Class Notes (809,255)
Canada (493,593)
Psychology (4,098)
PSY274H5 (129)
Lecture 5

Lecture 5 - Conversational Behavior

6 Pages
Unlock Document

University of Toronto Mississauga
Craig Chambers

PSY274: Intro to Human Communication Lecture 5: October 15, 2013 Conversational Behavior So Far: - The nature of human communication - Human vs. animal communication; spoken language, sign language, co-speech gesture Beginning today: Unit II - Language in relation to other psychological domains Language In Relation to Other Psychological Domains - Cognition (thinking) - Social decisions (case study: advertising) - Interpersonal interaction (conversation) Today: Conversational Behavior - Starting point: It is not enough to know the inventory of symbols in your language, accompanying paralinguistic and gestural cues, etc. - Successful communication involves knowing HOW to use language in a socially sophisticated way Topics to be Covered - Starting point: using artificial intelligence (AI) to gain insights into conversation behavior - “Cooperation” in conversation o Turn-taking o Sentence structuring o Common ground and reference - Details in the act of speaking: disfluencies and their effects o Production o Preparation Using artificial intelligence to gain insights into the psychology of conversational interacton Psychology of Communication: Goals of Discipline - Essentially an enterprise of “Reverse engineering” - Start with fully functioning system (mind/brain), try to deconstruct how it works scientifically Artificial Intelligence - Goal runs in opposite direction: to create/build a fully functioning system - Successes and failures can highlight the important properties of the fully functioning system Historical Development - 1960s conversational simulation (e.g, ELIZA) using text-based system “Turning Test” Competition - Features: o Stimulus/response (doesn‟t speak unless spoken to) o “Canned” sentences, selected based on occurrence of particular words OR partial repetition of previous sentence o Simple pronoun reversals “I” (in input)  “you” (in response) PSY274: Intro to Human Communication Lecture 5: October 15, 2013 More Contemporary Versions (1990s, 2000s): - “Chatterbots”/ “Chatbots” - Loebner Prize o More sophisticated: e.g., “ALICE” o Uses ever-expanding knowledge base to tailor sentences to the user by “guessing” age, background, interests, occupation, etc. through context clues The “Lessons” From Al Initiatives: - Even simple things (from the perspective of human communication) can require a lot of build-in knowledge and complex decision systems - … an initial example: the problem of “context” in conversation (see Napoli text for numerous illustrations) Examples Showing Context-Dependence - Elliptical utterances o Q: Where are you going? o A: School - Superficial nonsequiturs o Q: Can you come to my party on Sunday? o A: I have an exam on Monday - Ambiguities o There‟s a cup of coffee on the table. I just made it. o Fred ate the pie on the windowsill o Fred ate his dinner in the restaurant Enrichment in Interpretation - Have you been to Paris? (ie. ever, in your life) - Have you eaten breakfast? (ie. today) - The new Boeing 767 is very aerodynamic - The new Boeing 767 is very comfortable A Second Example Highlighting the Complex “Ingredients” of Conversation: - Makin a chat-room avator produce deictic gestures at relevant points, based on typed input…. A Third Example: Embodied Agents PSY274: Intro to Human Communication Lecture 5: October 15, 2013 - Video-based representations of individuals who “speak”, “listen”, acquire knowledge, recognize intentions, produce appropriate gestures, etc. - Used as an interface with systems having a specific application (Educational, commercial, etc.) - The virtual conversational partner must be lifelike/natural to be useful (ie., to be worth the effort, compared to a keyboard + mouse interface) Embodied Agents: - With more human-like interfaces, it is even more important to approximate natural conversation - One illustration: when the natural rhythm of conversation clashes with the time lags required for a computer to generate responses… - “Is this house in a good neighborhood?” o Compare 1 second to respond: “Yes”  3 seconds to respond: “Yes” Next Topic: COOPERATION in Conversation Turn-Taking - Typical conversation: each participant speaks and listens - How do we know when it‟s our turn? - “Reactive” accounts of turn-taking - Current speaker provides signals that turn is ending - Listener “reacts” by responding to signals and taking over Reactive Approaches: Turn-Yielding Signals (according to Duncan, 1972) - Drop in pitch - Lengthened syllable at turn-final utterance - Drop in loudness - Termination of hand gestures - Certain stereotyped expressions (example, “you know”, “or something”) - Completion of well-formed sentence Testing the Reactive Account: Can These Signals Predict When a Listener Attempts to Take a Turn? - 0 cues produced  listener attempts to take a turn 10% of time - 3 cues  listener attempts to take a turn 33% of time - All 6 cues  listener attempts to take a turn 50% of time - Answer: Sort of. (Still some explaining to do) Alternatives to “Reactive” Accounts: - “Projective” Accounts of Turn-Taking o Speaker and listener recognize conversation has some goal(s) o Turn-taking occurs when listener must now contribute to move conversation towards goal(s) o This explanation: based on utterance meaning and intention, not sounds/ sentence structure/ gestures…. Projective Approaches – Evidence - Beattie and Barnard (1979): analysis of telephone conversations o Listener often takes over less than 200 ms after speaker finishes PSY274: Intro to Human Communication Lecture 5: October 15, 2013 o = Consistent with the idea that listener anticipates end of speaker‟s turn - Sometimes the goals of conversational segment are made clear from the outset can informs listener of the goal for a particular part of using pre=sequences o “I wanted to ask you something…” o “Did I tell you what happened on Sunday night?” o … gives advance information about what the speaker is about to contribute; makes it easier for the listener to
More Less

Related notes for PSY274H5

Log In


Don't have an account?

Join OneClass

Access over 10 million pages of study
documents for 1.3 million courses.

Sign up

Join to view


By registering, I agree to the Terms and Privacy Policies
Already have an account?
Just a few more details

So we can recommend you notes for your school.

Reset Password

Please enter below the email address you registered with and we will send you a link to reset your password.

Add your courses

Get notes from the top students in your class.