Preference Structure

Within Conversation Analysis, “Preferences” are organizational principles that guide speakers in how best to construct their talk, and how to understand what they’re hearing when they are interacting socially. Preferences assist people in managing their behaviour when they talk, so that what they say conforms to societal expectations. Every time we speak about people in conversation, we  make choices about the most appropriate expressions to use. Sometimes a first name (Fred) will be ideal, whereas at other times a title and surname might be more appropriate (Mr Smith), whereas on yet other occasions a pronoun will be sufficient (he). Preferences pertaining to person reference items help us make these decisions. There are at least three preferences that assist Murrinh-Patha speakers to make these decisions. They also appear to be the same preferences driving the grammatical changes in their language.

When we want to speak about someone we haven’t spoken about previously, that we suspect is known to the people we’re conversing with, there is an expectation that we will use words that can allow the people we’re talking to to recognize the person in question. For this we use ‘recognitionals’.

RECOGNITION: If possible, use recognitional reference forms – forms that invite targeted recipients to identify the person being referred to from among the universe of people that they know (about) AND that they suspect their interlocutor expects them to know (about).

In many cultures, personal names are particularly good recognitionals (especially first names in most English speaking cultures). When you hear a (first) name, you invariably try to match the name to a person you know. Names tend to trigger this recognition process. In most cultures, names also tend to be relatively succinct. As such, they satisfy a preference for Minimization.

MINIMIZATION: If possible, use single (or minimal) reference forms, as opposed to multiple (or excessively lengthy) reference forms.

Because in many cultures names are both highly recognitional and succinct, they tend to be the default forms to use in conversation for referring to a person for the first time. Using kin terms (your cousin), descriptions (your friend and mine) are other possibilities that might be sufficiently recognitional and perhaps also minimal, though not necessarily. In Murrinh-Patha, names ARE the default recognitionals for initial mentions. They are regularly used to secure recognition. However, there are also plenty of occasions when it is not appropriate to use a particular person’s name. In these situations, other forms should be chosen to secure that recognition. The preference that handles this expectation can be technically expressed as follows:

CIRCUMSPECTION: If possible, observe culturally specific and/or situationally specific constraints on reference and avoid the default reference forms.

This preference advises speakers on how to respond to a range of taboos that limit the use of certain people’s names (the recently deceased, opposite-sex siblings, certain in-laws, etc.).

I won’t demonstrate the interaction of these preferences here. For that you are directed to the paper. However, note that when worded in this fashion, none of these preferences instruct speakers which forms to choose. Rather, they are guidelines that advise them HOW to make the appropriate choices. They provide design constraints for the sorts of words to select. In conversational interaction, taboos that limit the use of personal names tend to bias the selection of kin-based expressions in daily conversation. Over the millennia, the same biasing effects have given rise to kinship inflections in grammar.

Next, Invisible Hand processes


Joe Blythe is a research fellow at the School of Language Studies, University of Melbourne, Australia and an affiliate of the Language and Cognition Department at the Max Planck Institute for Psycholinguistics, Nijmegen in the Netherlands. The research was funded by the Australian Research Council (DP110100961 and DE130100399) and the European Research Council (240853).
Advertisements