Turn-Taking
Presented by Sarah Friedman
April 4, 2005
Based on:
Sacks, H., Schegloff, E.A., Jefferson, G. (1974). A
Simplest Systematics for the Organization of
Turn-Taking for Conversation. Language, 50, 696735.
Introduction
 Report
on turn-taking system for
conversation
 Research using audio recording of
naturally occuring conversations
Need for a New Turn-Taking Model


Investigators of small group interaction
have difficulty explaining behaviors using
turn-taking systems
Researchers of “interview” behavior are
concerned with
–
–
–
The distribution of talk among the parties
Silences
The way talk shifts from one speaker to
another
These researchers dealt with turn-taking
questions, but unsatisfactorily because of
weaknesses in turn-taking models
Reasons for Research


Obvious that one party talks at a
time, there are techniques for
allocating turns, etc.
Can be applied to many contexts
Turn Allocation

Next turn is allocated by
1. Current speaker
2. Self-selection
Rules Governing
Turn Construction
1. For any turn, at a transitionrelevant place:
a. If “current speaker selects next” then
the person selected is obligated to
take the next turn.
b. If not, self-selection may be instituted.
c. Or, the current speaker may continue
unless someone self-selects.
2. If 1c occurs, then rules a-c reapply
at the next transition-relevant place
The System Accounts for:



Speaker change
One party talking majority of the
time
Occcurrences of more than one
speaker at a time are common but
brief
Types of Overlap

Accounted for by rule 1b (self-selection):
Competing for next turn.
Lil:
Bertha’s lost, on our scale, about
fourteen pounds.
Damora: Oh[::no::,
Jean:
[Twelve pounds I think
wasn’t it =
Daisy: [Can you believe it?
Lil:
[Twelve pounds on the Weight
Watcher’s scale.
Types of Overlap (cont…)

Assume you know how the speaker will
finish.
A: Well if you knew my argument why did you
bother to a:[sk.
B:
[Because I’d like to defend my
argument.

The speaker added optional elements
that can go after completion.
A:
B:
Uh you been down here before
[havenche,
[Yeh.
Turn Order/Size


Turn Order
– Not fixed yet it’s not random
– Bias: Speaker before the current
speaker is selected as the next
speaker
Turn Size
– Not fixed
– Why?
 Because of unit types (single-word
turns, single phrase turns, etc.)
 Because of rule 1C (current speaker
can continue) the speaker can
produce more than one unit type
More on Turns



The length and content of conversations
are not specified in advance
Anyone can be the next speaker
Number of parties can vary.
–


With 4 you can have multiple conversations
Turns begin at “possible completion
points”
Repairs exist for errors:
–
–
Who me?
Excuse me?
Continuous or Discontinuous

Talk can be continuous or discontinuous
–
–
Continuous = minimum gap or overlap
Discontinuous = a current speaker has stopped
and no speaker starts, and non-talk is a lapse.
J: Oh I could drive if you want me to.
C: Well no I’ll drive (I don’ m//in’).
J: hhh.
(1.0)
J: I meant to offah.
(16.0)
J: Those shoes look nice when you keep on putting
stuff on ‘em.
C: Yeah I ‘ave to get another can cuz cuz it ran
out….
Turn Allocation Techniques: Select
Next Speaker
1. “Type of Sequence” parts. (Adjacency
pairs).
Examples:
– Complaint/denial
Ken: Hey yuh took my chair by the way an’t
I don’t think that was very nice.
Al:
I didn’ take yer chair, it’s my chair.
– Compliment/rejection
A:
I’m glad I have you for a friend.
B:
That’s because you don’t have any
others.
Turn Allocation Techniques:
Select Next Speaker (cont…)
2. Repeat parts of prior utterance with
a question intonation or one word
question.
Ben:
Lori:
Ben:
They gotta – a garage sale
Where.
On Third Avenue.
Turn Allocation Techniques: Select
Next Speaker (cont…)
3. Tag question
 You know? Don’t you agree?
4. Introduce social identities
 Two couples speaking. An invitation is
made by a speaker to go to the
movies. The next speaker should be a
member of the other couple.
Turn Allocation Techniques:
Self-Selection
1. Starting First
2. “Second Starters”
R:
F:
K:
D:
R:
Hey::, the place looks different
Yeah::hh.
Ya have to see all ou[r new-*
[It does?*
Oh yeah
Consequences of the Model

Motivation for listening to all utterances
in a conversation
–
–

To see if you are the next speaker
If you want to speak, make sure no one else
was selected to speak
When a new speaker is selected, the
speaker has to perform the second part
of the adjacency pair.
–
Shows he understood the prior turn’s talk as
the first part