Constituent Structure - Middle East Technical University

Download Report

Transcript Constituent Structure - Middle East Technical University

Constituent Structure 1

Syntactic Categories  Parts of speech: Noun, Verb, Adjective, Adverb, etc.   Evidence for syntactic categories: child language. The Wug Test (Jean Berko Gleason, 1958) was designed to understand children’s understanding of inflection. 2

3

4

Tongue slips of adult native speakers (Spoonerisms) “Sir, you’ve hissed my mystery class”.

Intended: “Sir you’ve missed my history class.” 5

Ambiguity (Ambiguous sentences)   Lexical ambiguity John drove his car to the bank. The hunter went home with five bucks in his pocket. 6

Structural ambiguity   This type of ambiguity is caused by grouping words together in different ways. The [tall bishop]’s hat (The bishop is tall)  The tall [bishop’s hat]  (The hat is tall) 7

We can assign different grammatical structures to the same string of words.

This is evidence showing that words form sub groups (or CONSTITUENTS) within a phrase or sentence. These groupings are often crucial in determining the meaning of a sentence. 8

Words belonging to different syntactic categories a) b) Mistrust wounds.

“Suspicion hurts people.” “We should mistrust injuries.” Can you interpret the following sentence?

Time flies.

9

Amusing newpaper headlines Which word causes the ambiguity? 

Reagan Wins On Budget, But More Lies Ahead

Squad Helps Dog Bite Victim

10

How do we identify a constituent?

 She is crying.  The little girl wearing a red hat with a blue ribbon is crying. (1) Strings of words replacing a single word must be units (constituents.) 11

Malay   (i)

I Saya makan eat ikan besar itu fish big

‘I ate/am eating that big fish.’

that (ii)İkan besar itu saya makan fish big that I

‘That big fish I ate/am eating.’

eat

(2) When a group of words can be moved as a unit, we can assume that the group froms a syntactic unit. 12

Malay   Orang person tua old itu that makan ikan besar itu eat fish big that ‘That old person ate the big fish.’ ikan besar itu di-makan oleh anjing saya fish big that PASS-eat by dog my ‘That big fish was eaten by my dog.’ (3) The same string of words can occur in a variety of positions within the sentence, e.g. as subject and object. 13

Malay    Orang person ‘That old person ate the big fish.’ Siapa makan ikan besar itu Who tua old ate itu that fish ‘Who ate that big fish?’ makan ikan besar itu eat fish big big that that (4) When a group of words are replaced by a question word to form a content question, we can assume the group of words forms a unit. 14

Siapa makan ikan Who ate fish big ‘Who ate that big fish?’ Answer1: Orang tua person Answer2: *tua itu besar itu old old that itu that that ‘old that’ ‘that old person (5) Constituents can form the answer to a content question, whereas a string of words which is not a syntactic unit is not a possible answer. 15

Hierarchy Each constituent of a larger unit may itself be composed of smaller constituents. The CLAUSE is the smallest grammatical unit which can express a complete proposition.

16

A sentence may consist of several clauses.

Can you identify the clauses in the following lines?

Foxes have holes and birds of the air have nests, But the Son of Man has no place to lay his head.

17

PHRASE A single clause may contain several phrases.

The coach’s wife introduced her little sister to the captain of the football team.

[to [the captain [of [the [football team]]]]].

18

Football team Of the football team The captain of the football team To the captain of the football team 19

A single word may contain several morphemes.

Dis-taste-ful Read-abil-ity Dis-en-tangle

20

This kind of structural organization is called a PART-WHOLE HIERARCHY: Each unit is entirely composed of smaller units belonging to a limited set of types. This is important in morphology, syntax, and phonology. 21

Identifying syntactic categories Traditional definitions of parts of speech are based on semantic properties. A NOUN is a word than names a person, place, or thing.

A VERB is a word that names an action or event.

An ADJECTIVE is a word that describes a state. 22

   Traditional definitions fail to identify nouns like

happiness

,

love, destruction,

etc.

They cannot distinguish between the noun

love

and the adjective

fond of

.

They cannot distinguish the noun

fool

from the adjective

foolish

. 23

In Jabberwocky, we were able to distinguish most parts of speech even though they were mostly nonsense words. Also, children are able to form the plurals of nonsense words or words they’ve never heard before. 24

The identification of syntactic categories cannot be based on semantic factors.

We need to address the following problems separately:   Which words belong together in the same class?

What name (or label) should we assign to a given class? 25

Answering Question 1 Words that share a number of grammatical characteristics are assumed to belong to the same class. Words that have distinct grammatical characteristics are assumed to belont to differen classes.

26

Identifying grammatical characteristics

Fool vs foolish Modification by degree adverb vs adjective

    They are utter fools.

*They are utter foolish.

They are fools.

They are very foolish.  

Inflection for number

Fool, fools Foolish, *foolishes  

Comparative forms

Fool-*fooler/*more fool Foolish-more foolish

As subject of a clause

  Fools rush in where angels fear to tread.

*Fools rush in where angels fear to tread.

27

Answering Question 2 Once the word classes in a particular language have been identified in this way, they can be assigned a label (Noun, Verb, etc) based on universal notional patterns. If there is a class whose prototypical members include most of the basic terms for concrete objects (

dog, book,house

), we would label that class NOUN. 28

If there is a class whose prototypical members include most of the basic terms for volitional actions (

run, dance, eat

), we would label that class VERB. The grammatical criteria used to determine word classes are diagnostic features rather than definitions. E.g. In English, not all adjectives can take the comparative and superlative suffixes.

29

Almost all languages have the lexical categories Noun and Verb, but there is a significant range of difference among languages. 30

PHRASES and PHRASAL CATEGORIES  A phrase must be a group of words which form a constituent.

 A phrase is lower in the hierarchy than clauses. 1. Which phrases belong together in the same class?

2. What name (or label) should we assign to a given class?

31

Answering Question 1  Internal structure of phrases e.g. An English noun phrase often begins with a DETERMINER ( a, the, that, this )  Mutual substitutability: two phrases of the same category could potentially occur in the same positions. e.g. Phrases occuring in Object and Subject positions are NOUN PHRASES. 32

Answering Question 2   In most phrases, there is a core word, called the HEAD of the phrase. We name a phrase by the category of the head. e.g.

That big fish

noun (

fish)

. is a NOUN PHRASE because its head is a

e.g.very beautiful

is an ADJECTIVE PHRASE because its head is an adjective (

beautiful

).

33

How do we know which word in the phrase is the head? How do we distnguish the head from the DEPENDENTS (i.e all the other elements in the phrase)? 34

The head is important because  it determines the grammatical features of the phrase as a whole.  it may determine the number and type of other elements in the phrase.

 it is more likely to be obligatory than the modifiers or other non-head elements in the phrase.

35

The head determines the grammatical features of the phrase as a whole The new rice

is

in the barn.

The new kittens

are

in the barn. 36

The head may determine the number and type of other elements in the phrase.

 Prepositional phrases are complements of the adjective phrase I am [very grateful

to you

] John felt [sorry

for his actions

.] angry at someone, proud of someone, worried about something  Objects are complements of the verb phrase Mary is [reading

a book

]. James [showed

his photo

album

to us

]. Mary [runs] every morning. 37

The head is often obligatory in a phrase.  [The little girl wearing a red hat with a blue ribbon] was crying her eyes out.  [The little girl] was crying her eyes out.  [The girl] was crying her eyes out.

38

The head may be omitted in certain contexts  The third little girl was smarter than the second ___.  The good, the bad, and the ugly  The rich get richer and the poor get childen. 39

  Major categories (can function as heads of phrases) Noun, verb, adjective, adverb, preposition   Minor categories Conjunctions, interjections, determiners (includes articles, demonostratives, and quantifiers) 40

Tree diagrams representing the constituents of a clause In analyzing grammatical structure, we need to identify   The constituent parts which the sentence is formed.

The order in which these constitutents occur. The vertical lines inserted between the constitutents are helpful to describe grammatical structure. 41

Tree diagrams A Mother node B C Daughter nodes 42

 A DOMINATES all of its daughter nodes; i.e. The daughters of daughters, daughters of its grand-daughters, etc.  A mother IMMEDIATELY DOMINATES its own daughters. A CONSTITUENT is a string of words which is exhaustively dominated by some node. 43

PP P NP Det N on the beach 44

       N Noun A Adjective V Verb P Preposition Adv Adverb Det Determiner Conj Conjunction      NP Noun Phras AP ADjective Phrase VP Verb Phrase PP Prepositional Phrase S Sentence or Clause 45

  The top-most node in any tree diagram is called the ROOT NODE. The terminal nodes at the bottom are sometimes called LEAVES.   The No Crossing Constraint not cross.

: lines from mother to daughter must The Single Mother Constraint : each node after the root node must be the daughter of exactly one other node. 46

The motivation for imposing these constraints is that by allowing crossing lines or multiple parenthood, we would end up with potentially complex structures which are never found in real human languages. 47

Phrase Structure Rules The task of the linguist is to find out the rules which allow the speakers of a language to construct and comprehend novel sentences.

The rules needed to produce Phrase Structure Trees are known as Phrase Structure Rules and have the following form: A B C 48

A B C  This rule says that a node labelled A may immediately dominate two daughters labelled B and C in that order.  This is a CONTEXT FREE rule, i.e. there is no conditioning environment stated in this rule. 49

Each node of a Phrase Structure tree must be permitted (or LICENSED) by a phrase structure rule in order to be legal. To license (or, to generate) the prepositional phrase “on the beach” (slide 44), we would need these rules: 50

  PP P NP NP Det N  We also need rules to insert the terminal elements (lexical elements), i.e. to hang leaves on the tree.  P {on, in, at, under, over ...}  N {beach, house, boy, girl, cat ...}  Det {the, a, an, this, that, ...} 51

The LEXICON (the speaker’s mental dictionary)   The lexicon includes much more than a simple list of words.

The lexical entry for each word must include phonological, semantic, morphological, and syntactic information.  Instead of having lexical rules like the ones in the previous slide, we can simply assume that there is a general rule of LEXICAL INSERTION which will licence a word of any given category to appear as the only daughter of a node which bears the corresponding category label. 52

Lexical Insertion Rule Any lexical category (N, V, etc) may have a sinlge daughter node which is a specific lexical item of the same category. 53

Notational devices to combine two or more Phrase Structure Rules  a) A b) A c) A B (C) B B C  a) X Y Z b) X Y c) X Z 54

Pronouns and proper names In traditional grammar, pronouns and proper names are not considered as “phrases” in the sense we use them in linguistics.

  I collapsed.

John collapsed.

(pronoun) (proper name)  The old school collapsed. (noun phrase) 55

The subject of a clause may be expressed as a pronoun, a proper name, or a common noun phrase.

S pronoun proper name V noun phrase 56

The object of a preposition can be a pronoun, a proper name, or a common noun phrase.

behind me behind John behind the old school house PP P pronoun proper name noun phrase 57

Notice that the material inside the braces in PS rules in slides 56 and 57 are exactly the same.

The same set of alternatives may show up in other PS rules as well, i.e., in almost every position where a name can occur, we can substitute a pronoun or a common noun phrase.

58

If we had to list all of these alternatives in every rule that mentions one of these positions, there would be a large amount of redundancy in the rules. We would be missing an important generalization.

In order to avoid this massive redundancy, we will use the term NP to refer to any unit which can appear in a name-like position in the phrase structure. 59

Two New Phrase Structure Rules S NP V PP P NP  Traditional grammars state that a pronoun “takes the place of a noun”, but in fact pronouns replace whole NPs. 60

Pronouns are never modified by adjectives (but common nouns are)  The quick red [fox] brown dog.

jumped over the lazy   *The quick red [she] jumped over the lazy brown dog.

She jumped over him. 61

Proper nouns are not modified by determiners or adjectives either. Some unusal cases exist:  You are the first Emily I’ve ever met.   We will assume that pronouns and proper names are lexical items whose lexical entry specifies that they belong to category NP, rather than N. They may appear in tree diagrams as immediate daughters of an NP node.

62

 This is the end of the lecture on constituency. You can now do the exercises in Kroeger, pp. 47-50.

63