TL;DR
This paper investigates how Korean speakers perceive foreign language plosives, specifically focusing on whether their perception patterns explain why certain sounds are adapted differently in Korean loanwords. The study reveals a disconnect between how Korean speakers perceive foreign sounds and how those sounds are phonologically integrated into Korean, suggesting that perception alone doesn't fully account for loanword adaptation.
What Problem Does This Paper Address?
When foreign words are borrowed into a language, their sounds are often adapted to fit the phonological system of the borrowing language. Korean has a unique three-way distinction in its oral stops (lenis, fortis, aspirated), while many other languages, like Japanese, French, and English, have a simpler two-way distinction (voiced/voiceless, or aspirated/unaspirated). Previous studies on Korean loanwords have observed interesting patterns: for example, Japanese initial voiceless plosives are often adapted as lenis in Korean, and French voiceless plosives are often adapted as fortis or aspirated. The authors question whether these loanword adaptation patterns are driven by how Korean speakers perceive these foreign sounds. They aim to bridge the gap between phonological adaptation theories, which often assume perception is the primary driver, and empirical perceptual data.
Methods
The researchers conducted a perception experiment where Korean speakers listened to plosive sounds from Japanese, French, English, and Chinese. They used CVCV (consonant-vowel-consonant-vowel) nonsense word tokens. To investigate the role of acoustic cues, they manipulated the stimuli, particularly focusing on pitch (F0) variations. They recorded natural speech tokens from native speakers of each language and then created resynthesized stimuli with altered F0 patterns. Korean listeners were then asked to categorize these plosives, and their responses were analyzed using statistical methods like ANOVAs. The study specifically examined the perception of initial and medial plosives and how different acoustic features (like VOT and F0) might influence categorization.
Datasets
The study used recorded speech data from native speakers of Tokyo Japanese, General American English, Parisian French, Seoul Korean, and Mandarin Chinese. These recordings formed the basis for the acoustic measurements and the stimuli used in the perception experiment. The paper does not introduce a new benchmark dataset but rather utilizes existing speech data to create experimental stimuli.
Key Insights and Ideas
- Perception vs. Adaptation Mismatch: A core finding is that Korean speakers' perception of foreign plosives doesn't always align with how those sounds are adapted in Korean loanwords. This challenges the assumption that perception is the sole or primary driver of loanword phonology.
- Japanese Voicing Neutralization: While Japanese initial voiceless plosives are often adapted as lenis in Korean loanwords, Korean speakers in the experiment perceived them more as aspirated. This suggests that the loanword adaptation pattern might not be directly explained by this specific perceptual cue.
- French Plosive Perception: French voiceless plosives were perceived as fortis or aspirated, and voiced plosives as lenis, aligning somewhat with loanword patterns but with less emphasis on aspiration than observed in some loanword analyses.
- English and Chinese Perception: English voiceless plosives were perceived as aspirated, while voiced plosives were categorized variably between fortis and lenis. Chinese unaspirated plosives were perceived as fortis, with high F0 values potentially playing a role.
- F0 Influence: Decreasing F0 generally led to a decrease in [+tense] responses and an increase in [-tense] responses, indicating F0 is a significant cue for Korean speakers, particularly for distinguishing tense/lax categories.
- VOT Dominance: In some cases, like with English plosives, VOT appeared to override F0 as a primary perceptual cue.
Key Results
- Japanese initial plosives were perceived as predominantly aspirated by Korean speakers, contrasting with their adaptation as lenis in loanwords.
- French voiceless plosives were perceived as fortis or aspirated, and voiced plosives as lenis.
- English voiceless plosives were perceived as aspirated in both initial and medial positions.
- Chinese unaspirated plosives were perceived as fortis.
- Statistical analyses revealed significant effects of language and position on plosive perception, with French and Japanese showing distinct patterns in initial positions.
Limitations and Future Work
The authors acknowledge that their study focused on acoustic cues like VOT and F0, and other factors like spectral tilt were not investigated. They also note that the listeners were university students with limited exposure to foreign languages, and their perception might differ from listeners with more language knowledge. Future work should explore how a listener's knowledge of the donor language influences perception and adaptation. The paper also raises questions about historical changes in Japanese phonology and the potential cancellation of Japanese pitch accent information in loanword adaptation. Further research is needed to understand the principles guiding the selection of acoustic cues for loanword grammar and the role of factors beyond acoustic perception.
Summary updated