PSY 369: Psycholinguistics

Language Production & Comprehension:

Conversation & Dialog

Dialog is the key Why so little research on dialog?

Most linguistic theories were developed to account for sentences in de-contextualized isolation

Dialog doesn’t fit the competence/performance distinction well

Hard to do experimentally Conversations are interactive and largely unplanned

Pickering and Garrod (2004) Proposed that processing theories of language

comprehension and production may be flawed because of a focus on monologues

Processing models of dialog Pickering and Garrod (2004)

Interactive alignment model Alignment of situation

models is central to successful dialogue

Alignment at other levels is achieved via priming

Alignment at one level can lead to alignment at another

Model assumes parity of representations for production and comprehension

Assumptions of the model

1. Alignment of situation models comes about via an automatic, resource-free priming mechanism

2. Representational parity between comprehension and production

3. Alignment at one level leads to alignment at other (interconnected) levels

4. There is no need for explicit perspective-taking in routine language processing

Assumptions of the model

1. Alignment of situation models comes about via an automatic, resource-free priming mechanism

Garrod & Anderson (1987) The maze game

Pairs played a co-operative computer game Move position markers through a maze of boxes

connected by paths

Each player can only see his/her own start, goal and current positions

Some paths blocked by gates (obstacles) which are opened by switches

Gates and switches distributed differently for each player

Players must help their partner to move to switch positions, to change the configuration of the maze

Assumptions of the model

1. Alignment of situation models comes about via an automatic, resource-free priming mechanism

Garrod & Anderson (1987) The maze game

1-----B: .... Tell me where you are?

2-----A: Ehm : Oh God (laughs)

3-----B: (laughs)

4-----A: Right : two along from the bottom one up:

5-----B: Two along from the bottom, which side?

6-----A: The left : going from left to right in the second box.

7-----B: You're in the second box.

8-----A: One up :(1 sec.) I take it we've got identical mazes?

9-----B: Yeah well : right, starting from the left, you're one along:

10----A: Uh-huh:

11----B: and one up?

12----A: Yeah, and I'm trying to get to ...

Assumptions of the model

1. Alignment of situation models comes about via an automatic, resource-free priming mechanism

Garrod & Anderson (1987) The maze game

41----B: You are starting from the left, you're one along, one up? (2 sec.)

42----A: Two along : I'm not in the first box, I'm in the second box:

43----B: You're two along:

44----A: Two up (1 sec.) counting the : if you take : the first box as being one up :

45----B: (2 sec.) Uh-huh :

46----A: Well : I'm two along, two up: (1.5 sec.)

47----B: Two up ? :

48----A: Yeah (1 sec.) so I can move down one:

49----B: Yeah I see where you are:

Assumptions of the model

1. Alignment of situation models comes about via an automatic, resource-free priming mechanism

Garrod & Anderson (1987) The maze game

Path descriptions (36.8%)

See the bottom right, go two along and two up

Co-ordinate descriptions (23.4%)

I’m at C4

Line descriptions (22.5%)

I’m one up on the diagonal from bottom left to top right

Figural descriptions (17.3%)

See the rectangle at the bottom right, I’m in the top left corner of that

Assumptions of the model

1. Alignment of situation models comes about via an automatic, resource-free priming mechanism

Garrod & Anderson (1987) The maze game

Pairs converge on different ways of describing spatial locations

Entrainment on a particular conceptualization of the maze

But little explicit negotiation Entrainment increases over the course of a game

Description schemes as local ‘languages’

Rules for mapping particular expressions onto interpretations with respect to a common discourse model

Once the meaning of a particular expression is fixed, players try to avoid an ambiguous use of that expression

Assumptions of the model

1. Alignment of situation models comes about via an automatic, resource-free priming mechanism

Garrod & Anderson (1987) The maze game

Entrainment emerges from a simple heuristic

Formulate your output using the same rules of interpretation as those needed to understand the most recent input

Representations used to comprehend an utterance are recycled during subsequent production

Leads to local consistency Helps to establish a mutually satisfactory description

scheme with least collaborative effort

Assumptions of the model

2. Representational parity between comprehension and production

Parity important for interactive alignment

We don’t go around repeating other people’s utterances!

Comprehension-to-production priming (BPC, 2000)

Priming from sentences which were only heard Suggests that representations shared across modalities

Equivalent to production-to-production effects?

E.g. Bock (1986), syntactic priming in language production tasks

Assumptions of the model

3. Alignment at one level leads to alignment at other (interconnected) levels

Cleland & Pickering (2003) Semantic boost

Primes either pre (the red sheep) or post nominally (the sheep that is red) modified NPs

Same (sheep to sheep), semantically related (goat to sheep), unrelated (knife to sheep)

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

RED SHEEP RED GOAT RED KNIFE

Pro

port

ion "

red s

heep"

targ

et

resp

onse

s

The red XXX The XXX that's red

Bigger priming effect when the prime noun is semantically related to the noun in the target

Branigan, Pickering, & Cleland (2000)

Lexical boost similar effect with same verb

Assumptions of the model

4. There is no need for explicit perspective-taking in routine language processing

If communication is successful, interlocutors’ situation models come to overlap

Implicit common ground

Overlap may be small to begin with

But via alignment, it increases over the course of a conversation

What looks like audience design is simply a by-product of good alignment

Full common ground only consulted when there are sufficient processing resources available

Summary “People use language for doing things with

each other, and their use of language is itself a joint action.” Clark (1996, pg387) Conversation is structured

But, that structure depends on more than one individual Models of language use (production and

comprehension) need to be developed within this perspective

Interactive Alignment model is a new theory attempting to do just this

Review for Exam 4 Chapters 13, 14, 15 (read 16 for interest, but I

won’t test on it) Same format as the last 3 exams General topics:

Language Production Conversation & dialog

I have fixed the link to the review sheet

Review for Exam 4 Language production

Paradox: form over meaning is preserved Speech errors - observational & experimental Tip-of-the-tongue Lexical bias Grammaticality constraint Models of speech production

Levelt’s model Dell’s model Lexical bias effect, mixed errors