this is for holding javascript data
test1234567 added Id_overview_tex_v_1__.tex
about 9 years ago
Commit id: 479cfe491a3c47799beca20b659ee71977acaf96
deletions | additions
diff --git a/Id_overview_tex_v_1__.tex b/Id_overview_tex_v_1__.tex
new file mode 100644
index 0000000..e3308f7
--- /dev/null
+++ b/Id_overview_tex_v_1__.tex
...
% $Id: overview.tex,v 1.39 1994/12/13 04:49:36 fergus Exp $
%\input{stuff}
%\makeindex
%\begin{document}
\section{An Overview of Basic Clause Structure}
\label{list:overview}
This section presents of an overview of basic structure, but it does not
attempt to summarize or justify the entire policy. For an overview of the
new bracketing style, with some notes about its usefulness, see
\cite{marcusetal94} (included on the {\sc cdrom} release as ``arpa94'').
For an overview of the Treebank Project in general, see \cite{marcusetal93}
(``cl93''). Note also that, for the most part, this manual focuses on {\em
problematic}\/ constructions rather than {\em common}\/ ones, for mostly
historical reasons.
The basic structure of an S \index{S} in the Treebank grammar was formerly
(Preliminary Release, Version 0.5, 1992):
\begin{verbatim}
(S (NP Casey)
(VP throws
(NP the ball)))
\end{verbatim}
More complicated structures were annotated with an AUX \index{AUX} node.
\begin{verbatim}
(S (NP Casey)
(AUX will)
(VP throw
(NP the ball)))
\end{verbatim}
\begin{verbatim}
(S (NP Casey)
(AUX should)
(VP have
(VP thrown
(NP the ball))))
\end{verbatim}
The basic approach to simple sentences has not changed. However, we no
longer have a specially bracketed and labeled AUX; the node that was
formerly AUX \index{AUX} now corresponds to the highest level of the VP.
\begin{verbatim}
(S (NP-SBJ Casey)
(VP will
(VP throw
(NP the ball))))
\end{verbatim}
\begin{verbatim}
(S (NP-SBJ Casey)
(VP should
(VP have
(VP thrown
(NP the ball)))))
\end{verbatim}
\subsection{Basic elements of S} \index{S!basic elements}
\subsubsection{Predicate} \index{predicate}
The predicate is either the lowest (right-most branching) VP or (after
copular verbs and in ``small clauses'') a constituent tagged
\index{PRD@-PRD} -PRD. Moved predicates leave a coindexed trace *T* in VP.
\subsubsection{Arguments of the predicate} \index{predicate!arguments of}
\begin{itemize}
\item{External:}
The {\bf surface subject} \index{subject!surface} is tagged -SBJ (subject).
\index{SBJ@-SBJ}
\item{Internal:}
\begin{enumerate}
\item {\bf Direct object NP}\/: \index{objects!direct} occurs after the
verb, has no function tag, and is not followed by another NP.
\item {\bf Indirect object NPs} \index{objects!indirect} are of the
following types:
(i) NPs that occur between the verb and its direct object and have no
function tags, as in dative shift constructions (e.g., {\it gave {\bf Mary}\/
the book}\/)
\index{dative shift}
(ii) dative PPs (e.g., {\it to Mary}\/), \index{dative PPs} tagged -DTV.
Only verbs that can undergo dative shift are considered to have
\index{objects!dative} dative objects.
\end{enumerate}
\end{itemize}
\subsubsection{Level of attachment} \index{attachment!levels of}
\begin{itemize}
\item S-level: \index{S}
The following are attached at S-level: subject NP, highest VP, fronted
constituents, initial and final punctuation, and most modifiers that
precede the verb phrase. When there is no VP (as in ``small clauses''),
the predicate is labeled -PRD \index{PRD@-PRD} , and it and any following
adjuncts are attached at S-level.
\item VP-level: \index{VP}
\begin{enumerate}
\item Almost all modifiers \index{modifiers!postverbal} that follow the
verb are attached under the lowest appropriate VP. When there is
conjunction and the modifier applies to both VPs, the modifier is attached
at conjunction level.
\item An exception is made for modifiers \index{modifiers!as appositives}
\index{apposition!to predicate}
that are interpreted as appositives to the event or the predicate itself.
Such modifiers are adjoined to VP. Some of them may also have a -ADV tag.
\begin{verbatim}
(S (NP-SBJ Investors)
(VP might
(VP (VP appear
(ADJP-PRD unenthusiastic
(PP about
(NP the new issue))))
(SBAR (WHNP-1 which)
(S (NP-SBJ *T*-1)
(VP might
(VP force
(S (NP-SBJ the government)
(VP to
(VP raise
(NP the coupon)
(PP-CLR to
(NP (QP more than 7)
%))))))))))))
\end{verbatim}
\begin{verbatim}
(S (NP-SBJ they)
(VP would
(VP (VP negotiate
(NP rates)
(ADVP-MNR individually)
(PP-CLR with
(NP advertisers)))
,
(NP-ADV (NP a practice)
(ADJP common
(PP-LOC in
(NP broadcasting)))))))
\end{verbatim}
\begin{verbatim}
(S (NP-SBJ factory inventories)
(VP (VP fell
(NP-EXT 0.1 %)
(PP-TMP in
(NP September)))
,
(NP (NP the first decline)
(PP-TMP since
(NP February 1987)))))
\end{verbatim}
\end{enumerate}
\end{itemize}
\subsubsection{Complementation within syntactic categories}
\index{complementation}
The complement is attached inside the VP, NP, ADJP, or PP.
\begin{itemize}
\item{Verbs:}
The term ``complement'' as it is used here refers to: \index{complements}
\begin{enumerate}
\item internal arguments \index{arguments!internal} such as NP objects, S
and SBAR with no adverbial dash tags (including some {\it if}\/-clauses, as
in {\it I wonder if the Cubs are winning}\/), and quoted constituents
(including SINV and FRAG)
\item the passive logical-subject {\it by}\/-phrase \index{by-phrases@{\it
by}\/-phrases} \index{subject!logical}
\item VP \index{VP}
\item constituents tagged -BNF, -CLR, -DTV, -PRD, and -PUT
\end{enumerate}
\begin{verbatim}
(S (NP-SBJ-1 the guide)
(VP was
(VP given
(NP *-1)
(PP-DTV to
(NP Arthur))
(PP by
(NP-LGS Ford)))))
\end{verbatim}
\begin{verbatim}
(S (NP-SBJ-1 Casey)
(VP ought
(S (NP-SBJ *-1)
(VP to
(VP have
(VP thrown
(NP the ball)))))))
\end{verbatim}
\item{Nouns:} \index{nouns}
Since it is difficult to consistently annotate an argument/adjunct
distinction, all PP modifiers \index{modifiers!PP} of nouns are
Chomsky-adjoined to the NP:
\begin{verbatim}
(NP (NP a teacher)
(PP of
(NP chemistry)))
\end{verbatim}
However, clausal complements \index{complements!clausal} are recognized:
\begin{verbatim}
(NP the belief
(SBAR that
(S the world is flat)))
\end{verbatim}
\item{Adjectives:} \index{adjectives}
Except in comparatives, any modifier \index{modifiers!postadjectival}
following an adjective is bracketed as a complement.
\begin{verbatim}
(ADJP eager/likely/ready
(S to believe anything))
\end{verbatim}
\begin{verbatim}
(ADJP full
(PP of
(NP life)))
\end{verbatim}
\item{Prepositions:} \index{prepositional phrases}
The NP or S complement \index{complements!of prepositions} of a preposition
is placed inside the PP.
\end{itemize}
\subsubsection{Modification}
\begin{itemize}
\item{Premodifiers:} \index{premodifiers}
Premodifiers generally are placed inside the phrase they are associated
with:
\begin{verbatim}
(NP the red ball)
\end{verbatim}
\begin{verbatim}
(ADJP extremely delicious)
\end{verbatim}
\begin{verbatim}
(ADVP (NP one year) ago)
\end{verbatim}
VP premodifiers, \index{modifiers!preverbal} however, are more often
attached at S-level:
\begin{verbatim}
(S (NP-SBJ Sandy)
(ADVP-TMP often)
(VP throws
(NP curves)))
\end{verbatim}
but they may also be attached inside the VP (see \RefShared\ for more details):
\begin{verbatim}
(S (NP-SBJ Sandy)
(VP (ADVP-MNR sneakily)
threw
(NP a curve)))
\end{verbatim}
\item{Postmodifiers:} \index{postmodifiers}
Postmodifiers of NPs and comparative ADJPs are adjoined to the NP or ADJP:
\begin{verbatim}
(NP (NP a book)
(PP about
(NP toads)))
\end{verbatim}
\begin{verbatim}
(ADJP (ADJP as tall)
(PP as
(NP him)))
\end{verbatim}
Postmodifiers of VP \index{modifiers!postverbal} are attached under VP,
with adverbial function tag(s) where appropriate:
\begin{verbatim}
(VP reading
(PP-CLR about
(NP toads))
(PP-LOC on
(NP the Internet)))
\end{verbatim}
When it is not clear whether a modifier within the VP should be attached at
VP-level or to an object NP, the {\sc default} is to attach at VP-level (see
\RefPseudo).
% (was III.A)
\end{itemize}
\subsection{Clause types} \index{clause types}
We distinguish among a number of basic clause types: S, SINV, SBAR, RRC,
SBARQ, SQ, S-CLF, {\it it}\/-extraposition, and FRAG.
\subsubsection{S} \index{S}
\begin{itemize}
\item{Simple declarative sentences:}
\begin{verbatim}
(S (NP-SBJ Casey)
(VP threw
(NP the ball)))
\end{verbatim}
\item{Passives:} \index{passives}
The surface subject is tagged -SBJ \index{subject!surface} , the passive
trace \index{passive traces} is indicated with (NP *) and coindexed to the
surface subject, the {\it by}\/-phrase \index{by-phrases@{\it
by}\/-phrases} is a child of VP, and the logical subject
\index{subject!logical} is tagged -LGS
\index{LGS@-LGS} . (Note that the -LGS tag goes on the NP and not on the PP of
the {\it by}\/-phrase.)
\begin{verbatim}
(S (NP-SBJ-1 The ball)
(VP was
(VP thrown
(NP *-1)
(PP by
(NP-LGS Casey)))))
\end{verbatim}
\item{Imperatives:} \index{imperatives}
Imperatives are labeled S and given a null subject (NP-SBJ *).
\begin{verbatim}
(S (NP-SBJ *)
(VP Throw
(NP the ball))
!)
\end{verbatim}
If the name of the addressee appears with the imperative (at either the
beginning or end), it is tagged -VOC (vocative) \index{VOC@-VOC} . The
vocative is NOT coindexed to the null surface subject.
\begin{verbatim}
(S (NP-VOC American imperialists)
,
(NP-SBJ *)
(VP go
(ADVP-DIR home))
!)
\end{verbatim}
\begin{verbatim}
(S (NP-SBJ *)
(VP Close
(NP the door)
,
(NP-VOC John))
.)
\end{verbatim}
\item{Questions with declarative word order:} \index{questions}
Sentences that end with a question mark but have non-inverted word order
are labeled S:
\begin{verbatim}
(S (NP-SBJ This)
(VP is
(NP-PRD Japan))
?)
\end{verbatim}
\begin{verbatim}
(S (NP-SBJ You)
(VP did
(NP what))
?)
\end{verbatim}
However, questions that are missing both subject and auxiliary are labeled
SQ \index{SQ} .
\begin{verbatim}
(SQ (NP-SBJ *)
(VP See
(NP that cute dog))
?)
\end{verbatim}
\item{Infinitives:} \index{infinitives}
Infinitives are labeled S and take (NP-SBJ *) as the null subject,
where {\it to}\/ represents the highest level of the VP. (See \RefInf\
for more on the annotation of infinitivals.)
\begin{enumerate}
\item{Complement clauses.} \index{complements!clausal}
When the infinitive is a VP complement, the null subject of the infinitive
is coindexed as usual to its logical subject (usually the subject of the
matrix clause, but sometimes the object of the verb or not coindexed at
all).
\begin{verbatim}
(S (NP-SBJ-1 Casey)
(VP wants
(S (NP-SBJ *-1)
(VP to
(VP throw
(NP the ball))))))
\end{verbatim}
\item{Purpose clauses.} \index{purpose clauses}
Purpose clauses are attached at S and labeled -PRP (purpose/reason)
\index{PRP@-PRP} . The subject is coindexed to the surface subject of the
matrix clause when there is a coindexed interpretation.
\begin{verbatim}
(S (NP-SBJ-1 Sue)
(VP arrived
(ADVP-TMP early)
(S-PRP (NP-SBJ *-1)
(VP to
(VP get
(NP a good seat))))))
\end{verbatim}
\item{Infinitival relatives.} \index{infinitival relatives}
In the case of infinitival relatives, the relative is adjoined to NP and
dominated by SBAR with a zero {\it wh}\/-complementizer labeled according
to the role played by the gapped constituent. A *T* in the position of the
gap is coindexed to the {\it wh}\/-complementizer. The (NP-SBJ *) is not
indexed.
%A- never subject IRC??
\begin{verbatim}
(NP (NP a movie)
(SBAR (WHNP-1 0)
(S (NP-SBJ *)
(VP to
(VP see
(NP *T*-1))))))
\end{verbatim}
\end{enumerate}
\item{Participial and gerund clauses:} \index{participles} \index{gerunds}
Participial clauses have full clause structure, with either a lexical or
null (NP-SBJ *) subject. (See \RefGP\ for more on the annotation of
participial and gerund clauses.)
``Floating'' participles are tagged -ADV.
\index{participles!floating@``floating''}
\begin{verbatim}
(S (S-ADV (NP-SBJ The crowd)
(VP cheering
(ADVP-MNR madly)))
,
(NP-SBJ Willie)
(VP caught
(NP the ball)))
\end{verbatim}
When appropriate, the null subject is coindexed to an NP in the matrix
clause (generally the logical/structural subject).
\begin{verbatim}
(S (S-ADV (NP-SBJ *-1)
(VP Running
(PP-DIR toward
(NP Casey))))
,
(NP-SBJ-1 Willie)
(VP caught
(NP the ball)))
\end{verbatim}
Gerunds \index{gerunds} that act as the surface subject or as the object of
a preposition are tagged -NOM \index{NOM@-NOM} .
\begin{verbatim}
(S (S-NOM-SBJ (NP-SBJ *)
(VP Baking
(NP pies)))
(VP is
(ADJP-PRD fun)))
\end{verbatim}
\begin{verbatim}
(S (PP With
(S-NOM (NP-SBJ interest rates)
(VP rising)))
,
(NP-SBJ the market)
(VP is
(VP moving
(ADVP-MNR slowly))))
\end{verbatim}
Verb complement gerunds \index{gerunds!as complement of verb} , including
those in ``serial'' verb constructions, are all bracketed as a simple S,
with coindexed subject when appropriate.
\begin{verbatim}
(S (NP-SBJ I)
(VP do not
(VP mind
(S (NP-SBJ you(r))
(VP leaving
(ADVP-TMP early))))))
\end{verbatim}
\begin{verbatim}
(S (NP-SBJ-1 The audience)
(VP keeps
(S (NP-SBJ *-1)
(VP leaving
(ADVP-TMP early)))))
\end{verbatim}
SBAR complement gerunds \index{gerunds!as complement of SBAR} are also
bracketed as a simple S.
\begin{verbatim}
( (S (NP-SBJ-1 He)
(VP ate
(NP television)
(SBAR-TMP while
(S (NP-SBJ *-1)
(VP watching
(NP dinner)))))
.))
\end{verbatim}
\end{itemize}
\subsubsection{SINV} \index{SINV}
The SINV label is used for subject-auxiliary inversion \index{inversion} in
the case of negative inversion, conditional inversion, locative inversion,
and some topicalizations. (SINV is not used with questions. See
section \ref{list:overview:clausetype:sq:y/n:inversion} and
section \ref{list:overview:clausetypes:sbarq:wh-questions:inversion} for the
treatment of subject-auxiliary inversion in the case of yes/no questions
and {\it wh}\/-questions, respectively.) Inverted auxiliaries are
unlabeled.
\begin{verbatim}
(SINV (ADVP-TMP Never)
had
(NP-SBJ I)
(VP seen
(NP such a place)))
\end{verbatim}
When the inversion results in a conditional clause \index{conditionals}
(i.e., when it is equivalent to (SBAR-ADV if... ), the SINV is enclosed in
SBAR-ADV).
\index{SBAR-ADV} \index{if-clauses@{\it if}\/-clauses}
\begin{verbatim}
(S (SBAR-ADV (SINV had
(NP-SBJ Casey)
(VP thrown
(NP the ball)
(ADVP-MNR harder))))
,
(NP-SBJ it)
(VP would
(VP have
(VP reached
(NP home plate)
(PP-TMP in
(NP time))))))
\end{verbatim}
When subject-aux inversion \index{inversion} is triggered by the predicate
(or a piece thereof) moving out of the VP to a position preceding the
subject, the moved predicate phrase is tagged -TPC (topicalized)
\index{TPC@-TPC} and leaves a coindexed trace in the VP. Note that in this
case, the auxiliaries that precede the subject are labeled VP (whereas in
other cases of inversion they are left unlabeled), so that there's a place
to properly attach the VP trace.
\begin{verbatim}
(SINV (VP-TPC-1 Marching
(PP-DIR past
(NP the reviewing stand)))
(VP were
(VP *T*-1))
(NP-SBJ 500 musicians))
\end{verbatim}
\begin{verbatim}
(SINV (ADJP-PRD-TPC-2 (ADVP Even more)
unusual)
(VP is
(ADJP-PRD *T*-2))
(NP-SBJ (NP the mating behavior)
(PP of
(NP the praying mantis))))
\end{verbatim}
\begin{verbatim}
(SINV (ADVP Also)
(ADJP-PRD-TPC-11 present)
(VP will
(VP be
(ADJP-PRD *T*-11)))
(NP-SBJ (NP (NP the bride 's)
children)
,
(NP Joan and Kirkland)))
\end{verbatim}
\begin{verbatim}
(SINV (ADVP-DIR-TPC-3 Out)
(VP might
(VP have
(VP popped
(ADVP-DIR *T*-3))))
(NP-SBJ a jack-in-the-box))
\end{verbatim}
\begin{verbatim}
(SINV (S-TPC-4 (NP-SBJ We)
(VP will
(VP win)))
,
(VP said
(SBAR 0
(S *T*-4))
(NP-SBJ Casey)))
\end{verbatim}
\subsubsection{SBAR} \index{SBAR}
SBAR is used for relative clauses and subordinate clauses, including
indirect questions. \index{questions!indirect}
\begin{verbatim}
(S (NP-SBJ (NP The person)
(SBAR (WHNP-1 who)
(S (NP-SBJ *T*-1)
(VP threw
(NP the ball)))))
(VP is
(ADJP-PRD very athletic)))
\end{verbatim}
\begin{verbatim}
(S (NP-SBJ Willie)
(VP knew
(SBAR that
(S (NP-SBJ Casey)
(VP threw
(NP the ball))))))
\end{verbatim}
\begin{verbatim}
(S (NP-SBJ Willie)
(VP asked
(SBAR (WHNP-1 who)
(S (NP-SBJ *T*-1)
(VP threw
(NP the ball))))))
\end{verbatim}
The {\it wh}\/-prefixed labels, WHNP, \index{WHNP} WHADVP, \index{WHADVP}
WHADJP, \index{WHADJP} WHPP, \index{WHPP} are used only when there is {\it
wh}\/-movement, \index{wh-movement@{\it wh}\/-movement} and they always
leave a trace *T*. (See \RefWHph\ \index{wh-clauses@{\it
wh}\/-clauses} for more information.)
\begin{verbatim}
(S (NP-SBJ-1 The committee)
(VP continued
(NP its meeting)
(SBAR-TMP while
(S (NP-SBJ *-1)
(VP eating
(NP lunch))))))
\end{verbatim}
\subsubsection{RRC (reduced relative clause)} \index{RRC} \index{relative
clauses}
Reduced relative clauses are adjoined to the NP they modify.
%How are we going to refer readers of the overview who want more detailed
%info to the appropriate section of the manual? The section on reduced
%relatives, for example, is deep in Gerunds-Participles and might require
%some searching through the table of contents to find. Shall we give
%references for every section heading in here?
%
%A- the reference below seems OK to me.
\begin{verbatim}
(NP (NP An orangutan)
(VP foaming
(PP-CLR at
(NP the mouth))))
\end{verbatim}
All sentential modifiers in reduced relatives are attached at VP level,
even when they precede the verb itself.
\begin{verbatim}
(NP (NP a car)
(VP not built
(NP *)
(PP by
(NP-LGS Mazda))
(PP-TMP for
(NP the last five years))))
\end{verbatim}
The RRC label \index{RRC} is used only in cases where there is no VP and an
extra level is needed for proper attachment of sentential modifiers (see
\RefGP).
%was VII.A.2
\begin{verbatim}
(NP (NP 110 titles)
(RRC not
(ADVP-TMP presently)
(PP-LOC in
(NP the collection))))
\end{verbatim}
In the case of passives, \index{relative clauses!passive} a passive trace
(NP *) is inserted under the VP, but it is not indexed, since the null
subject to which the passive trace \index{passive traces} would be
coindexed is not present in the annotation. The logical subject is tagged
-LGS.
\begin{verbatim}
(S (NP-SBJ I)
(VP bought
(NP (NP a car)
(VP built
(NP *)
(PP by
(NP-LGS Mazda))))))
\end{verbatim}
Compare with the unreduced:
\begin{verbatim}
(S (NP-SBJ I)
(VP bought
(NP (NP a car)
(SBAR (WHNP-5 which)
(S (NP-SBJ-6 *T*-5)
(VP was
(VP built
(NP *-6)
(PP by
(NP-LGS Mazda)))))))))
\end{verbatim}
\subsubsection{SBARQ} \index{SBARQ}
\label{list:overview:clausetypes:sbarq:wh-questions:inversion}
The SBARQ label marks {\it wh}\/-questions \index{wh-questions@{\it
wh}\/-questions} (i.e., those that contain a gap and therefore require a
trace). A further level of structure, SQ, \index{SQ} contains the inverted
auxiliary (if there is one) and the rest of the sentence. The inverted
auxiliary in {\it wh}\/-questions is not labeled. \index{inverted auxiliary}
\begin{verbatim}
(SBARQ (WHNP-1 Who)
(SQ (NP-SBJ *T*-1)
(VP threw
(NP the ball)))
?)
\end{verbatim}
\begin{verbatim}
(SBARQ (WHNP-2 What)
(SQ did
(NP-SBJ Casey)
(VP throw
(NP *T*-2)))
?)
\end{verbatim}
\begin{verbatim}
(SBARQ (WHNP-3 Who)
(SQ (NP-SBJ *T*-3)
(VP will
(VP throw
(NP the ball))))
?)
\end{verbatim}
If the main verb is inverted, \index{inversion} most arguments and adjuncts
go at SQ level:
\begin{verbatim}
(SBARQ (WHADVP-1 Why)
(SQ (PP-LOC in
(NP these movies))
is
(NP-SBJ the unwed pregnant woman)
(ADVP-TMP always)
(PP-PRD from
(NP Ohio))
(ADVP-PRP *T*-1))
?)
\end{verbatim}
\subsubsection{SQ} (See also section \ref{list:overview:s-clf:sq-clf}.)
\index{SQ}
\begin{itemize}
\item{inside SBARQ:}
As described above, inside {\it wh}\/-questions, SQ holds the subject,
inverted auxiliary (if any), main verb phrase, and some adjuncts.
\item{yes/no questions:}
\label{list:overview:clausetype:sq:y/n:inversion}
SQ is used for yes/no questions \index{questions!yes/no} (i.e., those with
inversion but no {\it wh}\/-movement).
\begin{verbatim}
(SQ Did
(NP-SBJ Casey)
(VP throw
(NP the ball))
?)
\end{verbatim}
\item{subject-less yes/no questions:} \index{questions!yes/no}
In questions where the auxiliary and subject do not appear, the auxiliary
is unlabeled and a null subject (NP-SBJ *) is used.
\begin{verbatim}
(SQ (NP-SBJ *)
(VP See
(NP that cute dog))
?)
\end{verbatim}
Note that questions with overt subjects and auxiliaries that show
declarative word order are simply labeled S.
\item{Tag questions:} \index{questions!tag}
Tag questions are treated as an adjunction of SQ to S. The resulting
structure is labeled SQ, since the whole thing is interrogative in nature.
The lower SQ is annotated to show predicate deletion; that is, an
appropriate null *?* is inserted.
\begin{verbatim}
(SQ (S But
(NP-SBJ you)
(VP knew
(NP that)))
,
(SQ did n't
(NP-SBJ you)
(VP *?*))
?)
\end{verbatim}
\begin{verbatim}
(SQ (S (NP-SBJ That)
(VP 's
(NP-PRD the problem)))
,
(SQ is
n't
(NP-SBJ it)
(NP-PRD *?*))
?)
\end{verbatim}
\end{itemize}
\subsubsection{S-CLF} ({\it it}\/-cleft or ``true'' cleft) \index{CLF@-CLF}
\index{clefts} \index{it-clefts@{\it it}\/-clefts}
Declarative {\it it}\/-clefts are labeled S-CLF, expletive {\it it}\/ is
tagged as the surface subject (-SBJ), the SBAR is attached at VP-level, and
a trace is coindexed to the {\it wh}\/-complementizer of the clefted
portion. (See \RefIt\ for more information.)
\begin{verbatim}
(S-CLF (NP-SBJ It)
(VP was
(NP Casey)
(SBAR (WHNP-1 who)
(S (NP-SBJ *T*-1)
(VP threw
(NP the ball))))))
\end{verbatim}
Traces for all adverbials, including purpose clauses, are labeled (ADVP
*T*) and coindexed to the WHADVP complementizer. If a preposition is
pied-piped, WHPP and (PP *T*) are used.
\begin{verbatim}
(S-CLF (NP-SBJ It)
(VP is
(ADVP-TMP-PRD then)
(SBAR (WHADVP-1 that)
(S (NP-SBJ-2 young queens)
(VP begin
(S (NP-SBJ *-2)
(VP to
(VP appear)))
(ADVP-TMP *T*-1))))))
\end{verbatim}
Interrogative {\it it}\/-clefts are labeled SQ-CLF:
\index{clefts!interrogative}
\label{list:overview:s-clf:sq-clf}
\begin{verbatim}
(SQ-CLF Was
(NP-SBJ it)
(NP-PRD John)
(SBAR (WHNP-1 who)
(S (NP-SBJ *T*-1)
(VP came
(PP-DIR to
(NP the party))
(PP-CLR in
(NP a dress)))))
?)
\end{verbatim}
Note: {\it wh}\/-clefts do not receive special treatment in the corpus.
They contain a free/headless relative, followed by a form of the verb {\it
be}\/, followed by a predicate (e.g., {\it What sustained the worst
injuries / was / the car}\/). (See \RefNotation\ for a bit more on free
relatives.)
\index{wh-clefts@{\it wh}\/-clefts} \index{clefts}
\subsubsection{{\it it}\/-extraposition} \index{it-extraposition@{\it
it}\/-extraposition}
Clauses that are extraposed from subject position are labeled S or SBAR.
The extraposed clause is attached at VP level and adjoined to the ``it''
with *EXP*-attach. The NP containing {\it it}\/ and *EXP* is tagged -SBJ.
(See \RefExp\ for more information.)
\begin{verbatim}
(S (NP-SBJ (NP It)
(S *EXP*-1))
(VP is
(NP-PRD a pleasure)
(S-1 (NP-SBJ *)
(VP to
(VP teach
(NP her))))))
\end{verbatim}
\subsubsection{FRAG} (fragment) \index{FRAG}
FRAG marks those portions of text that appear to be clauses, but lack too
many essential elements for the exact structure to be easily determined
(e.g., answers to questions). Predicate argument structure therefore
cannot be extracted from FRAGs. Some examples of what we have called FRAG:
\begin{verbatim}
(SBARQ (WHNP-9 Who)
(S (NP-SBJ *T*-9)
(VP threw
(NP the ball)))
?)
(FRAG (NP Casey)
,
(NP-TMP yesterday))
\end{verbatim}
\begin{verbatim}
(S (SBAR-ADV if
(FRAG not
(NP-TMP today)))
,
Casey will throw the ball tomorrow)
\end{verbatim}
\begin{verbatim}
(SBAR-ADV Though
(FRAG (ADJP limited)))
\end{verbatim}
\begin{verbatim}
(SBAR-ADV if
(FRAG (ADJP possible)))
\end{verbatim}
\begin{verbatim}
(SBAR-ADV (WHNP Whatever)
(FRAG (NP the long-term economic effect)))
\end{verbatim}
\begin{verbatim}
(FRAG (PP-LOC Among
(NP (NP the Guinness disk 's)
wonders))
:
(NP (NP the world 's)
loudest recorded belch)
.)
\end{verbatim}
\begin{verbatim}
( (FRAG (NP (NP Two guys)
(PP from
(NP (NP Gary)
,
(NP Ind.))))
?
''))
\end{verbatim}
\begin{verbatim}
( (FRAG Not
(ADJP so)
.))
\end{verbatim}
\begin{verbatim}
( (FRAG (VP Guaranteed
(NP *)
(PP by
(NP-LGS India)))
.))
\end{verbatim}
\subsection{Clause combinations}
\subsubsection{Coordination} \index{coordination}
See \RefShared\ and \RefCoord\ for details concerning coordination
structures.
\begin{itemize}
\item Phrase coordination
Coordination of phrases is represented in the annotation at the lowest
level possible. Single words are assumed to coordinate at word level
rather than projecting their own phrases, and only the highest level is
represented.
\begin{verbatim}
(S (NP-SBJ Girls and boys)
(VP throw and catch
(NP balls)))
\end{verbatim}
However, the addition of modifiers generally forces a higher level of
coordination, which is shown with Chomsky-adjunction structure.
\begin{verbatim}
(S (NP-SBJ (NP These girls)
and
(NP those boys))
(VP (VP throw
(ADVP-MNR well))
and
(VP catch
(ADVP-MNR badly))))
\end{verbatim}
\item Clause coordination
When like clauses are coordinated, the level of coordination has the same
label as the coordinated clauses.
\begin{verbatim}
(S (S (NP-SBJ Casey)
(VP threw
(NP the ball)))
and
(S (NP-SBJ Willie)
(VP caught
(NP it))))
\end{verbatim}
\begin{verbatim}
(S (NP-SBJ Jackie)
(VP knew
(SBAR (SBAR (WHNP-1 who)
(S (NP-SBJ *T*-1)
(VP threw
(NP the ball))))
and
(SBAR (WHNP-2 who)
(S (NP-SBJ *T*-2)
(VP caught
(NP it)))))))
\end{verbatim}
Coordinated unlike clauses are dominated by S.
\begin{verbatim}
(S (S (NP-SBJ He)
(VP was
(VP performing
(PP for
(NP pay)))))
, and
(SBARQ (WHADVP-1 why)
(SQ should
(NP-SBJ anyone)
(VP expect
(NP (NP anything)
(ADJP more))
(ADVP-PRP *T*-1))))
?)
\end{verbatim}
Note that coordination of unlike phrases (i.e., non-clauses) is dominated
by UCP (Unlike Coordinated Phrase). \index{UCP}
\item Coordinating conjunctions \index{conjunctions!coordinating}
\index{coordination}
Besides the usual {\it and, or, but}\/, etc., certain prepositions and
subordinating conjunctions can be used as coordinating conjunctions.
Multi-word coordinating conjunctions are labeled CONJP \index{CONJP} (see
\RefCoord).
%was I.
\begin{verbatim}
(S (NP-SBJ Willie
(CONJP as well as)
Casey)
(VP saw
(NP the ball)))
\end{verbatim}
\begin{verbatim}
(S (NP-SBJ Casey)
(VP will
(VP throw
(NP the ball)
(NP-TMP tomorrow
(CONJP if not)
today))))
\end{verbatim}
Otherwise, these are annotated as PPs or SBARs, as appropriate. (See
\RefCoord.)
%was Coordination.I.B.)
\begin{verbatim}
(S (SBAR-ADV if
(FRAG not
(NP today)))
,
(NP-SBJ Casey)
(VP will
(VP throw
(NP the ball)
(NP-TMP tomorrow))))
\end{verbatim}
\end{itemize}
\subsubsection{Subordination: the use of SBAR} \index{SBAR}
See \RefWHph\ and \RefSubord\ for more detail.
\begin{itemize}
\item Relative clauses are adjoined to the NP that they modify.
\index{relative clauses}
\begin{verbatim}
(S (NP-SBJ (NP The person)
(SBAR (WHNP-1 who)
(S (NP-SBJ *T*-1)
(VP caught
(NP the ball)))))
(VP is
(ADJP-PRD very athletic)))
\end{verbatim}
\item SBARs can be verbal complements.
\begin{verbatim}
(S (NP-SBJ Willie)
(VP knew
(SBAR that
(S (NP-SBJ Casey)
(VP hid
(NP the ball))))))
\end{verbatim}
\index{SBAR}
\item Conditional, temporal, and other such adverbial SBARs are
attached under either S or VP, depending on whether they precede or follow
the main clause, and given the appropriate adverbial function tag.
\begin{verbatim}
(S (SBAR-ADV if
(S (NP-SBJ Casey)
(VP throws
(NP the ball))))
(NP-SBJ Willie)
(VP catches
(NP it)))
\end{verbatim}
\begin{verbatim}
(S (NP-SBJ Willie)
(VP catches
(NP the ball)
(SBAR-ADV if
(S (NP-SBJ Casey)
(VP throws
(NP it))))))
\end{verbatim}
\end{itemize}
\subsubsection{Fronted elements} \index{fronted constituents}
Fronted elements are those that appear before the subject in a declarative
sentence. They are placed inside the top clause level (e.g. S, SINV, SQ,
SBAR). (See the section on -TPC in \RefNotation\ for more details on the
use of the -TPC tag and the section on *T* with fronted elements in
\RefNulls\ for more details on the distribution of *T*.)
\begin{itemize}
\item Fronted arguments.
Fronted arguments are attached under the main clause level. They always
leave a *T* and are tagged -TPC. This holds whether the argument is
fronted within a single clause or crosses more than one clause boundary.
\begin{verbatim}
(S (NP-TPC-5 This)
(NP-SBJ every man)
(VP contains
(NP *T*-5)
(PP-LOC within
(NP him))))
\end{verbatim}
\item Fronted adjuncts. \index{fronted constituents} \index{TPC@-TPC}
Fronted adjuncts receive function tags (e.g, -ADV, -TMP) as appropriate.
However, note that they receive the -TPC label only in cases where they are
associated with a *T* in a lower clause. In particular, a *T* appears in
the annotation only if the adjunct is fronted over more than one clause
boundary:
\begin{verbatim}
(S (PP-TPC-1 Excluding
(NP (NP (NP an increase)
(PP-LOC in
(NP the tax rate)))
and
(NP (NP the effects)
(PP of
(NP foreign currency translations)))))
,
(NP-SBJ Mr. Millis)
(VP said
(SBAR 0
(S (NP-SBJ (NP the company 's)
results)
``
(VP were
(ADVP still)
(ADJP-PRD (NP-ADV a little)
disappointing)
(PP *T*-1)))))
. '')
\end{verbatim}
\begin{verbatim}
(S (SBAR-ADV-TPC-1 If
(S (NP-SBJ profits)
(VP do n't
(VP improve))))
,
(NP-SBJ Mr. Whitten)
(VP says
(SBAR 0
(S (NP-SBJ he)
(VP may
(VP quit
(NP the exchange)
(SBAR-ADV *T*-1))))))
.)
\end{verbatim}
There is no *T* in the annotation if the adjunct is fronted within a single
clause:
\index{fronted constituents}
\begin{verbatim}
(S (NP-TMP Yesterday)
(NP-SBJ I)
(VP went
(PP-DIR to
(NP the store))))
\end{verbatim}
\begin{verbatim}
(S (S-ADV (NP-SBJ *-1)
(VP Running
(PP-DIR toward
(NP Casey))))
,
(NP-SBJ-1 Willie)
(VP caught
(NP the ball)))
\end{verbatim}
\begin{verbatim}
(S (SBAR-TMP (WHADVP-7 When)
(S (NP-SBJ I)
(VP do n't
(VP get
(NP enough sleep)
(ADVP-TMP *T*-7)))))
(NP-SBJ-8 I)
(VP have
(NP (NP trouble)
(S-NOM (NP-SBJ *-8)
(VP staying
(ADJP-PRD awake))))))
\end{verbatim}
Elements that are fronted within questions (either {\it wh}\/- or yes/no)
are put inside the highest level of structure (SBARQ or SQ, respectively).
\index{questions}
\index{fronted constituents}
\begin{verbatim}
(SBARQ (SBAR-ADV If
(S (NP-SBJ Casey)
(VP throws
(NP the ball))))
(WHNP-1 who)
(SQ will
(NP-SBJ *T*-1)
(VP catch
(NP it)))
?)
\end{verbatim}
\begin{verbatim}
(SQ (SBAR-ADV If
(S (NP-SBJ Casey)
(VP throws
(NP the ball))))
will
(NP-SBJ Willie)
(VP catch
(NP it))
?)
\end{verbatim}
\end{itemize}
\subsubsection{Quotations} \index{quotations}
\begin{itemize}
\item Direct quotations \index{quotations!direct}
A direct quotation is considered to be the argument of the verb of saying.
\begin{verbatim}
(S (NP-SBJ Casey)
(VP said
``
(S (NP-SBJ Willie)
(VP caught
(NP the ball))))
.
'')
\end{verbatim}
When the quotation appears before the quoting verb, it treated as a fronted
argument: the quote is attached at S level and given a -TPC tag, and a
trace is shown under the VP.
\begin{verbatim}
(S ``
(S-TPC-1 (NP-SBJ Willie)
(VP caught
(NP the ball)))
,
''
(NP-SBJ Casey)
(VP said
(S *T*-1))
.)
\end{verbatim}
\begin{verbatim}
(SINV ``
(S-TPC-1 (NP-SBJ Willie)
(VP caught
(NP the ball)))
,
''
(VP said
(S *T*-1))
(NP-SBJ Casey)
.)
\end{verbatim}
If the quotation is discontinuous, the interruptive material is annotated
as a parenthetical (-PRN). Note that a trace appears under the VP in the
parenthetical, but that the fronted portion is not labeled -TPC:
\begin{verbatim}
(S-7 ``
(NP-SBJ Willie)
''
(PRN ,
(S (NP-SBJ Casey)
(VP said
(S *T*-7)))
,)
``
(VP caught
(NP the ball))
.
'')
\end{verbatim}
\item Indirect quotations \index{quotations!indirect}
The (SBAR 0) level marks indirect quotations and cases where only a portion
of the quote is direct.
\index{SBAR 0}
\begin{verbatim}
(S (NP-SBJ Stokely)
(VP says
(SBAR 0
(S (NP-SBJ stores)
(VP revive
(NP (NP specials)
(PP like
(NP (NP three cans)
(PP of
(NP peas))
(PP for
(NP 99 cents)))))))))
.)
\end{verbatim}
\begin{verbatim}
(S (NP-SBJ Mr. Millis)
(VP said
(SBAR 0
(S (NP-SBJ (NP the company 's)
results)
``
(VP were
(ADVP still)
(ADJP-PRD (NP-ADV a little)
disappointing)))))
. '')
\end{verbatim}
\end{itemize}
\subsubsection{Gapping} \index{gapping}
See \RefCoord\ for more details on gapping.
\begin{itemize}
\item Intrasentential gapping \index{gapping!template}
\index{gapping!intrasentential}
When the gapped construction exists alongside a complete (ungapped) clause
of parallel structure, the complete clause is used as a template to which
elements in the gapped clause are mapped via the ``='' notation (referred
to as ``gap coindexing''). \index{gap coindexing}
\begin{verbatim}
(S (S (NP-SBJ-1 Mary)
(VP likes
(NP-2 Bach)))
and
(S (NP-SBJ=1 Susan)
,
(NP=2 Beethoven)))
\end{verbatim}
In the above, the equal sign notation maps constituent NP=1 over NP-1 so
that the following predicate argument structure can be extracted: {\sc
like}\/ (Mary, Bach) and {\sc like}\/ (Susan, Beethoven). Similarly:
\begin{verbatim}
(S (NP-SBJ John)
(VP (VP gave => GIVE (John, Mary, book)
(NP-1 Mary)
(NP-2 a book))
and
(VP (NP=1 Bill) => GIVE (John, Bill, pencil)
(NP=2 a pencil))))
\end{verbatim}
All constituents in gapped constructions receive function tags as
appropriate.
\begin{verbatim}
(S (NP-SBJ I)
(VP (VP eat
(NP-1 breakfast)
(PP-TMP-2 in
(NP the morning)))
and
(VP (NP=1 lunch)
(PP-TMP=2 in
(NP the afternoon)))))
\end{verbatim}
\item Intersentential gapping \index{gapping!intersentential}
The template approach to gapping is not used across sentences. If a
template is not available within the same sentence, the gapped constituent
is labeled FRAG instead.
{\it What is Tim eating? Mary Ann thinks chocolate.}
\begin{verbatim}
(S (NP-SBJ Mary Ann)
(VP thinks
(SBAR 0
(FRAG (NP chocolate)))))
\end{verbatim}
\end{itemize}
%\printindex
%\end{document}
