Conversational Reference and Perspective Taking

One aspect of meaning = denotation -> reference

Problem: Conversationalists must identify the referent of a Ss referring expression

  Many referents completely contextualized:

  Deictic expressions (this, that, etc.)

  Definite reference (the x, his x)

  To disambiguate these referents requires coordination: H attempts to take S's perspective, and S aids H in this attempt

Perspective taking required for referent identification (research emphasis), speech act performance, interpersonal effects (e.g., politeness)

Example: "Could you shut the door."

You = who? Which door? Shut it now? Why asking about my ability to shut some door? Etc.

Reciprocal language - perspective-taking relationship: Perspective taking required for successful language use;  perspective taking also an affordance of language use

Common Ground and Mutual Knowledge

Difficulty of perspective taking; each person's perspective is unique (Kelly); ability develops with time (initially egocentric; Piaget); underlies development of 'self' (Mead).

Perspectives (although unique) overlap


Interactants must determine what part is shared


Involves coordination (Herb Clark); similar to other social activities involving coordination (car driving; store purchases)

Coordination > simple perspective taking

H must recognize that S is attempting to take H's perspective into account

E.g.: "Have you seen him?"


Coordination involves establishing what is in common ground (what conversationalists mutually "know")


Much common ground is conventional

Conventions: behavioral regularities (agreed upon) evolved to solve coordination problems (driving on left side of road)

Lexical meaning (arbitrary), syntax, etc.


Conventionality alone is not sufficient:

Deictic expressions (context dependent)

Demonstrative reference: That dog is mean.

Polysemous words (e.g., web)

On-the-spot constructions (Clark): The ham sandwich wants his check.

Mutual knowledge: some theorists (Clark; Schiffer) argue that mutual knowledge of some type is require

A and B mutually know p: iff

A knows p

B knows p

A knows that B knows p

B knows that A knows p

A knows that B knows that A knows p

B knows that A knows that B knows p

ad infinitum

Mutual knowledge thus reflexive: S aware of what S and H know including an awareness of each other's awareness

Mutual knowledge = shared knowledge (A and B know p)

Not plausible psychologically

Heuristic approach (Clark & Marshall, 1983; 1996):

Mutual knowledge based on mutual awareness of shared evidence:

1. Physical co-presence; aspects of environment simultaneously available are in common ground.

E.g., "Can everyone see?"

"Can you open the door?"

Demonstrative reference based on physical co-presence = deictic expressions

E.g., "This theory is problematic."

Past, immediate, future (the exam...)

Problem: Interactants may attend to different aspects of environment

2. Linguistic co-presence

Prior conversations mutually available to all participants; continually expending discourse context

Can treat prior topics as in common ground;

Sometimes marked (e.g., So, like we were talking about earlier....)

Demonstrative reference based on linguistic co-presence = anaphoria (e.g., Shared knowledge isn't enough, it must by mutual knowledge)

Problem: Differential memory for prior discourse

3. Community membership

Certain pieces of information can be assumed to be known by all members of a community

E.g.: BSU community -> John Worthen is pres.

Mascot = Cardinals

English speakers -> syntax, word meanings, etc.

Demonstrative reference based on community membership = proper names (e.g., the "Bulls", Doc Watson, etc.)

Problem: difficult to identify group membership (e.g., type of psychologist);

not all members may know "x" (e.g., John Worthen as pres. may not be identified by all BSU members)

Above problems may be noted and repaired during a conversation;

Common ground is fluid, negotiable, tentative

Alternative view (Sperber & Wilson): mutuality difficult to defend; Instead, conversationalists assume (without actually determining)existence of common ground; i.e., S assumes H can work backward from remark and context to identify referent)

E.g.: "That's a difficult theory to understand"

I assume you can identify referent, but not necessarily that you are aware of the reasoning process by which I make that assumption (i.e., mutuality).

Empirical Research on Perspective Taking

Some Issues:

Do people take the perspective of their interlocutors?

How good are people at doing this?

Do systematic perspective taking errors occur?

I. Is perspective taking attempted?

People vary language as function of interlocutor characteristics (e.g., status);

Not necessarily perspective-taking; variability needs to be function of awareness of others' mental states


Explanations/Directions to others

Kingsbury: directions to out-of-towners (self-identified or inferred via dialect) longer and more detailed directions than to locals

Consider directions on BSU campus.

Laboratory Research

R. Krauss referential communication task (mid 60s)

Fussell & Krauss (1989); describe nonsense figure so can be identified by self or other

Sample figures:


Self: high common ground; descriptions short and figurative

Other: low common ground;descriptionss longs and literal

Identification success:

own self-descriptions > other descriptions > others' self-description

Referential identification moresuccessfull if generated by friend than by stranger (greater common ground)

Greater familiarity = greater common ground =more implicit talk (e.g., Kent, Davis, & Shapiro; easier to reconstruct conversations of strangers than of friends)

II. Perspective-taking biases


Referential identification in Krauss study not perfect

(e.g., 60% success in other identification condition)

- reflects inherent difficulty in perspective taking

Major bias: false consensus effect (social psychology; Ross) - tendency to overestimate extent to which others are similar to oneself; egocentric perspective: assume others know what I know (without checking)

Fussell & Krauss (1991; 1992) - Ss more likely to judge others could identify landmarks (NYC) if they themselves could do so.

Keysar's Perspective adjustment model

Initially, perspective is egocentric; use any available information to interpret, regardless of whether mutually known; fast and effortless (automatic Adjustment stage: attempt to correct initial perspective; slow and effortful

Ss (recipients)took longer to identify referents that are potentially ambiguous (multiple referents exist), even when they know others don't have this information


Example: "Is he there?" H slower if >1 males present, even when S obvious could not be aware of multiple males present.

Common Ground and Conversational Interaction

Common ground changes during conversation (attention shifts, environment changes)

Talk can be used to negotiate common ground (feedback)

Common ground cumulative (linguistic copresense)- referring expressions shortened (e.g.: "looks like a martini glass with legs on one side" -> "martini glass with legs" -> "martini glass" -> "martini")


Shortening a result of common ground verification (via feedback)

Grounding and Collaboration (Herb Clark)

Detailed analysis of task oriented dialogues

Grounding - collaborative process of making one's contribution part of common ground

Presentation phrase: conversational contribution (referent)

Acceptance phase: evidence for understanding contribution

Evidence for acceptance:


1. Direct Acknowledgment - backchannel responses;

part of turn; continuers, assessments, gestures

Important because absence of evidence is unclear

Negative evidence (e.g., clarification request) not always provided when unclear (e.g., Chen)


2. Relevant response - Conversation analytic framework:

Adjacency pairs (e.g., question-answer); appropriate "second" indicates acceptance

Important because relevant response is a simultaneous

acceptance and presentation; short-circuits infinite regress of acceptances


3. Attention to current speaker; (e.g., gaze); S may await confirmation before proceeding.

Grounding techniques demonstrating mutual orientation

1. Alternate description (acceptance of prior and presentation)

2. Indicative gesture (presentation); indicates mutual awareness via physical co-presence

3. Referential installments; left-dislocution - pause - await confirmation before continuing

4. Trial reference; try markers - rising intonation follows from referent

Principle of Least Collaborative Effort


Minimize communicative effort required of all participants (vs. minimal individual effort)

Examples: initial descriptions (Krauss task) brief, and await feedback tailor description



Strength - Emphasis on collaboration; mutual establishment of common ground is collaborative, a process

Problems - understanding always revealed in talk?

interpersonal explanations (e.g., self-repair, other- identification, acceptances (save face)

Perspective Taking and Attitudes

Lexical choice can indicate attitude toward referent

But only if H takes S's perspective

Higgins & Rholes (1977); describe target for another (who likes/dislikes target); S's alter descriptions (stubborn vs. independent) as function of H's attitude

- later memory affected

- represents shared not mutual knowledge (similar to self- presentation - don't want others to recognize intention to self-present)

Perspective Taking and Speech Act Recognition

Speech act - action performed with utterance

Coordinated perspective taking required for successful recognition; beyond conventionality for nonconventional ISAs

Egocentric bias - Keysar - overhearers fail to consider possibility of alternative interpretations (take an egocentric perspective); Mark - restaurant - to recipient "Marvelous, simply marvelous"; Ss believe recipient interprets as sarcasm if Mark had a negative experience (even though recipient would have no awareness of this)

Holtgraves; Ss take perspective of H and S; interpretations vary as function of perspective

Eg: "What did you think of my presentation?"

"It's difficult to give a good presentation"

H perspective: convey negative opinion

S perspective: avoid topic