Dec 8, 2010

2010/12/08

Lee, C.-Y. (2007). Does horse activate mother? Processing lexical tone in formriming. Language and Speech, 50, 101–123.

Presentation: Chris
Summary: Shelly

Lexical tones in tone languages are phonemically contrastive, but when and how they are used during lexical processing are issues in need of more investigation. The present study used four perceptual experiments with the form-priming paradigm in attempt to address these issues. In Experiment 1, four types of primes were identified: ST primes share both the segmental structure and the tone with the targets (e.g., lou2 ‘hall’–lou2 ‘hall’); S primes share only the segmental structure (e.g. lou3 ‘hug’–lou2 ‘hall’); T primes share only tone (e.g. cang2 ‘hide’–lou2 ‘hall’); UR primes are unrelated to targets neither in tones, nor in segments (e.g. pan1 ‘climb’–lou2 ‘hall’). Subjects were asked to perform a lexical decision task, and the ISI was set at 250 ms, which was a direct comparison with Yip (2001). Results showed that only ST primes displayed facilitative priming, and S primes or T primes did not facilitate target responses reliably. The lack of priming from segmental overlap or tonal overlap alone may be due to the long ISI, at which potential activation in the prime might have disappeared before it could have an impact on the target. Therefore, in Experiment 2, a shorter ISI, 50 ms, was used. However, similar results were obtained as in Experiment 1, in which segmental or tonal overlap alone showed no priming effect. In order to see whether this failure of priming was caused by activation inhibition on incompatible elements between primes and targets, or because of some unwanted response strategies that subjects utilized in the direct priming test, a mediated priming paradigm was used in Experiment 3. In a mediated paradigm, the prime and the target are not directly related, but the prime is form-related with a third word, which is semantically related to the target. Therefore, if there is facilitation of the target fu4qin0 ‘father’ by the prime ma3 ‘a horse’, the effect should be from the mediated third word ma1 ‘mother’. In Experiment 3, the 250 ms ISI was used, and still, results showed no evidence that the S or T primes produced facilitation priming. In Experiment 4, a shorter ISI (50 ms) was used, and eventually, reliable facilitation by S primes was shown, which implied that S primes did activate ST primes, which further facilitated the response of the targets. According to the results of the above experiments, the authors concluded that listeners indeed use tonal information in the early phase of lexical access, and such activation is short-lived, which is used on-line for disambiguation and disappears fast.