Lecture 10 (Comprehension).pptx

Comprehension: Written and Spoken Language Lecture 10 Overview • We focus here on Miller’s fourth and fifth levels of language analysis-- the conceptual and belief levels. • Example: Ambiguous sentences, revisited: Mary and John saw the mountains while they were flying to California. Traditional...

Comprehension: Written and Spoken Language Lecture 10 Overview • We focus here on Miller’s fourth and fifth levels of language analysis-- the conceptual and belief levels. • Example: Ambiguous sentences, revisited: Mary and John saw the mountains while they were flying to California. Traditional Comprehension Research • Linguistic intuitions: Judgements about the acceptability of sentences. • Sachs (1967) • Jarvella (1970) People hold information about a sentence in memory until the sentence, or a meaningful clause, is completed. Limitations of Traditional Research • Raised more questions than it answered. • We need more precise methods of investigating comprehension. • Measures that allow us to study comprehension as it happens. On-Line Measures of Comprehension • Measure comprehension as it happens. • Written Language: A sentence appears on-screen, followed by a word. Subjects decide if the word was in the just-read sentence. • Spoken Language: The interruption technique. Monitoring. Comprehension and Mental Structure Building • Gernsbacher (1990) • Language comprehension is a process of building mental structures. • Involves: Laying a foundation, mapping information onto the structure, and shifting to a new structure. The Structure Building Framework Evidence for Structure Building • Advantage of First Mention: Ideas mentioned in the first sentence retain a special significance. • Advantage of Clause Recency: Where the most recently presented character shows an advantage. Definite Versus Indefinite Articles • “The” is a definite article. • “A,” “an,” and “some” are indefinite articles. • Gernsbacher predicts sentences with definite articles will be more coherent and sensible then sentences with indefinite articles. Sample Sentences Enhancement and Suppression • Enhancement: Where memory nodes are boosted or enhanced in their level of activation. • Suppression: Where activated nodes that become unrelated to the focus decrease in activation. Situation Models and Comprehension • A situation model is a representation of the real world situation described in a passage of text. • They include temporal and spatial information, information about the objects, locations and people mentioned in the stories, and the inferences we draw while comprehending the stories. Reference • Involves finding connections between elements in a sentence or text passage. • Dave studied hard for his statistics exam. • Antecedent: “Dave” • Anaphoric Reference: “His” • Reference: Linguistic process of alluding to a concept by using another name. Implication and Inference • Implication: An intended-- but not explicitly mentioned-- reference in a sentence. • Inference: The process by which the reader draws connections between concepts, determines referents and derives conclusions. Types of Reference and Implication (Clark, 1977) Bridging • Clark’s term for the process of constructing a connection between concepts. • Speakers and listeners must both build the same bridges. • Authorized: When the listener draws the inference intended by the speaker. • Unauthorized: A mistaken inference on the listener’s part. Multiple Processes in Drawing Inferences • Three steps: Retrieving related information from memory. Storing this information in memory so you can make connections / draw inferences. Integrating meanings -- making connections between antecedents and referents. The Extent of Drawing Inferences • How extensively do people draw inferences when comprehending? • Minimalist position: Only to the extent that the information cannot be retrieved automatically. • Does this position underestimate the extent of inference drawing during normal comprehension? Reading • Just and Carpenter (1980) • Gaze Duration Procedures: On line technique for studying reading comprehension. Uses an eye-tracker and camera to measure how long the eye’s dwell on each word. Gaze Duration Assumptions • Immediacy: Readers interpret each content word of a text as they encounter it in the passage. • Eye-Mind: The eye remains fixated on a word as long as the word is being actively processed during reading. Online Reading Effects • Regressive eye movements (back to a previouslyread portion of text). • Most content words are fixated on. • Saccades are shorter here than in scene perception. • Meaningless words (the, of) are rarely fixated on. Sample Eye Fixations The Just and Carpenter Model • Evidence at two levels of comprehension. Microscopic / Word level processes. Macroscopic processes such as comprehension time at the idea and proposition level. Just and Carpenter Model Spoken Language and Conversation • Studying Spoken Comprehension: • The auditory spoken window: Recorded passages are coded into units, and then played to subjects one unit at a time. Subjects press a button to advance to the next unit. Each button press is timed. The Structure of Conversations (Taking Turns) • Typically, there is little overlap between participants’ utterances. • Rules: 1) The current speaker gets to select the next speaker. 2) If Rule 1 is not used, anyone can become the next speaker. Social Roles and Settings • Formal settings among strangers lead to more structured, rule-governed conversations. • Superior’s have more leeway in breaking the turntaking rules. • Men more likely to interrupt women than the reverse. Conversational Rules • Grice (1975) • Are derived from the Cooperative Principle: Each participant in a conversation implicitly assumes that all speakers are following the rules and that each contribution to the conversation is sincere and appropriate. Grice’s Conversational Maxims Topic Maintenance • Process of making our contributions relevant and to the topic. • Schank (1977) Studied permissible and impermissible moves participants can make with regard to topic shifts. Online Theories During Conversation • Direct: Model of what our conversational partner knows and is interested in, what the partner is like. • Second-Order: An evaluation of the other participant’s direct theory -- what you think he/she believes about you. Empirical Effects in Conversation • Indirect Requests: When we ask someone to do something by an indirect and presumably more polite statement. • “Do you have the time,” • -Versus • “What time is it?” Face Saving (Holtgraves, 1998) • Nick’s class presentation either went well or went poorly. Nick asks you how he did? • Violations of the relevance maxim are acceptable if meant to save face (e.g., we change the topic rather than tell Nick how bad his presentation was). Egocentric Speech • Optimal Design Principle: Speakers design utterances so listeners can understand them. • Keysar (1998) Adults-- much like children-- often disregard this principle by speaking egocentrically (by failing to appreciate another person’s perspective). • Our first pass at an utterance is typically egocentric.