研究采用长时重复启动范式，操纵学习和测验材料中说者身份的一致性，考察说者身份信息对较浅的词汇通达 (实验1) 和较深的概念理解 (实验2) 的影响。实验1词汇决定任务发现，对于学过的词，说者身份一致条件的正确率显著高于不一致条件，表明说者身份信息和语言信息以整合的方式影响词汇通达。实验2类别决定任务发现，对于未学过的词，说者身份一致条件的正确率显著高于不一致条件，表明说者身份信息以独立的方式影响概念理解。基于上述结果和先前的相关理论，研究尝试提出了包含身份信息加工的口语词汇加工的新观点，有助于更具社会性和生态性地解释口语词汇的认知加工。

Considering the speaker’s identity information provides a more social and ecological explanation of the cognitive processing of spoken words. However, whether and how speaker’s identity information affects spoken word processing is controversial. The abstractionist view (including the early and developmental abstractionist views) and the episodic view hold different opinions on this issue. Moreover, previous studies have employed different experimental tasks that provide different evidence for these views. Based on our analyses of these previous studies, we propose that existing views may each be suitable for explaining different processes in spoken word processing. It is necessary to examine the role of speaker’s identity information in spoken word processing requiring different processing depths. Based on this background, the present study focused on whether and how speaker’s identity information affected lexical access and conceptual comprehension in spoken word processing. Addressing these issues can help us better understand spoken word processing.

The present study conducted two behavioral experiments and adopted the classic long-term repetition priming paradigm to minimize possible interference from explicit experimental tasks. Specifically, Experiment 1 adopted a lexical decision task to examine whether and how speaker’s identity information affected lexical access in spoken word processing. Eighty-eight participants were recruited for the experiments and randomly divided into two groups (speakers’ identities were consistent vs. inconsistent). The experiment contained learning and test phases. In the consistent group, participants would hear stimuli spoken by a male in both the learning and test phases; in the inconsistent group, participants would hear stimuli spoken by a male in the learning phase and by a female in the test phase. The experimental materials consisted of 36 real words (e.g., “/yi1fu2/”, which means clothes in English) and 36 pseudowords (i.e., pronounceable but meaningless nonwords, e.g., “/ju4hong2/”). Participants needed to judge whether the auditory word was real or pseudo. Experiment 2 adopted a category decision task to examine whether and how speaker’s identity information affected conceptual comprehension in spoken word processing. The participants and design were the same as Experiment 1, with 36 biological words (e.g., “/xiao3cao3/”, which means grass in English) and 36 non-biological words (e.g., “/qian1bi3/”, which means pencil in English) as experimental materials. Participants needed to judge whether the auditory word was biological or non-biological.

In Experiment 1, the performance of learned words was better than that of unlearned words, indicating a stable repetition effect. More importantly, in the overall analysis (including real words and pseudowords), for learned words, the accuracy of the consistent condition was significantly larger than the inconsistent condition; for unlearned words, there was no significant difference between the consistent and inconsistent conditions. Further analysis revealed that the results for pseudowords were the same as the overall analysis, but for real words, there were no significant differences in either accuracy or reaction time between the consistent and inconsistent conditions for both learned and unlearned words. In Experiment 2, the response times of learned words were significantly shorter than those of unlearned words, suggesting the repetition effect of learned words. However, in contrast to Experiment 1, the accuracy of the consistent condition was significantly larger than the inconsistent condition for unlearned words, while there was no such difference for learned words.

Speaker’s identity information influences the processing of spoken word differently depending on the processes. Specifically, speaker’s identity consistency facilitation for learned words in the lexical decision task suggested that the representation of the speaker’s identity was integrated with linguistic information and would affect lexical access integrally, supporting the episodic view. In contrast, speaker’s identity consistency facilitation for unlearned words in the category decision task suggested that the speaker’s identity and linguistic information would be represented separately and affect conceptual comprehension independently, supporting the developmental abstractionist view. Integrating the developmental abstractionist and episodic views helps us better understand spoken word processing.