TL;DR
This paper investigates the acoustic underpinnings of a historical sound change: the palatalization of /l/ in consonant clusters. By analyzing speech recordings, the authors found that velar plosives are more likely to trigger this change than labial ones, and that "darker" sounding /l/ sounds are more resistant to palatalization. This work sheds light on the phonetic mechanisms driving language evolution.
What Problem Does This Paper Address?
Historical linguists have long observed a phenomenon called /l/-palatalization, where the sound /l/ in consonant clusters (like "pl" or "kl") changes to a more "palatalized" sound (similar to the "y" in "yes") even when there's no following vowel that would typically cause such a change (like "i" or "e"). This change has occurred in various languages, including Italian and Thai. While dialectal data suggests that this change is more common after velar plosives (/k/, /g/) than labial plosives (/p/, /b/), and that "dark" /l/ sounds are less likely to palatalize than "clear" /l/ sounds, the precise acoustic reasons why this happens have remained largely unexplored. Previous acoustic studies relied on visual inspection of spectrograms and didn't systematically investigate the influence of the preceding consonant's place of articulation or the acoustic properties of the /l/ itself. This paper aims to fill that gap by providing a detailed acoustic analysis.
Methods
The researchers designed an experiment to elicit specific consonant clusters. They recruited four speakers: two from Occitan (classified as a "clear-/l/" language) and two from Catalan (classified as a "dark-/l/" language). These speakers read disyllabic nonsense words containing target consonant clusters followed by either /a/ or /u/. Importantly, they excluded tokens where the /l/ was followed by /i/ or /j/ (the palatalized counterpart) because these sequences are phonotactically constrained and might already exhibit palatalization, confounding the results.
The core of their analysis involved acoustic measurements of the speech signal. They extracted acoustic features, specifically the first and second formant frequencies (F1 and F2), at specific time points after the onset of the lateral (/l/) or its palatalized counterpart (/j/). The difference between F2 and F1 (F2-F1) was used as a key metric, as it's known to correlate with the degree of palatalization (higher F2-F1) and velarity/darkness (lower F2-F1) of a sound. These F2-F1 values were then analyzed using statistical methods (ANOVAs) to determine the significance of various factors, such as the type of consonant cluster (/Cl/ vs. /Cj/), the place of articulation of the preceding consonant, and the following vowel.
Datasets
The study utilized a newly created dataset of spoken nonsense words. The specific tokens were derived from a list of disyllabic words designed to contain the target consonant clusters: plosive + lateral (/pl/, /bl/, /kl/, /gl/) and their palatalized counterparts (plosive + yod, e.g., /pj/, /bj/, /kj/, /gj/). The data was collected from four speakers of Occitan and Catalan. While the paper mentions that tokens with /i/ or /j/ following the cluster were recorded, they were not included in the primary analysis.
Key Insights and Ideas
- Velars Lead the Way: The study provides acoustic evidence supporting the historical observation that velar plosives (/k/, /g/) are more prone to triggering /l/-palatalization than labial plosives (/p/, /b/). This suggests that the articulatory transition from a velar sound to an /l/ might inherently create acoustic cues that listeners interpret as palatalization.
- Dark /l/ Resists Change: The research confirms the hypothesis that "dark" /l/ sounds are more resistant to palatalization. This is likely due to the more constrained tongue body articulation required for dark /l/, making it less susceptible to coarticulatory influences that drive palatalization.
- Acoustic Cues Drive Sound Change: The paper emphasizes that sound change isn't just about articulation but also about how sounds are perceived. The acoustic properties of velar + lateral clusters may sound "more palatalized" to listeners, even in the absence of a direct palatalizing trigger, thus driving the historical change.
- F2-F1 as a Diagnostic Tool: The study effectively uses the F2-F1 difference as a quantifiable measure to distinguish between palatalized and non-palatalized sounds and to assess the degree of "darkness" or "clearness" in lateral sounds.
Key Results
- Significant Differences: Statistical analysis revealed significant differences in F2-F1 values between lateral (/Cl/) and yod (/Cj/) clusters for all speakers, indicating a clear acoustic distinction.
- Place of Articulation Matters: The study found that velar + lateral clusters showed greater acoustic similarity to velar + yod clusters than labial + lateral clusters did to labial + yod clusters. This supports the idea that velars are more likely to undergo palatalization.
- Occitan vs. Catalan: Occitan speakers' laterals were found to be significantly clearer (higher F2-F1) than Catalan speakers', aligning with their classification as clear- and dark-/l/-languages, respectively.
- Vowel Influence Limited: The following vowel (/a/ or /u/) generally had a non-significant effect on the F2-F1 distance, suggesting that the primary drivers of palatalization in these clusters are the preceding consonant and the lateral itself.
- Specific Comparisons: Post-hoc tests showed that labial + lateral clusters were significantly different from their labial + yod counterparts in both Occitan and Catalan, suggesting that palatalization in these clusters is less widespread and potentially driven by different mechanisms than in velar clusters.
Limitations and Future Work
- Limited Speaker Pool: The study used only four speakers, which limits the generalizability of the findings. Future work could involve a larger and more diverse group of speakers.
- Nonsense Words: While useful for controlling phonetic environments, the use of nonsense words might not perfectly reflect natural speech patterns.
- Focus on Production: The study primarily focused on the acoustic production of these sounds. Future research could incorporate perceptual studies to directly test how listeners interpret these acoustic cues.
- Labial + Lateral Palatalization: The paper acknowledges that the reasons for palatalization in labial + lateral clusters remain less clear and are not as widespread as in velar clusters. Further research is needed to explore the subtle acoustic similarities and potential analogical sound change mechanisms in these cases.
- L-Epenthesis: The authors suggest future research should investigate the relationship between /l/-palatalization in onset clusters and /l/-epenthesis (the insertion of an /l/ sound) in obstruent + yod clusters.
Summary updated