This paper surveys and evaluates some techniques for eliciting requirements of computer-based systems, paying particular attention to how they deal with social issues. The methods surveyed include introspection, interviews, questionnaires, and protocol, conversation, interaction, and discourse analyses. Although they are relatively untried in Requirements Engineering, we believe there is much promise in the last three techniques, which grew out of ethnomethodology and sociolinguistics. In particular, they can elicit tacit knowledge by observing actual interactions in the workplace, and can also be applied to the system development process itself.

This paper examines conversational behavior which previous research suggests is differentiated on the basis of sex. Interaction is conceptualized in terms of a sexual,division of labor wherein men dominate conversation and women behave in a supportive manner. The literature raises the question of whether these differences in conversational patterns are tied to power as well as sex. A study was designed to determine which of a set of variables reflecting conversational duties and privileges are linked to power, to sex, or to both. The data were coded from interactions of intimate couples divided among those with both partners sharing power equally and those where one partner has more power. Three types of couples were compared: cross-sex couples, male couples, and female couples. Interruptions and back channels are linked to power regardless of the sex of the actor, as are tag questions, although the rarity of their occurrence makes any conclusions tentative. The more powerful person interrupts his or her partner more and produces lower rates of back channels and tag questions. Talking time and question asking seemed linked to both sex and power, though not in any simple way. The results of the analyses of minimal responses and overlaps proved inconclusive. In recent years there has been a growing re-search interest in sex differences in speech (e.g., Thorne and Henley, 1975; Thorne et al., 1983). Conversational behavior, it was once argued, can be viewed as having a "male dialect " and a "female dialect " (Kramer, 1974). More recent commentators feel that such a conceptualization exaggerates and at the same time oversimplifies the differences between men's and women's speech (Thorne et al., 1983: 14). However, neither these authors, nor any others, deny that there are significant sex differences in verbal interaction. As various sex differences were observed, some authors began to look at possible reasons for their existence and at their implications.

In this article the authors argue that claims of sex diflerences in interruption behavior should not be uncritically accepted as thereare limitations in prwious resmrch that make such acceptancequestionable. Thefrequencyof interruption was examinedovera portion of the early life span (Grades 4 and 9 and college). %enty-minute structured conversa-tions of 90 dyads (30 male, 30 fmale, and 30 mixed sex) were scoredfor four types of interruption, and both dewlopmental and sex diyemces in interruption behavior were examined. lnterruption frequency did not change over age or across dyads of different sex composition. Males did not interrupt any more than females did andfPmnles were interrupted by theirpartnm asfrequently as males were interrupted by theirs, with one exception: Grade 9 females were interrupted more by theirfemale partners. Interruptions were asymmetrically distributed in same-sex and opposite-sex dyads; however, the asymmetry in opposite-sexdyads was not predictablefrom sex ofsubject orsexofpartner. That is, males did not interrupt females any more than females interrupted males. The authors conclude that wholesale acceptance of sex diyerences in interruption behavior is not warranted. P articipants in conversations are expected to follow the tumtaking system, which specifies that only one speaker may talk at a time. Thus interruption is prohibited. An interruption event has been defined as an instance of simultaneous speech that involves "a deep intrusion into the internal structure of a speaker's

This paper surveys and evaluates some techniques for eliciting requirements of computer-based systems, paying particular attention to how they deal with social issues. The methods surveyed include introspection, interviews, questionnaires, and protocol, conversation, interaction, and discourse analyses. Although they are relatively untried in Requirements Engineering, we believe there is much promise in the last three techniques, which grew out of ethnomethodology and sociolinguistics. In particular, they can elicit tacit knowledge by observing actual interactions in the workplace, and can also be applied to the system development process itself.