Textbook Notes (368,217)
Canada (161,710)
Psychology (1,418)
PSYC 340 (12)
Chapter 11

chapter 11.docx

14 Pages
Unlock Document

PSYC 340
Debra Ann Titone

Word Meaning (semantics: study of meaning) Why this is important: derivation of meaning is the ultimate goal of language processing 3 Reasons why phonological and orthological representations of words can be separated from their meanings: 1. Can translate words from one language to another even though not every word meaning is represented by a simple, single word in every language 2. Imperfect mapping between words and their meaning= ambiguity and some words have the same meaning, synonymy 3. Meaning of words often depends on context eg. Blushed red vs. Red apple KEY TERMS episodic memory n. A type of long-term memory for personal experiences and events, such as being stung by a bee many years ago, getting married a few months agoIt accounts for only a small proportion of human memory, most of our memories having no basis in personal experience. -1972 by the Estonian-born Canadian psychologist Endel Tulving (born 1927) semantic memory n. A type of long-term memory for factual information about the world, excluding personal episodes in one's life, typical examples being knowledge of the dates of the Second World War, the chemical formula for water - 1972 by the Estonian-born Canadian psychologist Endel Tulving (born 1927) lexicon: a person’s mental dictionary or the entire vocabulary of an individual Fun fact: English is believed to have the largest lexicon of any language: the Oxford English Dictionary (2nd ed.) defines more than 500,000 words. cognitive economy n. The tendency for cognitive processes to minimize processing effort and resources. For example, we do not squander memory capacity on storing the information has two legs for every individual we know; instead, we store this fact as a general default assumption for people in general and add a separate memory has one leg or has no legs only in exceptional cases in which the default does not apply. Cognitive economy refers to the fact that properties of concepts are stored at the highest possible level in the hierarchy and not re-represented at lower levels (see Figure 12.1 in text). This model did a relatively good job at predicting the time required to verify simple sentences, with a small set of assumptions: retrieving properties and traversing the hierarchy take time times for dependent steps are additive time to retrieve a property is independent of level of hierarchy Associations arise from words regularly occurring together VS Semantic relations arise from shared contexts and higher level relations Jackendoff (1983) 2 constraints on a general theory of semantics 1. Grammatical constraint: we should prefer a semantic theory that explains – arbitirary generalizations about syntax and the lexicon 2. Cognitive constraint – there is a level of representation where semantics must interface with other psychological representations ie. Perception Classic Approaches to Semantics  Connotation: a word’s secondary implications Vs. Denotation: word’s essential meaning or core (people agree on denotation but connotations differ) Referential theory of meaning: (point to it) Problems: 1. Abstract concepts (ie. Point to the justice) 2. Dissociation between a word and the things to which it can refer intension The properties that define a word or concept. An intensional definition of a class will correspondingly specify the properties that something must have to be a member of it extension 1. In the ordinary sense. Thus analogical extension is an extension in the range of words formed in a particular way, seen as by analogy with those already so formed. ‘Extension of meaning’ is the development of a new sense of a lexical unit: thus, in particular, a figurative extension which involves a metaphor or other figure of speech. It is also used of the widening of an existing sense: cf. widening of meaning. Semantic Networks Collins and Quillian (1969) A Hierarchical Semantic Network  suggested that semantic networks  have a hierarchical organization, with different levels representing concepts ranging from the most  abstract (broad) to down to the most concrete (specific).     stored within a hierarchical structure  properties stored together with a concept following the principle of cognitive economy (see above)  Tested by sentence verification task Drawbacks to Collins and Quillian (1969)  Performance on false sentences: There are several methods that a model such as that of Collins and Quillian might use to predict the time required to indicate that a sentence is false unfortunately, none of these methods corresponded with how humans did the task  Not all information is easily represented in hierarchical form ie. Truth, justice, law  Confounds semantic distance travelled with conjoint frequency (or how frequently words co- occur)  Makes some incorrect predictions ie. “A cow is an animal” is verified faster than “a cow is a mammal”  Relatedness effect: we do not reject all untrue statements equally slowly, they closer they are related- the harder to disentangle  Prototypicality effect; advantage for more typical items ie. Robin trumps penguin, apple trumps kiwi  We list things that shouldn’t be included for all in a concept ie Birds fly Collins & Loftus’ (1975) Semantic Network Model Major Differences:  Many Connections (No Cognitive Economy)  Spreading Activation Replaces Search  Structure is more complex: links between nodes vary in strength and distance  Very difficult to test, hard to see what sort of experiments could falsify Decompositional Theories Meaning of word determined by its decomposition into smaller units Semantic primitives: atoms of meaning or concepts which can be understood without any definitions, and in terms of which all other concepts can be defined, are called semantic primitives (or primes) Katz­Fodor Semantic Theory (1963)    Decompositional theory of meaning- meaning of individual words broken down into semantic markers  Combination is governed by selection restrictions (ie. Selection restriction on kick – needs and animate subject (kicker) and may include optional object ie. Ball)  assumed a semantic component that assigns semantic representations to lexical items and, by means of recursive 'projection rules', to phrases and sentences  Semantic representations are concepts which are built up from semantic primitives Criticism of Katz and Fodor decomposition: • b/c you cannot become an unmarried man who can never be married > is impossible to become part of the semantic features of this sense of bachelor? Seems very implausible Sentence Verification Test Example: “This is a test of speed of comprehension. You see sentences on the screen, one at a time, for example Bicycles have wheels Tomatoes have wings If the sentence is true, you press YES, if not true, you press NO. This is a reaction time task, so you should always respond as quickly as possible, while avoiding errors. We record the reaction times averaged over the test, and the total number of errors.” Rips Feature List Model (1973) Assumption: the meaning of any word or concept consists of a set of elements called features. Features come in two types: a) Defining, meaning that the feature must be present in every example of the concept b) Characteristic, meaning the feature is usually, but not necessarily present.  Very closely tied to sentence verification paradigm  Many words do not have obvious defining features Smith and Medin’s Probabilistic Feature Model (1981)  assumes that concepts are abstractions, or summary representations  argues that for a property to be included in the summary it need have only a substantial probability of occurring in instances of the concept, i.e. it need only be characteristic of the concept, not defining. (An object will then be categorized as an instance of some concept A if, for example, it possesses some criterial number of properties, or sum of weighted properties, included in the summary representation of A.  Categorization =a matter of assessing similarity rather than of applying a definition Advantages:  Emphasizes the relation between meaning and identification  Can account for verification time data  Categories that have unclear boundaries= no longer problematic Is semantic decomposition obligatory?  Ie. When we see new see a word like bachelor is the retrieval of all of its features an automatic process? Evidence against: - PDNs (pure definitional negatives) two implied negatives should take longer to process than one but they don’t - Lexical causatives (verbs that cause a new state of affairs) should reflect deep structure differences but no evidence that participants decompose lexical causatives Evidence in favour: Memory tasks- people with aphasia are better @ retrieving verbs with rich semantic representation compared with vers with less rich Ie. Hurry is richer than go because it contains go  UMMARY :LIKELY  THAT  WE  REPRESENT  THE  MEANINGS OF  WORDS AS COMBINATIONS  OF SEMANTIC  FEATURES      Prototype Theories  Prototype: an average family member (Rosch, 1978) “best example of a concept” Eg. Robin>bird, apple>fruit (special type of schema) √ prototype theory would consider a category like bird as consisting of different elements which have unequal status (instead of a definition based model - e.g. a bird may be defined as elements with the features [+feathers], [+beak] and [+ability to fly],) √ leads to a graded notion of categories Basic Level: the level of representation in a hierarchy that is the default level (eg. Dog rather than terrier or   animal)  - Thus, when asked What are you sitting on? most subjects prefer to say chair rather than a subordinate such
More Less

Related notes for PSYC 340

Log In


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.