1 / 28

Analyzing Features of Grammatical Categories

Analyzing Features of Grammatical Categories. Show my head to the people; it is worth seeing. --Feature structure, to Ivan Sag in a dream. CFG’s: Chock Full of Goofs. Requires massively redundant rules Fails to capture generalizations S →NP-3p-sg VP-3p-sg S →NP-3p-pl VP-3p-pl S →NP1 VP1

druce
Download Presentation

Analyzing Features of Grammatical Categories

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Analyzing Features of Grammatical Categories Show my head to the people; it is worth seeing. --Feature structure, to Ivan Sag in a dream

  2. CFG’s: Chock Full of Goofs • Requires massively redundant rules • Fails to capture generalizations • S→NP-3p-sg VP-3p-sg • S→NP-3p-pl VP-3p-pl • S→NP1 VP1 • S→NP2 VP2 • Rules are arbitrary • S→NP1 VP2

  3. HPSG: How to Phix Subpar cfGs • Change atomic categories into categories that can be decomposed into features.

  4. type feature value university NAME Stanford Univ. FOUNDERS PRESIDENT TEL feature structure Let’s get our terms straight

  5. More complex structures • Embedding, see p. 54, fig.7a • Indexes Department TEL [1] 650-723-4284 CHAIR [TEL [1] ]

  6. HPSG Types and Features • Initial type hierarchy, p. 61, fig. 25

  7. HPSG Types and Features • Initial type hierarchy, p. 61, fig. 25 • First important subtypes divide expressions into words and phrases

  8. HPSG Types and Features • Initial type hierarchy, p. 61, fig. 25 • First important subtypes divide expressions into words and phrases • All expressions have the feature HEAD

  9. HPSG Types and Features • Initial type hierarchy, p. 61, fig. 25 • First important subtypes divide expressions into words and phrases • All expressions have the feature HEAD with a value from the pos types

  10. HPSG Types and Features • Initial type hierarchy, p. 61, fig. 25 • First important subtypes divide expressions into words and phrases • All expressions have the feature HEAD with a value from the pos types

  11. Agreement • Some pos types exhibit agreement, so we group them into their own subtype agr-pos, whichcarries the feature AGR • AGR feature takes the value agr-cat, which is a sub-type of feature-structure • It has (at least) the features PER and NUM • PER takes the values {1st,2nd,3rd} • NUM takes the values {sg, pl} • See p. 70, fig. 49

  12. How much of this is universal? Other languages have different values? All languages have all features and values, but optimality-like constraints ensure that only some show up?

  13. The VAL feature • A feature of expressions

  14. The VAL feature • A feature of expressions • Takes the value val-cat, which has the features COMPS and SPR

  15. The VAL feature • A feature of expressions • Takes the value val-cat, which has the features COMPS and SPR • These “represent the combinatoric potential of the word or phrase”

  16. The VAL feature • A feature of expressions • Takes the value val-cat, which has the features COMPS and SPR • These “represent the combinatoric potential of the word or phrase” • COMPS takes the value itr, str, or dtr; p.62, fig. 27 • All expressions have this feature (so far)

  17. Underspecification • A type can be unspecified for a particular feature • This picks out a larger class of feature structures (it includes more kinds of feature structures) • Underspecification allows some kinds of generalizations that we couldn’t get from CFGs, p. 63, fig.28

  18. The SPR feature • A generalization of the notion ‘determiner’ • Distinguishes N, NOM, and NP • SPR + for NP • SPR – for NOM

  19. The SPR feature • A generalization of the notion ‘determiner’ • Distinguishes N, NOM, and NP • SPR + for NP • SPR – for NOM • SPR + or SPR – for N

  20. The SPR feature • A generalization of the notion ‘determiner’ • Distinguishes N, NOM, and NP • SPR + for NP • SPR – for NOM • SPR + or SPR – for N • Also used to distinguish S and VP • S is SPR + • VP needs a subject NP to the left, so it is SPR – • Huh? p. 64, fig. 34

  21. Diagramming for Dollars

  22. Diagramming for Dollars

  23. Team 1: lexical entry for the noun “verb” As in, “Weird can be a verb” Team 2: lexical entry for the verb “verbs” As in, “Calvin verbs the word ‘weird’.”

  24. Diagramming for Dollars • Round 1: lexical entries • Round 2: p. 65, fig. 37a

  25. Diagramming for Dollars • Round 1: lexical entries • Round 2: p. 65, fig. 37a • Round 3: p. 69, fig. 47

  26. Agreement Rule • Agreement features get “passed up” from daughters to mothers • See rule, p. 70, fig. 50 • Tree in 51

  27. Head Feature PrincipleThere is a wisdom of the head… --Charles Dickens • Every headed phrase has a head daughter with the same head values • General form: p. 72, fig. 53 • Headed rules: p. 73, fig. 54, esp. 54d

  28. Diagramming for Dollars • Round 1: lexical entries • Round 2: p. 65, fig. 37a • Round 3: p. 69, fig. 47 • Round 4: “LING7420 loves HPSG.” “The professor cheers.”

More Related