1 Does gender-related variation still have an effect, even when topic and (almost) everything else is controlled? Hans-Jörg Schmid 1. Introduction 1 Previous research suggests that the gender of the participants involved in a conversation can affect their use of language in three different ways: – as a function of the gender of the person speaking: female vs. male (see Mulac, Bradac and Gibbons 2001 and Newman et al. 2008 for extensive surveys) – as a function of the gender of the person addressed vis-à-vis the person speaking: same-gender talk vs. mixed-gender talk (Bilous and Krauss 1988, Mulac et al. 1988, Hirschman 1994, Hancock and Rubin 2014) – as a function of the interaction of the two: effects of the gender of the speaker that are contingent on same-gender or mixed-gender talk, or vice versa (McMillan et al. 1977, Palomares 2008). Previous research also suggests, however, that the observable differences could simply be due women’s and men’s preferences regarding topics of conversation (Newman et al. 2008: 229). Women have been claimed to spend more time talking about people, past events and personal topics, while men’s favourites include job-related topics, sports, politics and technology. Of course, topic choices have a strong effect on linguistic choices. For example, talk about peo- ple and past events is much more likely to contain larger numbers of proper nouns, personal and possessive pronouns, temporal and spatial adverbials as well as past tense verbs than talk about politics or cutting-edge technology. Linguistic investigations that seek to identify the effect of gender on linguistic variation are thus well advised to take the confounding effect of topic into consideration. Since topic keeps changing and drifting in casual conversation, it has turned out to be extremely difficult to control this variable in quantitative corpus studies. It is precisely this dilemma which forms the backdrop and motivation for the present study. What is presented here is actually not much more than a methodological exercise whose key aim is to show to what extent gender-related effects on language use can still be observed if the variable TOPIC is kept constant. The characteristics of a very special dataset are exploited to reach this goal: the HCRC Map Task Corpus collected at the universities of Glas- gow and Edinburgh in the 1980s (see Anderson et al. 1991 and http://groups.inf.ed.ac.uk/maptask/#top for more information). 2 This corpus consists of tran- scipts of 128 dialogues, all of which had the same setup and involved the same task: […] two speakers sit opposite one another and each has a map which the other cannot see. One speaker -- designated the Instruction Giver -- has a route marked on her map; the other speaker -- the Instruction Follower -- has no route. The speakers are told that their goal is to reproduce the Instruction Giver's route on the Instruction Follower's map. The maps are not identical and the speakers are told this explicitly at the beginning of their first session. It is, however, up to them to discover how the two maps differ. (http://groups.inf.ed.ac.uk/maptask/maptask-description.html). 1 I am greatly indebted to Franziska Günther, Daphné Kerremans and Gill Woodman for their comments on an earlier version of this paper, and to Alexander Bauer from the Statistical Consulting Unit at Ludwig Maximil- ians University Munich directed by Helmut Küchenhoff for his expert advice on the statistics used in this pa- per. 2 I would like to thank the compilers of the Map Task Corpus for sharing their material with the scientific community and Jean Carletta from the University of Edinburgh for directing me to pertinent information on the HCRC Map Task corpus website. In: Jocelyne Daems, Eline Zenner, Kris Heylen, Dirk Speelman and Hubert Cuyckens (eds.), Change of Paradigms – New Paradoxes. Recontextualizing Language and Linguistics, Berlin - Boston: de Gruyter Mouton, 327-246, 2015.