Question and answer sequences in conversation

The study in a sentence

Different linguistic features are observed in preferred (or socially affiliative) responses to questions as compared to dispreferred (or socially disaffiliative) responses. We can see this pattern in close reading of naturally occurring data using the tools of Conversation Analysis, and the pattern generalises across a large dataset of phone calls. Words like 'um' and 'well' play an important communicative role in the sequence of conversation as part of a finely-tuned interactional system.


The question

Can you tell what kind of answer someone is going to give to a question?

  • Alternative potential responses to questions are not treated as equal; generally, one will be the preferred response and the other the dispreferred response.

  • Preferred responses support social solidarity between the two speakers whereas dispreferred responses threaten social solidarity.

Are preferred responses produced with different linguistic features from dispreferred responses, in terms of speed of response and/or complexity of language structures used?

Key concept

Language as action

In conversation, we use language to perform social actions: ask questions, report good news and bad, complain and compliment, agree and disagree, express surprise, disbelief, or disgust, request assistance or offer it and so on. In conversation, social actions do not occur in isolation; they occur together to form coherent sequences of action.

The most basic and common type of sequence in conversation is an adjacency pair: a pair of two turns/actions, each produced by a different speaker

  • The first pair-part (FPP) initiates the sequence and makes relevant a specific next turn/action

  • The second pair-part (SPP) responds to the first and completes the sequence

      • Examples: greeting-greeting, question-answer, invitation-acceptance/declination

Are preferred responses produced with different linguistic features from dispreferred responses?

Do these patterns generalise across larger samples of data?

speaker 1: why don't you come and see me sometimes? Speaker 2 I would like to
Speaker 1 If you'd care to come over and visit a little while this morning I'll give you a cup of coffee. Speaker 2 Uh huh well that's awfully sweet of you I don't think I can make it this morning hee uhm I'm running an ad in the paper and uh I have to stay near the phone.


Conversation Analysis gives us tools to draw out the difference between preferred and dispreferred responses to questions (i.e. second pair-parts, SPP):

  • preferred responses tend to be produced quickly and have simple forms

  • dispreferred responses tend to be delayed and have complex structures.

In this study the authors examined 185 telephone calls to find 195 responses to requests, offers, invitations, and proposals. They looked at two measures:

  • the duration of the silence between the first and second pair-parts

  • a count of three linguistic features at the beginnings of the second pair-parts: in-breaths, hesitation particles (e.g. “um” or “uh") and “well” prefaces.

Frequency plot of the duration of the silence between questions and answers in the corpus, split by preferred or dispreferred answers
Table showing percentage of turn-initial features observed at start of preferred versus dispreferred answers.

The answer

Responses to questions can be socially affiliative (preferred) or socially disaffiliative (dispreferred)

  • The timing and construction of preferred and dispreferred responses differs systematically

  • This allows listeners to predict whether a response is likely to be preferred or dispreferred

    • If a noticeable silence occurs before someone responds, chances are the response will be dispreferred

    • If a response begins with an in-breath, hesitation particle, or “well”, chances are the response will be dispreferred

Fillers are not just fillers

The findings of this study show that words like “um” and “well” are part of a finely-tuned interactional system – banning their use would damage communication, not improve it.

  • In responses to questions, these words are important signals that allow others to recognize quickly the type of response that will be produced and so plan their next turn at talk.

  • Efforts to ban words like “um” and “well” are misguided.

Classroom activities

Lead in task

What would you normally say to accept or decline an invitation?

Extension task

Try out some Conversation Analysis for yourself.

In more detail

A longer explanation of the research study

The slides handout from Kobin Kendrick's talk at the 2019 workshop

Meet the author

Kobin Kendrick

Kobin teaches modules in Conversation Analysis including the Language of Turn & Sequence, Language as Action and Multimodality.

Read the paper

Kendrick, K. H., & Torreira, F. (2015). Thetiming and construction of preference: Aquantitative study. Discourse Processes,52(4), 255-289. download