Abstract: Stuttering, speech fluency disorder, is divided into developmental and acquired stuttering. With the accumulation of research evidence and improvement of research means, the researchers found that there were different speech recognition, speech muscle movements and brain activity between persons who stutter (PWS) and persons who do not (PNS). Therefore, more people started to believe there might be some intrinsic stuttering speech defects. Multi-factorial model of stuttering specify variables that are hypothetically linked to speech motor instability or breakdown, including memory load, syntactic complexity or speaker anxiety. Researchers have launched studies on semantic encoding and syntactic encoding for those PWS, in addition to their phonological encoding defect. Two different experiments were conducted in the present study to investigate the semantic encoding performance of PWS in picture-word interference paradigm which is used to examine the semantic encoding and phonological encoding of speech production.
In Experiment 1, PWS and PNS performance were examined in basic-level and category-level naming task. Each of these target pictures has two conditions including semantically related condition and unrelated condition (e.g. the target picture “dog” was accompanied by the word “cat” in semantically related condition and by the word “pear” in the unrelated condition). All pictures came from Snodgrass and Vanderwart. In basic-level naming, Participants were instructed to ignore the context word and name the target pictures (basic level naming) or to name the category of the pictures (category-level naming task) as fast as possible while maintaining accuracy. In Experiment 2, the time process of semantic interference impacting on PWS was discussed with different SOA conditions (-200 ms, -100 ms, 0 ms). All participants, including PWS and PNS, were tested individually on computers with the same basic-level naming task as that used in Experiment 1.
Results showed semantic interference effect in the basic-level naming, while a relatively stronger effect occurred for PWS than for PNS, especially in the -100 ms and 0 ms SOA condition. In contrast, similar semantic facilitation effect occurred for both PWS and PNS with the category-level naming task. These results indicated that the semantic encoding of PWS was delayed, and this process was possibly flawed.
Basing on above results, we claimed that PWS were possibly flawed with semantic encoding, and selecting, morpheme extraction, speech coding, and movement execution. This study supported the Multi-factorial model of stuttering and the Two-step Interactive Activation Theory of spoken word production. At the same time, the study has important implications on diagnosis and treatment of stuttering, embodying high theoretical and practical value.

Key words: stuttering, semantic encoding, picture-word interference paradigm, multi-factorial model of stuttering