... node^2.1

A null adjunction constraint (NA) is systematically put on the foot node of an auxiliary tree. This disallows adjunction of a tree onto the foot node itself.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... substitution^2.2

Technically, substitution is a specialized version of adjunction, but it is useful to make a distinction between the two.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... form.^2.3

Notice the similarity of the definition of a lexicalized grammar with the off line parsability constraint ([#!kaplan83!#]). As consequences of our definition, each structure has at least one lexical item (its anchor) attached to it and all sentences are finitely ambiguous.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...^2.4

abbreviations in the figure: t=top feature structure, tr=top feature structure of the root, br=bottom feature structure of the root, U=unification

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...^2.5

abbreviations in the figure: t=top feature structure, b=bottom feature structure, tr=top feature structure of the root, br=bottom feature structure of the root, tf=top feature structure of the foot, bf=bottom feature structure of the foot, U=unification

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... features.^2.6

The remaining constraint, Null Adjunction (NA), must still be specified directly on a node.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... entries.^3.1

This number does not include trees assigned by default based on the part-of-speech of the word.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...tex2html_comment_mark^3.2

The nonterminals in the tree database are A, AP, Ad, AdvP, Comp, Conj, D, N, NP, P, PP, Punct, S, V, VP.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... tree^3.3

The feature structures associated with each note of the parse tree are not shown here.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... items^3.4

Appendix E explains the conventions used in naming the trees.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... grammar^4.1

See Chapter 3 for details on these levels of representation.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... words^4.2

Such as non-auxiliary verbs or predicative nouns.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... structure.^4.3

Iteration of a structure can also be used as a diagnostic: Srini bought a book at the bookstore on Walnut Street for a friend.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... S-structure.^4.4

There are certain problems with applying the case filter as a requirement at the level of S-structure. These issues are not crucial to the discussion of the English XTAG implementation of case and so will not be discussed here. Interested readers are referred to [#!lasnik-uriagereka88!#].

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... prepositions.^4.5

For also assigns case as a complementizer. See section 8.5 for more details.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... tree^4.6

Features not pertaining to this discussion have been taken out to improve readability and to make the trees easier to fit onto the page.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... anchored^4.7

The diamond marker () indicates the anchor(s) of a structure if the tree has not yet been lexicalized.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... tree^4.8

Again, the feature structures shown have been restricted to those that pertain to the V/NP interaction.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... instead.^4.9

See section 21.1 for a more complete explanation of how this relinking occurs.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... framework.^4.10

See Sections  8.1 and 8.9 for additional discussion of PRO.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... figure.^5.1

Since Chapter 6 has a brief discussion and a declarative tree for every tree family, page references are given only for other sections in which discussion or tree diagrams appear.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... main^6.1

Auxiliary verbs are handled under a different mechanism. See Chapter 21 for details.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... family^6.2

See section 3.1.2 for explanation of tree families.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... system.^6.3

An explanation of the naming convention used in naming the trees and tree families is available in Appendix E.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... tree^6.4

Before lexicalization, the indicates the anchor of the tree.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... family.^6.5

Numbers given are as of August 1998 and are subject to some change with further development of the grammar.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... well.^6.6

No great attempt has been made to go through and decide which adjectives should actually take this family and which should not.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... well.^6.7

No great attempt has been made to go through and decide which nouns should actually take this family and which should not.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... callergative^7.1

The terminology is from [#!Burzio86!#]. See also [#!Perlmutter78!#] and [#!Rosen81!#] for discussion within the Relational Grammar framework.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... categories.^8.1

i.e. empty complementizers. We do have PRO and NP traces in the grammar.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... complementizer^8.2

Although we will continue to refer to `null' complementizers, in our analysis this is actually the absence of a complementizer.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... gerundive^8.3

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...

Most gerundive phrases are treated as NP's. In fact, all gerundive subjects are treated as NP's, and the only gerundive complements which receive a sentential parse are those for which there is no corresponding NP parse. This was done to reduce duplication of parses. See Chapter 18 for further discussion of gerunds. 

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... phrase.^8.4

Other considerations, such as the relationship between the tense/aspect of the matrix clause and the tense/aspect of a complement clause are also important but are not currently addressed in the current English XTAG grammar.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... <mode>^8.5

<mode> actually conflates several types of information, in particular verb form and mood.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... anchor^8.6

See section 4.2 for a discussion of the difference between complements and adjuncts in the XTAG grammar.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... it.^8.7

Because root S's cannot have complementizers, the parser checks that the root S has <comp>=nil at the end of the derivation, when the S is also checked for a tensed verb.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... former.^8.8

See Chapter 21 for a discussion of do-support.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... tree.^8.9

This does not mean that elementary trees with more than one gap should be ruled out across the grammar. Such trees might be required for dealing with parasitic gaps or gaps in coordinated structures.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... S-subjs.^8.10

Some speakers also find if as a complementizer only marginally grammatical in S-comps.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... grammar.^9.1

This chapter is strongly based on [#!heycock91!#]. Sections 9.1 and 9.2 are greatly condensed from her paper, while the description of the XTAG analysis in section 9.3 is an updated and expanded version.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... verbs^9.2

with the exception of have in British English. See footnote 1 in Chapter 21.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... structures^9.3

There are actually two other predicative trees in the XTAG grammar. Another predicative noun phrase tree is needed for noun phrases without determiners, as in the sentence They are firemen, and another prepositional phrase tree is needed for exhaustive prepositional phrases, such as The workers are below.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... shift.^10.1

In languages similar to English that have overt case marking indirect objects would be marked with dative case. It has also been suggested that for English the preposition to serves as a dative case marker.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... .^11.1

This is fine as an adjunct, but not as a complement.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... it-clefts^12.1

See e.g. [#!Ball91!#], [#!Delin89!#] and [#!Delahunty84!#] for more detailed discussion of types of it-clefts.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... small-clauses.^12.2

For additional discussion of equative or predicative-be see Chapter 9.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... anchor^13.1

A reduced set of features are shown for readability.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...^14.1

Features not pertaining to this discussion have been taken out to improve readability.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... <wh>=+^14.2

How is the only <wh>=+ adjective currently in the XTAG English grammar.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... ^15.1

No adjunct traces are represented in the XTAG analysis of adjunct extraction. Relative clauses on adjuncts do not have traces and consequently feature equations of the kind shown here are not present.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... wh-phrase.^15.2

The convention followed in naming relative clause trees is outlined in Appendix E.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... node.^15.3

Complementizers in clausal complementation are introduced by adjunction. See section 8.4.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... node^15.4

The feature equation used is NP_w.t:<wh> = +. Examples of NPs that could substitute under NP_w are whose mother, who, whom, and also which but not when and where which are treated as exhaustive +wh PPs.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... PP_w.t:modeind/inf^15.5

As is the case for NP_w substitution, any +wh-PP can substitute under PP_w. This is implemented by the following equation:
PP_w.t:wh Not all cases of pied-piping involve substitution of PP_w. In some cases, the P may be built in. In cases where part of the pied-piped PP is part of the anchor, it continues to function as an anchor even after pied-piping i.e. the P node and the NP_w nodes are represented separately.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...),^15.6

The determiner tree shown has the <rel-clause> feature built in. The RC analysis would give two parses in the absence of this feature.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...).^15.7

It may be of some interest that ((277)) is acceptable in certain dialects of Belfast English.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... participal).^16.1

We considered allowing bare indicative clauses, such as He died that others may live, but these were considered too archaic to be worth the additional ambiguity they would add to the grammar.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... complementizers:^16.2

While these sound a bit like extraposed relative clauses (see [#!kj87!#]), those move only to the right and adjoin to S; as these clauses are equally grammatical both sentence-initially and sentence-finally, we are analyzing them as adjunct clauses.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... allowed.^18.1

an exception being the NP positions in ``equative BE'' sentences, such as, John is my father.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... needed.^18.2

This feature is also needed to restrict the selection of gerunds in NP positions. For example, the subject and object NP's in the ``equative BE'' tree (Tnx0BEnx1) cannot be filled by gerunds, and are therefore assigned the feature gerund = -, which prevents gerunds (which have the feature gerund = +) from substituting into these NP positions.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... N.^18.3

Note that the determiner can adjoin to the gerund only from within the gerund tree. Adjunction of determiners to the gerund root node is prevented by constraining determiners to only select NP's with the feature gerund = -. This rules out sentences like Private markets approved of (*the) [the selling of bonds].

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...)).^18.4

The trees for genitive pronouns and genitive NP's have the root node labelled as D because they seem to function as determiners and as such, sequence with the rest of the determiners. See Chapter 19 for discussion on determiner trees.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...UN^19.1

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...3pl^19.2

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...-^19.3

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...+/-^19.4

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...

We use the symbol UN to represent the fact that the selectional restrictions for a given feature are unspecified, meaning the noun phrase that the determiner selects can be either positive or negative for this feature.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...

Except one which is 3sg.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...

Except one which is compl+.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...

A partitive can be either quan+ or quan-, depending upon the nature of the noun that anchors the partitive. If the anchor noun is modified, then the quantity feature is determined by the modifier's quantity value.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...3pl^19.5

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...

one differs from the rest of CARD in selecting singular nouns

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... grammar,^19.6

A more detailed discussion of this analysis can be found in [#!ircs:det98!#].

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... these^19.7

In addition to this tree, these would also anchor another auxiliary tree that adjoins onto card+ determiners.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... compounds.^20.1

Relative clauses are discussed in Chapter 15.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... determiners^20.2

See Chapter 19 or [#!ircs:det98!#] for details of the determiner analysis.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... so.^20.3

There may be other classes of NPs, such as directional phrases, such as north, south etc., which behave similarly. We have not yet analyzed these phrases.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...Pss^20.4

Clove healthy is an adjective small clause

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...sARB,^20.5

In the naming conventions for the XTAG trees, ARB is used for adverbs. Because the letters in A, Ad, and Adv are all used for other parts of speech (adjective, determiner and verb), ARB was chosen to eliminate ambiguity. Appendix E contains a full explanation of naming conventions.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... obtain)^20.6

Note that there are semantic/lexical constraints even for the categories that these adverbs can modify, and no doubt invite a more in-depth analysis.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... AP's^20.7

They can also appear before NP's, as in, ``John wants a little sugar''. However, here they function as multi-word determiners and should not be analyzed as adverbs.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... adverbials^20.8

It is to be noted that this analysis, which allows these multiword adverbs to modify adjectival phrases as well as adverbials, will yield (not necessarily desirable) multiple derivations for a sentence like John is a little unecessarily stupid. In one derivation, a little modifies the AP and in the other case, it modifies the adverb.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... 26^20.9

Though nearly all of these adverbs are spatial in nature, this number also includes a few temporal adverbs, such as ago, that also select these trees.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... form^21.1

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...

There are American dialects, particularly in the South, which allow double modals such as might could and might should. These constructions are not allowed in the XTAG English grammar.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... be^21.2

Some dialects, particularly British English, can also invert main verb have in yes/no questions (e.g. have you any Grey Poupon ?). This is usually attributed to the influence of auxiliary have, coupled with the historic fact that English once allowed this movement for all verbs. 

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... sentences.^21.3

We saw this tree briefly in section 4.4.3, but with most of its features missing. The full tree is presented here.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... tree^21.4

The declarative transitive tree was seen in section 6.3.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... it.^21.5

Earlier, we said that indicative mode carries tense with it. Since only the topmost auxiliary carries the tense, any subsequent verbs must not have indicative mode.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... be.^21.6

The inversion of main verb have in British English was previously noted.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... order^21.7

Some speakers accept dare preceded by a modal, as in I might dare finish this report today. In the XTAG analysis, this particular double modal usage is accounted for. Other cases of double modal occurrence exist in some dialects of American English, although these are not accounted for in the system, as was mentioned earlier. 

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... but.^22.1

We believe that the restriction of but to conjoining only two items is a pragmatic one, and our grammars accepts sequences of any number of elements conjoined by but.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... feature^22.2

See section 8.3 for an explanation of the <mode> feature.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... feature^22.3

See section 8.5 for an explanation of the <assign-comp> feature.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... compar+^23.1

The analysis given later for adjectival propositional comparatives produces aggregated compar+ adjectives such as more bright, which will also be incompatible (as desired) with ARBaPa.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... more.^23.2

We ignore here the interpretation in which the comparison covers the eating event, focussing only on the one which the comparison involves the stuff being eaten.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... boar.^23.3

This sentence differs from the metalinguistic comparison That stuff on her face is more than mud in that it involves a comment on the quantity and/or type of the compared NP, whereas the other expresses that the property denoted by the compared noun is an inadequate characterization of the thing being described.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... restrictions.^25.1

This section is a repeat of information found in section 20.1.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... English.^25.2

This section is from [#!ircs:det98!#].

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... own^25.3

The behavior of own is sufficiently unlike other determiners that it most likely needs a tree of its own, adjoining onto the right-hand side of genitive determiners.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... family.^25.4

This analysis may need to be extended to the Transitive Verb particle family as well.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... application^27.1

Actually more than one output tree can be generated from the successful application of a rule to an input tree, as will be seen later in this chapter

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... ?3452.^27.2

Notice however that having the sole purpose of distinguishing between variables, a number like the one in the last example is not very likely to occur, and a metarule with more than three thousand variables can give you a place in the Guinness TagBook of Records.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... stem^27.3

This is different from not having a subscript, which is interpreted as checking that the matching node at the input tree has no subscript

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...inp^27.4

Notice that, unlike the case for the constant node, the inverse is not required, i.e., if lhs_j has no marker, inp_g(j) is still allowed to have one.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... specifier^27.5

If the type specifier has a `?' subscript, there is no restriction, and that is exactly its function: to allow for the matching to be independent of the subscript.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... parameter^27.6

The parameter is accessible from the Lisp interface by the name XTAG::*metarules-copy-unmatched-trees*. At the end of section C.4 it is shown how to change the value of this parameter through the XTAG interface.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... output^27.7

As will be seen in section C.4 the cumulative mode of application is not affected by this parameter.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... equations^27.8

Notice that what is really important is not the features themselves, but the feature equations that relate the feature values of nodes of the same tree.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...>=+).^27.9

Commutativity of equations is accounted for in the system. Hence an equation x=y can also be specified as y=x. Associativity is not accounted for and its need by an user is viewed as indicating misspecification at the input trees.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... side^27.10

Although a buffer is intended to implement the concept of a set (not a sequence) of trees we take advantage of the actual organization of the system to realize the concept of (ordered) tree pair in the implementation.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... irrelevant^27.11

So that even if we want to have mnemonic names resembling their distinct character - left or right hand side, - we have some flexibility in naming them, e.g. lhs23 or lhs-passive.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... tree^27.12

The reason we do not use the name of the metarule, that is, the name of the buffer, is because in some forms of application the metarules do not carry individual names, which, as we will see, is the case when a set of metarules from a file is applied.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... trees^27.13

Remember that a metarule application generates as many output trees as the number of matches.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... metarule^27.14

If the file contains more than 2 trees, the additional trees are ignored.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

....^28.1

In order to focus on the use of tree descriptions and to make the figures less cumbersome, we show only the structural aspects and do not show the feature value specification. The parent, (immediate dominance), relationship is illustrated by a plain line and the dominance relationship by a dotted line. The arc between nodes shows the precedence order of the nodes are unspecified. The nodes' categories are enclosed in parentheses.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... matching^28.2

Matching occurs successfully when frame s is compatible with r_l in the type of anchors, the number of arguments, their positions, categories and features. In other words, incompatible features etc. will block certain LRRs from being applied.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... grammars^28.3

Both grammars are still under development, so the contents of these two tables might change a lot in the future according to the analyses we choose for certain phenomenon. For example, the majority of work on Chinese grammar treat ba-construction as some kind of object-fronting where the character ba is either an object marker or a preposition. According to this analysis, an LRR rule for ba-construction is used in our grammar to generate the preverbal-object frame from the postverbal frame. However, there has been some argument for treating ba as a verb. If we later choose that analysis, the main verbs in the patterns ``NP0 VP'' and ``NP0 ba NP1 VP'' will be different, therefore no LRR will be needed for it. As a result, the numbers of LRRs, subcat frames and tree generated will change accordingly.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... Penn^28.4

We have not yet attempted to extend our coverage to include punctuation, it-clefts, and a few idiosyncratic analyses.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... TAGs.^30.1

The displ-const feature is also used in the ECM analysis.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... type.^30.2

We have already seen one instance of a feature that marks clause-type: extracted, which marks whether a certain S involves extraction or not.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... chunks^31.1

We treat a sequence of verbs and verbal modifiers, including auxiliaries, adverbs, modals as constituting a verb group.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

....^31.2

It is important to note in this comparison that the supertagger uses lexical information on a per word basis only to pick an initial set of supertags for a given word.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... compared^31.3

We used the parseval program written by Phil Harison (phil@atc.boeing.com).

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... sentences^31.4

The Treebank was obtained through Salim Roukos (roukos@watson.ibm.com) at IBM.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

....^31.5

We are aware of the fact that increasing the number of constituents also increases the recall percentage. However we believe that this a legitimate gain.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... system.^31.6

When CLARE-3 is tuned to the ATIS domain, performance increases to 90%. However XTAG has not been tuned to the ATIS domain.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.