Exploring the anatomy of articulation rate in spontaneous English speech: relationships between utterance length effects and social factors

TL;DR

This paper addresses how utterance length and social factors shape articulation rate in spontaneous English. It analyzes 116,020 utterances from 13 corpora using a distributional Bayesian multilevel model of the mean $\mu$ and variance $\sigma^2$ of $\log(\text{articulation rate})$, with utterance length captured via two measures (speaker mean and relative deviation) and modeled with cubic splines; age and gender are included as predictors, with random effects for corpus and speaker. The key finding is that utterance length exerts the largest and most consistent influence on both the mean and variance of articulation rate (approximately $2.1$–$2.4$ syll/sec increase from shortest to longest utterances), while age and gender have smaller effects (age about $-0.7$ to $-0.8$ syll/sec; gender about $0.21$–$0.24$ syll/sec). The results imply that production and perceptual constraints largely structure speech timing, with social factors contributing but to a lesser extent, and highlight the value of large, cross-corpus analyses in sociophonetic research.

Abstract

Speech rate has been shown to vary across social categories such as gender, age, and dialect, while also being conditioned by properties of speech planning. The effect of utterance length, where speech rate is faster and less variable for longer utterances, has also been shown to reduce the role of social factors once it has been accounted for, leaving unclear the relationship between social factors and speech production in conditioning speech rate. Through modelling of speech rate across 13 English speech corpora, it is found that utterance length has the largest effect on speech rate, though this effect itself varies little across corpora and speakers. While age and gender also modulate speech rate, their effects are much smaller in magnitude. These findings suggest utterance length effects may be conditioned by articulatory and perceptual constraints, and that social influences on speech rate should be interpreted in the broader context of how speech rate variation is structured.

Figures (4)

• Figure 1: Estimated articulation rate (left) and variance in articulation rate (right) for each corpus as a function of utterance length. Lines indicate posterior medians with shaded areas representing 95% Bayesian credible intervals.
• Figure 2: Estimated articulation rate for each speaker (faceted by corpus) as a function of utterance length (one line per speaker). Values indicate posterior medians.
• Figure 3: Top: estimated articulation rate as a function of speaker age across corpora, held at mean utterance length. Values indicate posterior medians (one line per dialect) with 95% Bayesian credible intervals. Bottom: histogram of speaker ages.
• Figure 4: Estimated male-female difference in articulation rate for each corpus, held at mean utterance length. Points and lines indicate posterior medians with 95% Bayesian credible intervals, with dashed vertical line indicating null gender difference.