PSY 369 Psycholinguistics Language Comprehension Semantic networks Overview

PSY 369: Psycholinguistics Language Comprehension: Semantic networks

Overview of comprehension Input The cat chased the rat. Language perception c a t /k/ /ae/ /t/ Word recognition cat dog cap wolf tree yarn cat claw fur hat Syntactic Semantic & pragmatic analysis S NP VP the cat V NP chased the rat

Different approaches n Immediacy Principle: access the meaning/syntax of the word and fit it into a syntactic structure n n Serial Analysis (Modular): Build just one based on syntactic information and continue to try to add to it as long as this is still possible Interactive Analysis: Use multiple levels (both syntax and semantics) of information to build the “best” structure

Minimal attachment n Garden path sentences (Rayner & Frazier, 1983) MA S NP VP the spy NP VP S’ the spy V PP P the cop saw NP NP S’ PP P NP with the revolver V the cop NP saw Non-MA S but the cop didn’t see him The spy saw the cop with the binoculars. . The spy saw the cop with the revolver <- takes longer to read Conclusion: participants didn’t use semantic information initially, built the wrong structure and had to reanalyze. Supports a serial model.

Interactive Models n Other factors (e. g. , semantic context, co-occurrence of usage & expectation) may provide cues about the likely interpretation of a sentence (e. g. overriding purely syntactic principles like Minimal Attachment) n n n Trueswell et al (1994). Local semantic feature like Animacy The evidence (that was) examined by the lawyer … The defendant (that was) examined by the lawyer… Taraban & Mc. Celland (1988). Expectation n The couple admired the house with a friend but knew that it was over-priced. The couple admired the house with a garden but knew that it was over-priced.

What about spoken sentences? n All of the previous research focused on reading, what about parsing of speech? n Methodological limits – ear analog of eye-movements not well developed n n n Auditory moving window Reading while listening Looking at a scene while listening

Summing up n Is ambiguity resolution a problem in real life? n n Yes (Try to think of a sentence that isn’t partially ambiguous) Many factors might influence the process of making sense of a string of words. (e. g. syntax, semantics, context, intonation, cooccurrence of words, frequency of usage, …)

Semantics n Two levels of analysis (and two traditions of psycholinguistic research) n Word level (lexical semantics, chapter 11) n n What is meaning? How do words relate to meaning? How do we store and organize words? Sentence level (compositional semantics) (chapter 12) n n How do we construct higher order meaning? How do word meanings and syntax interact?

Separation of word and meaning n Words are not the same as meaning n Words are symbols linked to mental representations of meaning (concepts) n n Even if we changed the name of a rose, we would not change the concept of what a rose is Concepts and words are different things n n n Translation argument – we can translate words between languages (even if not every word meaning is represented by a single word) Imperfect mapping - Multiple meanings of words n e. g. , ball, bank, bear Elasticity of meaning - Meanings of words can change with context n e. g. , newspaper

Semantics n Meaning is more than just associations Write down the first word you think of in response to that word. CAT “Dog”, “mouse”, “hat”, “fur”, “meow”, “purr”, “pet”, “curious”, “lion” n You cannot just substitute these words into a sentence frame and have the same meaning. n n Frisky is my daughter’s ______. Sometimes you get a related meaning, other times something very different.

Semantics n Referential theory of meaning (Frege, 1892) n Sense (intension) and reference (extension) n n n “The world’s most famous athlete. ” “The athlete making the most endorsement income. ” 2 distinct senses, 1 reference Now n Over time the senses typically stay the same, while the references may change In the 90’s 2013 Bleacher report

Word and their meanings n Semantic Feature Lists n Decomposing words into smaller semantic attributes/primitives n Perhaps there is a set of necessary and sufficient features Features “father” “mother” “daughter” “son” Human + + Older + + - - Female - + + -

Word and their meanings n n Semantic Feature Lists “John is a bachelor. ” What does bachelor mean? n What if John: n n n is married? is divorced? has lived with the mother of his children for 10 years but they aren’t married? has lived with his partner Joe for 10 years? Suggests that there probably is no set of necessary and sufficient features that make up word meaning n (other classic examples “game” “chair”)

Meaning as Prototypes n Prototype theory: store feature information with abstract prototype (Eleanor Rosch, 1975) Rate on a scale of 1 to 7 if these are good examples of category: Furniture TV bed chair table refrigerator couch desk 1) chair 1) sofa 2) couch 3) table : : 12) desk 13) bed : : 42) TV 54) refrigerator

Meaning as Prototypes n Prototype theory: store feature information with abstract prototype (Eleanor Rosch, 1975) n Prototypes: n Some members of a category are better instances of the category than others (prototypicality effect) n n Fruit: apple vs. pomegranate What makes a prototype? n n n Possibly an abstraction of exemplars More central semantic features n What type of dog is a prototypical dog? n What are the features of it? We are faster at retrieving prototypical of a category than other less prototypical members of the category

Meaning as Prototypes n The main criticism of the model n The model fails to provide a rich enough representation of conceptual knowledge n n How can we think logically if our concepts are so vague? Why do we have concepts which incorporate objects which are clearly dissimilar, and exclude others which are apparently similar (e. g. mammals)? How do our concepts manage to be flexible and adaptive, if they are fixed to the similarity structure of the world? If each of us represents the prototype differently, how can we identify when we have the same concept, as opposed to two different concepts with the same label?

Meaning as Exemplars n Instance theory: each concept is represented as examples of previous experience (e. g. , Medin & Schaffer, 1978) n n Make comparisons to stored instances Typically have a probabilistic component n Which instance gets retrieved for comparison dog

Meaning as Theories n A development of the prototype idea to include more structure in the prototype (e. g. , Carey, 1985; Keil, 1986) n Concepts provide us with the means to understand our world n n n They are not just the labels for clusters of similar things They contain causal/explanatory structure, explaining why things are the way they are n n A lot of this work came out of concepts of natural kinds Similar to “scientific theories” They help us to predict and explain the world

Meaning as Networks n Semantic Networks n Words can be represented as an interconnected network of sense relations n Each word is a particular node n Connections among nodes represent semantic relationships

Collins and Quillian (1969) Animal Lexical entry Bird has skin can move around breathes A IS IS A has feathers can fly Fish has wings n Semantic Features has fins can swim has gills Collins and Quillian Hierarchical Network model n n Lexical entries stored in a hierarchy Representation permits cognitive economy n Reduce redundancy of semantic features

Collins and Quillian (1969) n Testing the model n Semantic verification task n An A is a B True/False An apple has teeth Use time on verification tasks to map out the structure of the lexicon.

Collins and Quillian (1969) has skin can move around breathes Animal Bird Robin has feathers can fly has wings n Testing the model Sentence Robins eat worms Robins have feathers Robins have skin Verification time 1310 msecs 1380 msecs 1470 msecs eats worms n has a red breast Participants do an intersection search

Collins and Quillian (1969) has skin can move around breathes Animal Bird Robin has feathers can fly has wings Robins eat worms n Testing the model Sentence Robins eat worms Robins have feathers Robins have skin Verification time 1310 msecs 1380 msecs 1470 msecs eats worms n has a red breast Participants do an intersection search

Collins and Quillian (1969) has skin can move around breathes Animal Bird Robin has feathers can fly has wings Robins have feathers n Testing the model Sentence Robins eat worms Robins have feathers Robins have skin Verification time 1310 msecs 1380 msecs 1470 msecs eats worms n has a red breast Participants do an intersection search

Collins and Quillian (1969) has skin can move around breathes Animal Bird Robin has feathers can fly has wings Robins have feathers n Testing the model Sentence Robins eat worms Robins have feathers Robins have skin Verification time 1310 msecs 1380 msecs 1470 msecs eats worms n has a red breast Participants do an intersection search

Collins and Quillian (1969) has skin can move around breathes Animal Bird Robin has feathers can fly has wings Robins have skin n Testing the model Sentence Robins eat worms Robins have feathers Robins have skin Verification time 1310 msecs 1380 msecs 1470 msecs eats worms n has a red breast Participants do an intersection search

Collins and Quillian (1969) has skin can move around breathes Animal Bird Robin has feathers can fly has wings Robins have skin n Testing the model Sentence Robins eat worms Robins have feathers Robins have skin Verification time 1310 msecs 1380 msecs 1470 msecs eats worms n has a red breast Participants do an intersection search

Collins and Quillian (1969) n Problems with the model n Difficulty representing some relationships n n How are “truth”, “justice”, and “law” related? Effect may be due to frequency of association (organization and conjoint frequency confounded) n “A robin breathes” is less frequent than “A robin eats worms” n Assumption that all lexical entries at the same level are equal n The Typicality Effect n n A whale is a fish vs. A horse is a fish Which is a more typical bird? Ostrich or Robin.

Collins and Quillian (1969) Animal Robin and Ostrich occupy the same relationship with bird. Bird Robin has feathers can fly has wings eats worms Ostrich has a red breast has skin can move around breathes Fish has fins can swim has gills has long legs is fast Verification times: can’t fly “a robin is a bird” faster than “an ostrich is a bird”

Collins and Quillian (1969) n Problems with the model Animal Bird has feathers can fly has wings n Smith, Shoben & Rips (1974) showed that there are hierarchies where more distant categories can be faster to categorize than closer ones n Chicken lays eggs clucks n A chicken is a bird was slower to verify than A chicken is an animal

Spreading Activation Models n Collins & Loftus (1975) street n vehicle car truck house orange blue Fire engine fire red n pear roses flowers n n apple tulips n bus Words represented in lexicon as a network of relationships Organization is a web of interconnected nodes in which connections can represent: fruit categorical relations degree of association typicality

Spreading Activation Models n Collins & Loftus (1975) street n n vehicle n car bus truck blue n house orange Fire engine fire red apple tulips pear roses flowers Retrieval of information fruit Spreading activation Limited amount of activation to spread Verification times depend on closeness of two concepts in a network

Spreading Activation Models n Advantages of Collins and Loftus model n n n Recognizes diversity of information in a semantic network Captures complexity of our semantic representation (at least some of it) Consistent with results from priming studies

Spreading Activation Models n More recent spreading activation models n n Probably the dominant class of models currently used Typically have multiple levels of representations

Meaning as networks n Today’s focus There may be multiple levels of representation, with different organizations at each level Meaning based representations Grammatical based representations Sound based representations

Meaning beyond the word n Not all meaning resides at the level of the individual words. n n n Conceptual combinations Sentences Move to compositional semantics

Conceptual combination n How do we combine words and concepts n We can use known concepts to create new ones n Noun-Noun combinations n n Modifier noun Head noun “Skunk squirrel” “Radiator box” “Helicopter flower”

Conceptual combination n How do we combine words and concepts n Relational combination n Property mapping combination n Relation given between head and modifier “squirrel box” a box that contains a squirrel Property of modifier attributed to head “skunk squirrel” a squirrel with a white stripe on its back Hybrid combinations n n A cross between the head and modifier “helicopter flower” a bird that has parts of helicopters and parts of flowers

Conceptual combination n How do we combine words and concepts n n n Instance theory has problems Modification? (brown apple) Separate Prototypes? (big wooden spoon) n n But sometimes the combination has a prototypical feature that is not typical of either noun individually (pet birds live in cages, but neither pets nor birds do) Extending salient characteristics? n n When nouns are “alignable” (zebra horse) But non-alignable nouns are combined using a different mechanism (zebra house)