Introduction to LFG Kersti Brjars Nigel Vincent k
Introduction to LFG Kersti Börjars & Nigel Vincent {k. borjars, n. vincent}@man. ac. uk University of Manchester Winter school in LFG July 4 -8 2004 University of Canterbury Christchurch, New Zealand
Part 1 Basic building blocks and c-structure to f-structure mapping Kersti Börjars In order to make sense of this, you will need to run the presentation, it relies on animations and a printed version will not be very clear. Feel free to use it if you find it helpful, but please give source.
Types of information about linguistic units (1) The dog ate the rats. • Syntactic structure, c-structure the dog forms a constituent in (1) • Functional information, f-structure the dog is the subject in (1) is in past tense • Argument structure, a-structure eat takes two arguments • Information structure • Phonetic structure • …
Correspondence between dimensions (1) The dog ate the rats. Function subject object dog rats agent theme Semantic role (2) The rats were eaten by the dog. Function subject object dog rats
Non-one-to-one correspondence Parallel correspondence approach a-structure c-structure mapping relations f-structure
a-structure Information relating to thematic roles associated with a predicate (3) a. tickle < agent, patient> b. like < experiencer, theme> Arguments are represented as features. Lexical Mapping Theory maps the arguments onto fstructure.
f-structure: attributes A feature value matrix: an unordered set of feature-value pair Types of attributes: Functional features NUM PERS TENSE … Semantic feature PRED Grammatical relations SUBJ OBJ ADJUNCT COMP …
f-structure: values Types of values: Atomic value Value of the functional features: plu, past, 3, fem Semantic form value of PRED f-structure value of grammatical relations f-structures are reasonably invariant across languages
f-structure: examples Semantic feature she Sw Functional hon features Semantic feature The goal keeper smiled. Sw. Grammatical Målvaktenrelation log. Functional feature
c-structure Category labelled trees Categories Lexical S, N, V, P, A, (Adv) Functional C, I, D Cross linguistic variation Both endocentric (headed) and exocentric (non-headed) structures allowed
c-structure: functional categories Functional categories are used when certain functional features are associated with positional properties Warlpiri: (6) a. wita-jarra-rlu ka-pala wajili-pi-nyi small-DUAL-ERG PRES-3 DUSUBJ chase-NPAST yalumpu kurdu-jarra-rlu maliki that. ABS child-DUAL-ERG dog. ABS b. All permutations possible as long as inflectional bundle stays in second position c. IP I’ I ka-pala
c-structure: functional categories English: (7) a. The rats will eat the dog. (8) b. Will the rats eat the dog? (8) a. The rats ate the dog. (9) b. *Ate the rats the dog? Special status of auxiliary verbs: (9) IP I’ NP I will VP
c-structure: an interlude Principle of Economy of Expression: all phrase structure nodes are optional unless required by independent principles (completeness, coherence, semantic expressivity) (11) (10) The rats ate the dog. IP NP The rats I’ I VP VP V NP ate the dog
c-structure: “head to head movement” Xn C-structure heads are f-structure heads: Complements of functional categories are f-structure co-heads: X F’ XP
c-structure Latin: (10) a. Canis rattos devoravit. dog. NOM rat. ACC. PL eat. PERF. 3 SG b. all orders in possible under right information structural conditions Functional information on verb not associated with position, so no argument for a functional category. (11) S NP NP S V NP V Etc. NP
Mapping between f-structure and c-structure Position: Spec-IP IP NP I’ The dog VP V NP ate the rats Case: nominative Position: sister of V S NP NP Case: accusative V rattos canis devoravit S NP NP V canis rattos devoravit
Mapping between f-structure and c-structure (12) IP f 1 NP f 2 D f 4 The I’ f 3 VP f 6 N f 5 dog V f 7 ate NP f 8 D f 9 the N f 10 rats
Basic lexical entries the dog ate rats
Mapping between f-structure and c-structure (12) IP f 1 SUBJ = f 2 f 1 = f 3 NP f 2 D f 4 The I’ f 3 VP f 6 N f 5 dog f 4 PERS = 3 f 3 = f 6 f 5 PRED = ‘dog’ f 6 = f 7 f 5 NUM = sg f 6 OBJ = f 8 f 7 PRED = ‘eat <SUBJ OBJ>’ f 2 = f 4 V f 7 ate NP f 8 D f 9 the f 4 DEF = + N f 10 rats f 2 = f 5 f 8 = f 9 f 8 = f 10 f 7 TENSE = past f 9 DEF = + f 9 PERS = 3 f 10 PRED = ‘rat’ f 10 NUM = pl
Mapping between f-structure and c-structure f 1 SUBJ = f 2 f 4 f 5 f 4 DEF = + f 1 = f 3 f 5 PRED = ‘dog’ f 2 = f 4 f 5 NUM = sg f 2 = f 5 f 7 PRED = ‘eat <SUBJ OBJ>’ f 3 = f 6 f 7 TENSE = past f 6 = f 7 f 9 DEF = + f 6 OBJ = f 8 f 10 PRED = ‘rat’ f 8 = f 9 f 10 NUM = pl f 8 f 9 f 10 f 1 f 3 f 6 f 7 f 9 = f 10
Well-formedness conditions on f-structure Completeness condition: (i) all argument functions specified in the value of the PRED feature must be present in the local f-structure, (ii) all arguments so specified must have a PRED value. Coherence condition: (i) all argument functions in an fstructure must be specified by their local PRED, (ii) all functions which have a PRED value must be arguments of the element which specifies them. Functional uniqueness: Any attribute has only one value
Well-formedness conditions on f-structure (14) a. *Oscar tickled. Incomplete b. *Oscar tickled Sarah the dog’s tummy. Incoherent c. *The sky rained. Incoherent
Well-formedness conditions on f-structure (14) d. *Tickles her. e. tickles f. Coherent? Yes Complete? No, cf (ii), no PRED value for SUBJ
Well-formedness conditions on f-structure (14) g. *They tickles her. h. tickles i. they j. f. Functional uniqueness violated
c-structure: back to Latin (10) a. Canis rattos devoravit. dog. NOM rat. ACC. PL eat. PERF. 3 SG b. all orders in possible under right information structural conditions (11) S NP NP S V NP V Morphology builds f-structure: (15) a. b. Etc. NP
c-structure: back to Latin S (16) NP NP V canis rattos devoravit (15) a. b.
More on functions Grammatical relations SUBJ OBJ ADJUNCT COMP … Semantic feature PRED Functional features NUM PERS TENSE …
More on functions Non-argument TOP FOC Discourse function SUBJ Core OBJ Argument Non-core Non-argument OBL COMP ADJUNCT Non-discourse function
Discourse functions (17) a. Beans, I like. b. FOCUS In Spec-CP Functional uncertainty c. Incomplete? Complete
COMP function (18) a. Oscar claimed that he liked LFG. b. claim c. VP V claimed CP that he liked LFG
OBJ and OBL functions Both OBJ and OBL are argument functions which can occur in the PRED feature of a verb. In English, an OBL is always a PP, whereas an OBJ is an NP. A function subscripted with a is restricted to a certain thematic role. (19) a. Oscar gave Sarah a book. b. Oscar gave a book to Sarah. In (19 a), a book is OBJ restricted to the role of theme, hence it is an OBJTHEME. The PP to Sarah in (19 b), is restricted to having a recipient role, hence it is an OBLRECIP.
OBJ function (19) The teacher bought Sarah a book. Sarah is OBJ a book is OBJ • Provide the lexical entry for bought and the other words; • Provide the phrase structure rule that is required (remember that it does not have to be binary branching); • Add annotations to the phrase structure rule; • Draw the tree using the new rule and the ones we have used in the class; • Add numbered f-structure variables to each node (f 1, f 2, etc); • Write down the equations and solve them to give you the correct f-structure.
Reading The main reading for this part comes from: Falk, Yehuda 2001. Lexical-Functional Grammar. An introduction to parallel constraint-based syntax. Stanford, Ca: CSLI Publications. Chapters 1 -3 If you want to have some additional reading, try: Bresnan, Joan (2001) Lexical Functional Syntax. Oxford: Blackwell. Chapters 1 -4 Dalrymple, Mary (2001) Lexical Functional Grammar. [Syntax and Semantics 34]. New York: Academic Press. Chapters 1 -5.
- Slides: 33