許多母語的研究發現性別在會話互動中扮演重要的角色，但鮮少有研究探討在外語的對話裡性別如何影響外語學習者之間的會話互動，以及這些影響是否與性別而異的談話風格有關。本研究以互動假說（Interaction Hypothesis）為基礎，並採用Lakoff與Tannen（1984）的談話風格分類，探討性別如何影響任務導向互動（task-based interactions）中語言相關片段（language-related episodes）的多寡及解決情形，並進而影響外語學習者的學習機會。24位臺灣籍的英語學習者（共12位女性及12位男性）參與本研究。研究結果發現說話者性別和性別配對（異性組或同性組）都不會對語言相關片段的提問及回答頻率產生顯著影響。然而，性別配對卻顯著影響到回答的答案是否正確；比起同性組的表現，女性和男性在異性組時皆比較容易給予正確的解答。至於談話風格，疏遠型和清晰型兩種談話風格呈現性別差異；疏遠型有較多男性，而清晰型有較多女性。說話者的談話風格並不會顯著影響她／他自己在語言相關片段的提問頻率及回答情形。然而，有趣的是說話者的談話風格可能會影響到與談人的回答意願。疏遠型人（較多男性）較可能提高與談人的回答意願，因而可以聽到較多答案；相反地，清晰型的人（較多女性）較容易降低與談人的回答意願，因而聽到較少答案。

A wealth of research on native language conversational interaction has indicated that women and men talk differently, but little is known about how gender plays a role in L2 conversation, especially how gender-differentiated speech styles affect the interlocutor's willingness and ways to talk, thus creating different learning opportunities for female and male learners. Drawing on Long's Interaction Hypothesis, the present study investigated how gender affects incidence of negotiation for meaning in task-based interactions, focusing on language-related episodes (LREs). Subjects comprised 24 Chinese-speaking learners of English at the college level, with an equal number of females and males. Each subject completed two text-reconstruction tasks in English with a same-gender partner and a mixed-gender partner on different days. Each task was followed by a stimulated recall session in Chinese, which was to collect participants' comments on their interactions for the analysis of speech styles. Lakoff and Tannen's (1984) categorization of speech styles in native language conversation was adopted for the analysis of speech styles in L2 interactions. The results indicate that the gender of the learners participating in task-based interactions did not significantly influence the interactional patterns in terms of the incidence or resolution of LREs. However, it was found that both female and male learners resolved LREs in a more targetlike way when working with mixed-gender partners than with same-gender partners. Gender-differentiated speech styles were also observed, with more males adopting the distance style and more females the clarity style. Speakers of the distance style were found to be more likely to increase their interlocutors' willingness to offer LRE resolutions and thus received more input. In contrast, speakers of the clarity style tended to turn off their interlocutors' willingness to provide LRE resolutions and thus received less input. The results of the study shed some light on our understanding of the relationship between gender influence on conversational interaction and SLA.