ISCA Archive SWAP 2000
ISCA Archive SWAP 2000

Prosody and Word Recognition: A case study

Liang Tao

Three experiments tested the role of an emerging syntactic tone in spoken word recognition of Beijing Mandarin Chinese. The study tested whether there are two prelexical representations of a word following yi55 (one) with two different tones, and how synchronic tonal variations is mapped onto word recognition during online language processing.

Beijing Mandarin is currently undergoing syntactic changes in its noun phrase formation. The obligatory rule of the classifier is changing, caused by sound erosion on high-frequency units in oral discourse. Such changes leave a 'frozen' tone that does not follow the Mandarin tone sandhi rules, and a noun phrase without the classifier. Although Mandarin tones are lexical, the 'frozen' tone assumes a syntactic function (Tao, 1999). Specifically, the tone is a high-rising tone associated with the word yi55 (one). When yi55 is used with a noun that may be a classifier or a noun (e.g., che55: car), the frozen tone indicates the word is a noun (yi35 che55: a car). A lexical tone with the sandhi change indicates the word is a classifier (yi51 che55: a carload). The change may be the beginning of an article in the language.

Four passages were designed each ending with a phrase that may be either a noun phrase or an elliptical nominal expression, depending on the crucial tonal. Passages were recorded twice with two different tones on the word yi55 (one). Following each passage were four multiple-choice questions, focusing on the definition of the word following the crucial tone. Experiments 1 and 2 tested participants individually with the answers videotaped. Participants were asked to retell the passages using their own words in Experiment 2. Experiment 3 tested 100 participants in two groups. The answers were printed on paper.

Preliminary analysis indicates that the tonal variations on the same word did have an impact: The lexical tone was readily processed and its following word correctly recognized, yet the words following the syntactic tone were correctly identified only by those familiar with the multiple-choice test format. The others either ignored the syntactic tone during word recognition, or chose a heuristic answer to compensate. The results suggest a conflict between perception and mental representation of the spoken word, indicating that with reference to an emerging language change, the acoustic image of the word does not readily match the actual sound; thus the mental lexicon lags behind language change.


Cite as: Tao, L. (2000) Prosody and Word Recognition: A case study. Proc. Spoken Word Access Processes (SWAP), 175-178

@inproceedings{tao00_swap,
  author={Liang Tao},
  title={{Prosody and Word Recognition: A case study}},
  year=2000,
  booktitle={Proc. Spoken Word Access Processes (SWAP)},
  pages={175--178}
}