rates across different French varieties and speaker backgrounds, while controlling for other influencing factors.

Methods

The researchers employed a controlled experimental design to measure articulation rate. They recorded participants speaking French in two distinct styles: reading a journalistic text and engaging in a free conversation. The audio recordings were then meticulously transcribed and aligned with the speech signal using automated tools (Praat software and the EasyAlign script), with manual verification and correction to ensure accuracy. Crucially, the analysis focused on "accentual phrases" (APs), which are units of speech carrying a pitch accent. Within these APs, disfluencies were identified and excluded. The core metric, articulation rate, was calculated as the mean syllabic duration within each AP. Statistical analysis was performed using generalized linear models with repeated measures, incorporating speaker variety, gender, speech style, and the number of syllables within the AP as predictor variables.

Datasets

The study utilized recordings from four groups of French speakers:

  • Parisian French (PA): Native speakers from Paris.
  • Neuchâtel Swiss French (NE): Native speakers from Neuchâtel, Switzerland.
  • Bernese German speakers (BE): Non-native French speakers from the Emmental-Oberaargau area in the canton of Bern, who have been living in a French-speaking environment (Neuchâtel) for at least 20 years.
  • Zurich Swiss German speakers (ZH): Non-native French speakers from the city of Zurich, who have also been living in a French-speaking environment (Neuchâtel) for at least 20 years.

The participants were carefully selected to be elderly (over 55 years old) to mitigate potential age-related biases in articulation rate. The reading task involved a journalistic text from the PFC project, and the conversation task lasted for 20 minutes.

Key Insights and Ideas

  • Dialectal Differences are Real: The study provides robust evidence that dialectal origin significantly impacts articulation rate in French.
  • Parisian Speed vs. Swiss Slowness: Parisian speakers articulate faster than their Swiss counterparts, a finding that holds true for both native and non-native Swiss speakers.
  • Advanced L2 Speakers Blend In: Surprisingly, highly proficient non-native French speakers (Swiss German speakers with extensive exposure) do not necessarily articulate slower than native speakers of the same regional variety (Neuchâtel Swiss French). This suggests that with sufficient immersion, L2 articulation rate can converge with L1 norms.
  • Gender Matters, But It's Complicated: While males generally articulate faster than females, this difference is not uniform across all varieties. The effect of gender is less pronounced in some Swiss varieties compared to Parisian French.
  • Conversation is Faster: As expected, speakers articulate faster in conversational speech than in reading, likely due to the more spontaneous and less constrained nature of conversation.
  • Syllable Count Influences Pace: The number of syllables within an accentual phrase also affects articulation rate, with more syllables potentially leading to a slightly slower pace, though this effect is more pronounced in reading.

Key Results

  • Variety Effect: Parisian speakers exhibit a shorter syllabic duration (higher articulation rate) than Swiss speakers (NE, BE, ZH) (p < .001).
  • Non-Native Convergence: Non-native Swiss speakers (BE and ZH) show similar articulation rates to native NE speakers.
  • Gender Effect: Syllabic duration is shorter in males (215ms) than in females (235ms) (p < .001), but this difference varies significantly across varieties (p < .001).
  • Speech Style Effect: Speakers have shorter syllabic durations in conversation (211ms) compared to reading (239ms) (p < .001).
  • Syllable Number Effect: Syllabic duration decreases as the number of syllables within an AP increases (p < .001). This effect is stronger in reading than in conversation.

Limitations and Future Work

The authors acknowledge that while their study controls for several factors, further research is needed. They suggest that future work should:

  • Include a larger number of speakers to increase statistical power.
  • Investigate other temporal features, such as pause distribution and duration, and rhythmic variations (consonant and vocalic variability), to provide a more comprehensive understanding of speech pace.
  • Explore the reasons behind the observed differences, potentially linking them to phonological features of Swiss French (e.g., the use of duration as a distinctive feature) or the prevalence of penultimate syllables in Swiss varieties.
  • Further analyze the impact of L1 background on articulation rate in French, especially for non-native speakers with varying degrees of exposure and proficiency.
  • Examine the influence of age more closely, as the current study focused on older speakers to control for its effect.

Summary updated

BibTeX
@inproceedings{avanzi2012effects,
  author = {Mathieu Avanzi and Pauline Dubosson and Sandra Schwab},
  title = {{Effects of dialectal origin on articulation rate in French}},
  booktitle = {Interspeech},
  year = {2012},
}
Figure