CS 712: Topics in NLP Linguistic Phrases and Statistical Phrases Pushpak Bhattacharyya, CSE Department, IIT Bombay 18 March, 2013 (main text: Syntax by Adrew Carnie, Blackwell Publication, 2002)
Domination Essentially the specification of a tree (very familiar to computer scientists!) Axioms of domination (x <= D y means x dominates y) (a) X <= D X (b) if X <= D Y <= D Z then X <= D Z (c) if X <= D Y <= D X then X =Y (d) if X <= D Z and Y <= D Z then either X <= D Y or Y <= D X (or both if X=Y=Z)
Immediate Domination and Exhaustive Domination Immediate Domination: Direct Parent Child relation Exhaustive Domination: Node A exhaustively dominates a set of nodes {B, C,, D}, if it immediately dominates all the members of the set and and there is no node G immediately dominated by A that is not a member of this set.
Constituency Constituent: A set of nodes exhaustively dominated by a single node Constituent-of: B is a constituent of A iff A dominates B Immediate-constituent-of: B is an immediate-constituent-of A iff A immediately dominates B
Precedence ( said first relation) S NP VP S dominates NP and VP {NP VP} forms a constituent But NP precedes VP Definition: Node A precedes node B iff A is to the left of B and neither A dominates B nor B dominates A and every node dominating A either appears to the left of B or dominates B
No crossing branches constraint If one node X precedes another node Y then X and all nodes dominated by X must precede Y and all nodes dominated by Y
Axioms of Precedence Lets denote precedes by the symbol ~ (a) If X ~ Y then NOT(Y ~ X) (b) If X~Y~Z then X~Z (c) If X~Y or Y~X then NOT(X <= D Y) and NOT(Y <= D X) (d) X~Y iff for all terminals U, V, X <= D U and Y <= D V jointly imply U~V No crossing of branch; no discontinuous constituent
Exercise D 1 NP 1 O The N AP 1 boy little S T will V buy NP 2 N 2 apples VP P from 1.What nodes dominate grocer 2. What nodes immediately dominate D3 the 3. Do will and buy form a constituent? 4. What nodes does N 1 boy c-command? 5. What nodes does NP1 c-command? 6. What is V s mother? 7. What nodes does will precede? 8. List all the sets of sisters in the tree. 9. What is s mother? 10. Do NP1 and VP symmetrically or assymetrically c-command one another? 11. List all the nodes c-commanded by V NP 3 D 3 the N 3 grocer 12. What is the subject of the sentence? 13. What is the object of the sentence? 14. What is the object of the preposition? 15. Is NP 3 a constituent of VP? 16. What node(s) in NP 3 an immediate constituent of? 17. What node(s) does VP exhaustively dominate? 18. What is the root node? 19. List all the terminal nodes. 20. What immediately precedes grocer?
X-bar Theory
How deep should a tree be? Is there a principle in branching When should the constituent give rise to children? What is the hierarchy building principle?
Deeper trees needed for capturing sentence structure NP This wont do! Flat structure! The AP book with the blue cover big of poems [The big book of poems with the Blue cover] is on the table.
Other languages NP English The AP book with the blue cover big NP of poems AP Hindi kitaab niil jilda vaalii kavita kii badii [niil jilda vaalii kavita kii kitaab]
Other languages: contd NP English The AP book with the blue cover big NP of poems AP Bengali niil malaat deovaa kavitar motaa ti bai [niil malaat deovaa kavitar bai ti]
s are at the same level: flat with respect to the head word book NP No distinction in terms of dominance or c-command The AP book with the blue cover big of poems [The big book of poems with the Blue cover] is on the table.
Constituency test of Replacement runs into problems One-replacement: I bought the big [book of poems with the blue cover] not the small [one] One-replacement targets book of poems with the blue cover Another one-replacement: I bought the big [book of poems] with the blue cover not the small [one] with the red cover One-replacement targets book of poems
More deeply embedded structure NP N 1 The AP N 2 big N 3 N book with the blue cover of poems
To target N 1 I want [ NP this [ N big book of poems with the red cover] and not [ N that [ N one]]
Bar-level projections Add intermediate structures NP (D) N N (AP) N N () N () () indicates optionality
New rules produce this tree NP N-bar N 1 The AP N 2 big N 3 N book with the blue cover of poems
As opposed to this tree The AP NP book with the blue cover big of poems
V-bar What is the element in verbs corresponding to one-replacement for nouns do-so or did-so
As opposed to this tree The AP NP book with the blue cover big of poems
I [eat beans with a fork] VP eat NP with a fork beans No constituent that groups together V and NP and excludes
Need for intermediate constituents I [eat beans] with a fork but Ram [does so] with a spoon VP V 1 VP V V V () V V (NP) V 2 V NP with a fork eat beans
How to target V 1 I [eat beans with a fork], and Ram [does so] too. VP V 1 VP V V V () V V (NP) V 2 V NP with a fork eat beans
Case of conjunction VP V 1 V 2 V eat V 3 NP Conj and V drink V 4 NP In the afternoon beans coffee
A-bar: adjectives AP A 1 AP A A (AP) A A A () AP A 2 Very AP A 3 A 4 Conj and A 5 AP A 6 bright blue dull green
So-replacement for adjectives Ram is very serious about studies, but less so than Shyam
P-bar: prepositions P 1 P P P () P P (NP) AP P 1 right P off P 2 NP Conj and P 3 P into NP the table the trash
So-replacement for Prepositions Ram is utterly in debt, but Shyam is only partly so.
Complements and Adjuncts or Arguments and Adjuncts
Rules in bar notation: Noun NP (D) N N (AP) N N N () N N ()
Rules in bar notation: Verb VP V V V () V V (NP)
Rules in bar notation: Adjective AP A A (AP) A A A ()
P P P () P P (NP) Rules in bar notation: Preposition
Introducing the X factor Let X stand for any category N, V, A, P Let XP stand for NP, VP, AP and Let X stand for N, V, A and P
XP to X Collect the first level rules NP (D) N VP V AP A P And produce XP (YP) X
X to X Collect the 2 nd level rules N (AP) N or N () V V () A (AP) A P P () And produce X (ZP) X or X (ZP)
X to X Collect the 3 rd level rules N N () V V (NP) A A () P P (NP) And produce X X (WP)
Basic observations about X and X X (WP) X X (ZP) X is called Head X Phrases must have Heads: Headedness property Category of XP and X must match: Endocentricity
Basic observations about X and X X (WP) X X (ZP) X Sisters of X are complements correspond to arguments Sisters of X are Adjuncts s and Adjectives are typical adjuncts We have adjunct rules and complement rules
Structural difference between complements and adjuncts XP X X ZP X WP Adjunct Complement
Complements and Adjuncts in NPs NP N N ZP book N with red cover of poems
NP Any number of Adjuncts N N N ZP from Oxford Press N with red cover book of poems
Digression
Kinds of Verbs Verb finite Non-finite Participle infinitive -ed form -ing form He walked. When contacted, he acceded to the request. Seeing the car, he moved. To see is to believe
Verb types are universal English Hindi Bengali He walked. वह चल ও হ ট ল When contacted, he acceded to the request. स पक कए ज न पर उ ह न अन र ध व क र कर लय ওন র স সmক ন লন কর ত উ ন ম ন Seeing the car, he moved. ग ड़ द खत ह वह म ड़ गय গ ড়ট দ খই ও স র গ ল To see is to believe. द खन ह व व स करन ह দখ ই ব স কর
More xbar trees
S NP N N VP V V AP Ram A reads N A N frequently booksof poems
S NP N VP V N V N Ram bought N P red A AP N A N Shirts P with NP pink stripes
S NP N VP V N V NP Ram bought N N AP Shirts red With pink stripes
S NP N VP V N V Ram V with pink stripes bought red Shirts
Ram read books of poems frequently
Ram bought red shirts with pink stripes
Ram bought red shirts with pink stripes (another parse)
Ram bought red shirts with pink stripes (yet another parse)
I want him to swim and I want that he swims (normal parse) S NP VP I want him to swim S NP VP I want that he swims
I want him to swim (xbar parse) S NP VP I V V XP want him to swim
I want that he swims (xbar parse) S NP VP I V V XP want that he swims