Sentences change in conversation. They can “shrink” to fragments. But they can also “grow” to include material that can only be interpreted in the context of a conversation. For example, Canadian eh as in “I have a new dog, eh?” is used to ask the addressee (A) for confirmation that s/he knows that the proposition expressed in the utterance (p) is true. Hence, eh is A-oriented in at least two ways:
i) … the speaker (S) indicates that s/he believes that – but is not sure if – A knows that p and
ii) … S requests a response from A and hence marks the end of his/her turn
Based on the empirical properties of confirmationals (such as eh) as well as other discourse markers across a number of unrelated languages, I argue that the human language faculty includes a Conversation Competence. That is, I argue that our conceptualization of Universal Grammar (UG) has to include a layer of structure that hosts units of language dedicated to expressing S’s attitude towards p; S’s assessment of A’s attitude towards p; as well as what S wants A to do with A. The analysis is couched within the Universal Spine Hypothesis (Wiltschko 2014) which allows for an exploration of the association of sound and meaning to an abstract functional spine.
The exploration of conversation competence brings with it a methodological problem: how do we investigate our abstract knowledge of conversational sentences? While the traditional method of conducting corpus analyses allows the researcher to discover certain patterns it does not allow us to tap into Conversation Competence (no negative data, no possibility to control specific contextual factors, no straightforward way for a comparative investigation). In this talk I report on how we (the eh-lab at UBC) explore Conversation Competence by means of corpus analyses alongside controlled story-board elicitations, online surveys, as well as traditional elicitation techniques (field-work).