Here’s the abstract to an ICCA 2018 paper I’m working on with J.P. de Ruiter at the Human Interaction Lab at Tufts. The goal is to use computational linguistic methods (that often use the term ‘backchannel’) to see if all these responsive particles really belong in one big undifferentiated ‘bucket’.

Many studies of dialogue use the catch-all term ‘backchannel’ (Yngve ,1970) to refer to a wide range of utterances and behaviors as forms of listener-feedback in interaction. The use of this wide category ignores nearly half a century of research into the highly differentiated interactional functions of ‘continuers’ such as ‘uh huh’ or ‘wow’ (Schegloff, 1982, Goodwin, 1986), acknowledgement tokens such as ‘yeah’, ‘right’ or ‘okay’ (Jefferson, 1984; Beach, 1993) and change-of-state markers such as ‘oh’ or ‘nå’ (Heritage, 1984; Heinemann, 2017). These studies show how participants use responsive particles as fully-fledged, individuated, and distinctive words that do not belong in an undifferentiated functional class of ‘backchannels’ (Sorjonen, 2001). For this paper we use the Conversation Analytic British National Corpus (CABNC) (Albert, L. de Ruiter & J. P. de Ruiter, 2015) – a 4.2M word corpus featuring audio recordings of interaction from a wide variety of everyday settings that facilitates ‘crowdsourced’ incremental improvements and multi-annotator coding. We use Bayesian model comparison to evaluate the relative predictive performance of two competing models. In the first of these, all ‘backchannels’ imply the same amount of floor-yielding, while the second CA informed model assumes that different response tokens are more or less effective in ushering extended turns or sequences to a close. We argue that using large corpora together with statistical models can also identify candidate ‘deviant cases’, providing new angles and opportunities for ongoing detailed, inductive conversation analysis. We discuss the methodological implications of using “big data” with CA, and suggest key guidelines and common pitfalls for researchers using large corpora and statistical methods at the interface between CA and cognitive psychology (De Ruiter & Albert, 2017).


  • Albert, S., De Ruiter, L., & De Ruiter, J. P. (2015). The CABNC. Retrieved from 9/09/2017
  • De Ruiter, J. P., & Albert, S. (2017). An Appeal for a Methodological Fusion of Conversation Analysis and Experimental Psychology. Research on Language and Social Interaction, 50(1), 90–107.
  • Goodwin, C. (1986). Between and within: Alternative sequential treatments of continuers and assessments. Human Studies, 9(2), 205–217.
  • Jefferson, G. (1984). Notes on a systematic Deployment of the Acknowledgement tokens ‘Yeah’ and ‘Mmhm’ Papers in Linguistics, 17(2), 197–216.
  • Schegloff, E. A. (1982). Discourse as an interactional achievement: Some uses of ’uh huh’ and other things that come between sentences. In D. Tannen (Ed.), Analyzing discourse: Text and talk (pp. 71–93). Georgetown University Press.
  • Sorjonen, M.-L. (2001). Responding in conversation: A study of response particles in Finnish. Amsterdam: John Benjamins Publishing.
  • Yngve, V. H. (1970). On getting a word in edgewise. In Proceedings of the 6th Regional Meeting of the Chicago Linguistic Society, volume 6, pages 657–677, Chicago, IL.