400 likes | 419 Views
Language, Mind, and Brain by Ewa Dabrowska. Chapter 10: The cognitive enterprise. 1. Towards a psychologically realistic grammar. Q: What do we know about language processing?. 1. Towards a psychologically realistic grammar. Q: What do we know about language processing?
E N D
Language, Mind, and Brainby Ewa Dabrowska Chapter 10: The cognitive enterprise
1. Towards a psychologically realistic grammar • Q: What do we know about language processing?
1. Towards a psychologically realistic grammar • Q: What do we know about language processing? A: It is flexible, fast, robust, and relies on low-tech, general-purpose mechanisms, including ability to perceive, categorize, and store information.
1. Towards a psychologically realistic grammar • Q: How do we account for these facts?
1. Towards a psychologically realistic grammar • Q: How do we account for these facts? • A: • fast – because it relies on chunks • flexible and robust – because of redundancy • intimate relationship between grammar and lexicon • These facts should constrain our linguistic theory, and the best choice is: Cognitive Grammar.
2.1 Linguistic expressions are symbolic units • Q: What’s a “symbolic unit”?
2.1 Linguistic expressions are symbolic units • Q: What’s a “symbolic unit”? • A: A pairing of a phonological form and a meaning. A unit can be simple or complex (varying levels of abstractness, partly or wholly schematic). + Plural = kæt + s = kæts
2.2 Imagery • Q: What is a base and what is a profile?
2.2 Imagery • Q: What is a base and what is a profile? • A: A base is a knowledge structure (often referred to as “extralinguistic knowledge”) within which an expression is understood. A profile is a substructure designated for a concept within the base.
2.2 Imagery • Q: What is a landmark and what is a trajector?
2.2 Imagery • Q: What is a landmark and what is a trajector? • A: A landmark is the ground against which a trajector stands out as the figure, or most salient entity. ‘I put my lipstick in my handbag.’
2.2 Imagery • Q: What factor decides which element will be the trajector and which one the landmark?
2.2 Imagery • Q: What factor decides which element will be the trajector and which one the landmark? • A: Construal! There are some strong trends (animate or moving or most salient entity), but speakers can manipulate this distinction.
2.2 Imagery • Q: What do we know about categorizing relationships?
2.2 Imagery • Q: What do we know about categorizing relationships? • A: We can understand the same things at different levels of abstractness, both at the level of the superordinate schema and the subordinate instantiation, and this is true for all kinds of linguistic units (phonological, lexical, syntactic)
2.3 Things, processes and atemporal relations • CG can provide “ a semantically based account of grammatical distinctions which in other theories are dealt with by means of arbitrary syntactic features”. • Thing: “a region (set of interconnected entities) in some domain”, where are domain can be, for example time (week), color (blue), etc. • Process: a temporal relation (profiled by a verb) • Atemporal relations: profiled by prepositions adjectives, adverbs
2.4 Constructional schemas • Q: What is a constructional schema?
2.4 Constructional schemas • Q: What is a constructional schema? • A: A symbolic unit which is both complex and schematic. Constructional schemas capture the types of relationships described by rules in other theories.
2.4 Constructional schemas • Q: How does the CG account of syntax differ from that in other theories?
2.4 Constructional schemas • Q: How does the CG account of syntax differ from that in other theories? • A: CG represents lexical and syntactic information in the same way, rather than as separate different functions. Lexicon and syntax differ only in their degree of specificity vs. schematicity.
2.5 Language as a structured inventory of conventional linguistic units • “In the CG framework, linguistic knowledge is seen as a complex, richly interconnected network (structured inventory) of linguistic units.” • Q: What three types of units are there?
2.5 Language as a structured inventory of conventional linguistic units • “In the CG framework, linguistic knowledge is seen as a complex, richly interconnected network (structured inventory) of linguistic units.” • Q: What three types of units are there? • A: Phonological, semantic, and symbolic.
2.5 Language as a structured inventory of conventional linguistic units • Q: What are the three types of relationships between the three types of units?
2.5 Language as a structured inventory of conventional linguistic units • Q: What are the three types of relationships between the three types of units? • A: • Symbolization: the relationship between semantic structure and its phonological form • Composition: relationships between component units and the composite unit they form • Schematicity: the relationship between a superordinate schema and a subordinate instantiation (Note that both composition and schematicity can hold for both semantic and phonological units.)
2.5 Language as a structured inventory of conventional linguistic units • Q: What is partial schematicity (extension)?
2.5 Language as a structured inventory of conventional linguistic units • Q: What is partial schematicity (extension)? • A: When the specific content (phonological or semantic) does not fully match the schema. (E.g. the polysemy of fly, as in Her hair was flying in the wind; variant pronunciations)
2.6 A usage-based model • Phonological, semantic, and symbolic units are generalizations over actual usage events. Every time a particular unit is used, it becomes more entrenched. A person’s mental grammar is a dynamic system constantly shaped by experience with language.
2.6 A usage-based model • CG: • Is maximalist: linguistic knowledge is represented at various levels of abstraction • Emphasizes low-level schemas, for they are necessary to capture actual patterns of usage • Higher-level generalizations may exist, but they have “an organizing function rather than an active computational one”.
2.7 Meeting the specifications • CG’s conception of linguistic knowledge as a network of symbolic units of varying size and schematicity accommodates chunks. • CG is flexible and accounts for robustness of language, also allows for variation in individual mental grammars. • CG’s sensitivity to entrenchment accounts for frequency effects. • CG does not rely on any language-specific biological adaptions. • CG treats lexical and grammatical information similarly
3. Language production in a CG framework • Linguistic communication involves finding the relevant parts of form-meaning pairing. • New expressions can be formed by: • Juxtaposition (come here + now) • Superimposition: a filler elaborates a subpart (slot) of a frame (want + NP, NP = my desk)
3. Language production in a CG framework • Q: When constructing a novel expression, what types of units do speakers prefer to use?
3. Language production in a CG framework • Q: When constructing a novel expression, what types of units do speakers prefer to use? • A: The most specific units that are most concrete and well-entrenched. Thus overgeneralization is avoided.
4. A cognitive view of language acquisition • Q: What do we know about the input children receive?
4. A cognitive view of language acquisition • Q: What do we know about the input children receive? • A: It is mostly multi-word utterances in a rich and fairly predictable context with clues about meaning, and most utterances are stereotypical (frequently repeated in similar contexts).
4. A cognitive view of language acquisition • Child hears: You want milk? Do you want milk? D’you want milk? • Partial understanding of whole phrase • Memory is content-accessible (content is both phonological and semantic) • Components get entrenched, pattern gets analyzed.
4. A cognitive view of language acquisition • Some overall features of this system: • “The capacity for grammatical productivity emerges gradually as result of rote-learned phrases and is a by-product of the way symbolic units are stored in long-term memory.” • Relational words and near-synonyms are learned in context emerge from their context
5. More on regularity • Q: How regular are the rules that are posited by linguists?
5. More on regularity • Q: How regular are the rules that are posited by linguists? • A: No rules are completely regular, and rules tend to overemphasize regularity. Most patterns show some lexical specificity. Note that languages show tendencies both for generalization and for emergence of irregular structures.
6. Future directions • Some future challenges for CG: • Account for all syntactic phenomena • More work on phonology • Empirical studies • Language acquisition and processing