Abstract: Substantial evidence has suggested that the processes of speech plan of people who stutter (PWS) may be impaired. Recently, discussions focused mainly on the exact process in which PWS have deficit. The difference of word length effect between PWS and people who do not stutter (PWNS) provided important support for the phonological encoding deficit hypothesis of stuttering. By reviewing the literatures on word length effect, three alternative explanations for this difference were put forward. First, frequencies of short words are usually lower than those of long words. The word length effect difference between PWS and PWNS may be due to frequency effect. Second, stuttering may easily occur when phases or sentences are longer and more complex, but none of the two factors can singly account for stuttering. So, syntax may be another explanation for their difference. Third, in block design, pronunciation length would influence reaction strategies, and may also contribute to the difference. Thus, 3 picture naming experiments were used to test these hypotheses respectively.
Method and Results
In experiment 1, a 2 x 2 x 2 factorial design was set, with word length and word frequency as within subject factors. Ten PWS and ten PWNS participated in this experiment. The result indicated that stable differences of word length effect were obtained between PWS and PWNS, which were not influenced by word frequency. So, deficit of PWS may not exist in phoneme retrieval process.
In experiment 2, A 2 x 2 factorial design was set, with word class (noun phase vs. verb phase) as within subject factors. In both conditions, names of pictures consisted of 3 Chinese characters. Ten PWS and ten PWNS participated in this experiment. The differences between two word class conditions were not significant in PWNS, but prominent in PWS. This result suggested PWS were sensitive to syntactical complexity after controlling the word length.
In experiment 3, three conditions (1 character picture naming, 3 character picture naming and 1 character picture naming 3 times) were designed. The first two conditions were used to obtain the word length effect, and the third condition was used to control the length of pronunciation. Contrary to our prediction, the increase of pronouncing length did not prolong naming latency of PWS, and word length effect was still prominent after controlling pronouncing length.
Conclusion
The results showed that both the syntactical and phonological encoding processes may be impaired in PWS. And a slots-and-filler was applied to illustrate how these two processes contributed to the difference of word length effect between PWS and PWNS. Meanwhile, it was suggested that stuttering therapists should pay more attention on speech training with long and complex sentences

Key words: stuttering, word length effect, phonological encoding, picture naming