TL;DR
This study investigates how aging affects the rhythmic patterns of Italian speech, specifically focusing on the proportion of vowels within utterances (%V) and the timing between vowels (VtoV). They found that older speakers tend to have a higher vowel percentage, suggesting a shift towards a mora-timed rhythm, and this effect is independent of speech rate.
What Problem Does This Paper Address?
While many studies have explored how aging impacts acoustic speech features like pitch and speech rate, the influence of aging on speech rhythm has been less examined. Speech rhythm, which relates to the temporal organization of speech, is often characterized by the relative durations of vowels and consonants. This paper aims to fill this gap by investigating how the proportion of vowels (%V) and the timing between vowels (VtoV) change with age in Italian speakers. Understanding these rhythmic variations is important because they can influence how listeners perceive and process speech, and how they judge the speaker's age.
Methods
The researchers conducted a controlled experiment involving eight Italian native speakers, divided equally into two age groups: young adults (20-25 years old) and older adults (75-80 years old). Participants were asked to read four sentences of varying lengths at four different speech rates (slow, medium, fast, very fast), controlled by synchronizing their speech with a visual cue.
The recorded speech was then analyzed using spectro-acoustic methods, specifically the Praat software. The key measurements extracted were:
- Vowel Percentage (%V): The proportion of time spent producing vowels within an utterance.
- Vowel-to-Vowel Interval (VtoV): The time elapsed between the onset of consecutive vowels.
These measurements were compared between the age groups and across different speech rates.
Datasets
The study utilized a newly collected corpus of read speech from 8 Italian speakers (4 young, 4 old) reading 4 sentences at 4 different speech rates, resulting in 32 utterances. This corpus was analyzed and compared to previously published data from the "Piero Angela" (PA) corpus, which included recordings of the same Italian speaker at ages 40 and 79, as well as data from multilingual corpora of TV news, drug advertisements, and recited speech.
Key Insights and Ideas
- Aging shifts Italian speech rhythm towards mora-timing: The most significant finding is that older speakers exhibit a higher vowel percentage (%V) compared to younger speakers. This suggests a tendency for Italian speech to move from a more syllable- or stress-timed rhythm towards a mora-timed rhythm with advancing age.
- Rhythmic changes are age-dependent, not speech-rate dependent: Crucially, the increase in %V observed in older speakers was consistent across all tested speech rates. This indicates that the underlying rhythmic shift is primarily driven by age, rather than being a compensatory mechanism for speaking faster or slower.
- VtoV intervals are more variable but reflect speech rate: While %V showed a clear age-related trend, VtoV intervals displayed more variation, reflecting the different speech rates. Longer VtoV intervals correlated with slower speech, and shorter intervals with faster speech.
- Previous findings on PA corpus are validated: The study's results align with and expand upon earlier findings from the PA corpus, confirming the age-related increase in %V and suggesting that this phenomenon is not unique to a single speaker but is a general trend in aging Italian speakers.
Key Results
- Young adults (20-25) had %V values ranging from 44.1% to 47.4%.
- Older speakers (75-80) had significantly higher %V values, ranging from 49.4% to 53%.
- These differences in %V between age groups were statistically significant (p<0.01).
- Intra-group differences (within young adults or within old speakers) for %V were not statistically significant (p>0.05).
- Gender did not significantly affect %V values within the same age group.
- Speech rate variations did not cause statistically significant changes in %V for either young or old speakers.
Limitations and Future Work
- Limited speech styles: The study focused on read speech. The authors suggest future research should explore informal and spontaneous speech to better understand the effect of aging on speech rate variations and rhythmic patterns in more naturalistic contexts.
- Age band gaps: The study included young adults and elderly speakers but not intermediate age groups (25-75 years old). Future work should investigate these age bands to precisely define the temporal limits of %V variations and the associated changes in speech production mechanisms.
- Language comparison: The research could be extended to speakers of other languages, particularly those classified as isosyllabic, stress-timed, and mora-timed, to see if similar age-related rhythmic shifts occur across different linguistic rhythmic typologies.
Summary updated