DICTIONARY TEXT ENTRIES AS A SOURCE OF KNOWLEDGE 
FOR SYNTACTIC AND OTHER DISAMBIGUATIONS 
Karen Jensen and Jean-Louis BinoO 
IBM Thomas J. Watson Research Center 
P.O. Box 704 
Yorktown Heights, New York 10598 
Abstract 
Online reference books may be thought of as 
knowledge bases. We describe here how infor- 
mation in the text of machine-readable dictionary 
entries can be processed to help determine the 
proper attachment of prepositional phrases and 
relative clauses; the resolution of some cases of 
pronoun reference; and the interpretation of 
dangling modifiers. This approach also suggests 
the possibility of bypassing conventional efforts 
at hand-coding semantic information, efforts 
which ate time-consuming and usually incom- 
plete. 
0. INTRODUCrION 
Online reference books may be thought of as 
knowledge bases, with data structures encoded in 
natural language. We have developed a system 
that reasons heuristically about the comparative 
likelihood of various potential attachments for 
prepositional phrases in English sentences by 
analyzing relevant definitions in Webster's online 
dictionary (W7) in their original text form (Binot 
and Jensen 1987, Jensen and Binot forthcoming). 
This paper reviews that earlier work and then 
extends it by suggesting how additional informa- 
tion (particularly example .,entences from an- 
other dictionary, the Longman Dictionary of 
Contemporary English (I,DOCE)) might be used 
to cope with throe additional problems: the at- 
tachment of relative clauses, the resolution of 
some cases of pronoun reference, and the inter- 
pretation of dangling modifiers. The earlier work 
on PP attachments has been implemented, but 
we have only begun work on the implementation 
of these additional disambiguation problems. 
Nevertheless, it seems like a good idea to indicate 
that this dictionary-based approach should be 
feasible for mora than PP attachments. 
Our objective is to consult the dictionary to 
find the kind of information that has previously 
been supplied by means of scripts, frames, tem- 
plates, and other hand-crafted devices. This ap- 
proach offers hope for reducing time-consuming, 
and usually incomplete, hand-codings of seman- 
tic information: and it should be of particular 
interest for non-restricted text processing appli- 
cations such as machine translation and critiqu- 
ing. 
We are concerned here with emulating, in 
some sense, the way a person uses a dictionary: 
look up one entry, study the defmitions and the 
examples, look up other entries, and so on. We 
feel that natural language itself can be a reason- 
able knowledge representation language. More 
needs to be learned about how to access and 
manipulate this knowledge; but the flexibility 
afforded by natural language is an advantage for 
the task, not a drawback. 
This research is related to other work being 
done with machine-readable dictionaries, e.g. 
Markowitz et al. 1986, in the sense that we all 
share the goal of automatically extracting se- 
mantic information from these rich sources. 
l lowever, in other respects our approaches are 
quite different. 
I. A'VI'ACIIMENT OF PREPOSITIONAL 
PIIRASF^~; 
The relationships in which we am interested can 
be illustraled by the following sentences from 
Binot Iq85: 
(!) I ate a fish with a fork. 
(2) l ate a fish with bones. 
(See Appendix A, Tree I.) In both cases, the 
ambiguity resides in tile placement of the "with" 
prepositional phrase, which might modify either 
"fish" or "ate'. The parse tree shows the PP at- 
tached to the closest possible head, "fish," with a 
question mark showing that it could alternatively 
be attached to the verb "ate ~. 
! The second author currently works for B.I.M., Belgium. 
152 
Focussing on (1), another way to phrase the 
key question is "Is it more likely that a fork is 
associated with a fish or with an act of eating?" 
To answer that question, the system evaluates 
separately the plausibility of the two proposed 
constructs: 
(la) eat with a fork 
(lb) a fish with a fork 
then orders the solutions, and picks the one with 
the highest rating. 
In the heuristics we are currently using, the 
basic way to rate the likelihood of a construct is 
to try to establish, through the dictionary, some 
relevant semantic connection between the words 
of that construct. Easier (or shorter) connections 
yield better ratings. Long connections, or con- 
nections making use of approximate inferences, 
will lead to lower ratings. For example, the de- 
finition of "fork" contains the phrase "used for 
taking up," and "eating" is defined as a kind of 
"taking" in the dictionary. By establishing these 
relationships, we see a plausible semantic con- 
nection between "fork" and "eat," and (la) re- 
ceives a high rating. 
The relationships are established (aS by iden- 
tifying equivalent function-word patterns in the 
definitions, such as the equivalence of "used for" 
and the instrumental "with'; (b) by linking im- 
portant definition words (i.e., central terms in 
definitional phrases, such as heads of phrases, or 
else synonyms). This is done by parsing the de- 
fruitions, identifying the central word(s), and then 
following hierarchical chains of definitions 
through the dictionary. 
Heuristic answers are expressed in terms of 
certainty factors which, as in the MYCIN system 
(Shortliffe 1976), take their values in the range 
(-1,+ 15: "-I" expresses absolute disbelief; "0" 
expresses complete uncertainty; "1" expresses ab- 
solute befief. Intermediate values express varying 
degrees of belief or disbelief. 
The two main heuristics that are used to 
evaluate the plausability of (la) against (Ib) can 
be described in English as follows: 
HI- for checking for an INSTRUMENT relation 
between a head and a "with" complement: 
I. if the head is not a verb, the relation 
doesn't hold (certainty factor = -15; 
2. if some "instrument pattern" (see be- 
low) exists in the dictionary def'mition 
of the complement, and if this pattern 
points to a defining term that can be 
linked with the head, then the relation 
probably holds (certainty factor = 0.7); 
3. else assume that there is more chance 
that the relation doesn't hold (certainty 
factor = -0.35. 
checking for a PARTOF relation be- 
tween a head and a "with" complement: 
I. if the head is not a noun, the relation 
doesn't hold (certainty factor = - 15; 
2. if some "part-of pattern" (see below) 
exists in the dictionary definition of the 
complement, and if this pattern points 
to a defining term that can be linked 
with the head, then the PARTOF re- 
lation probably holds (certainty factor 
= 0.7); 
3. else assume that there is more chance 
that the relation doesn't hold (certainty 
factor = -0.35. 
H2- for 
Each certainty factor refers to the specific 
proposition (or goal) to which the heuristic is 
applied. Thus, if clause 3 of heuristic 112 is used 
when applied to the proposition (lb), the result- 
ing certainty factor -0.3 will indicate a relatively 
moderate disbelief in this proposition, stemming 
from the fact that the system has not been able 
to find any positive evidence in the dictionary to 
sustain it. 
The above heuristics make use of the fact that 
there are specific words and/or phrases in dic- 
tionary definitions, forming patterrt~, which are 
almost systematically used to express specific se- 
mantic relations (Markowitz et ai. 19865. For the 
two relations considered here, some of these pat- 
terns are: 
INS'I'RUMI:.NT: for, used for, used to, a 
means for, etc. 
PARTOF: part of, arises from, end of, 
member of, etc. 
These patterns generally take, as their objects, 
some cen',rai term (or terms) in the definition of 
the complement word. We can then try to link 
that term with the head of the construct that is 
being studied. 
Focussing again on example sentence (15, the 
syslem starts by examining the first construct, 
153 
(la). It parses the definition of the complement 
"fork," and discovers at least one INSTRU- 
MENT pattern, "used for': 
fork: An implement with two or more pron~ 
used esp for taking up (as in eating), 
pitching or digging. 
Taking the conjunction into account, the 
system finds three possible terms: "taking up," 
"pitching," and "digging," which it tries to link 
with "eat." (For the present, we deliberately 
avoid the phrase "as in eating" -- which offers a 
direct match -- in order to show that our ap- 
proach does not rely on such lucky coincidences.) 
The system is able to establish that "eat" is a di- 
rect hyponym of "take" according to W7: 
eat: to take in through the mouth as food... 
to take food or a meal. 
The link is thus probably established, and the 
system moves on to consider (lb). Since no 
PARTOF pattern can be found in the definitions 
of "fork," this second possible construct will be 
ranked as much less likely -- (la) receives a cer- 
tainty factor of +0.7, but (Ib) gets a certainty 
factor of only -0.3. Therefore the system recom- 
mends attaching the PP to the main verb in (!). 
For sentence (2), the constructs to be com- 
pared are "eat with bones" and "a fish with 
bones." In the definition of "bone," no useful 
INSTRUMENT pattern is found; so "eat with 
bones" cannot be easily validated. But the first 
definition of "bone" gives the following 
PARTOF pattern: 
bone: One of the hard parts of the skeleton of a 
vertebrate. 
This yields two possible links for "fish': 
%keleton" and "vertebrate." "Fish" can be iden- 
tiffed as a direct hyponym of "vertebrate" ac- 
cording to W7. 
fish: Any of numerous cold-blooded strictly 
aquatic craniate vertebrates... 
Therefore, "a fish with bones" receives a higher 
certainty factor than "eat with bones," and the 
system recommends attaching the prepositional 
phrase to the direct object in sentence (2). 
The above examples are among the simplest. 
In more difficult cases, heuristics may perform 
various kinds of inferences in order to establish 
connections. It is also possible for several 
heuristics to be applied to a given construct, with 
their results then being combined. The cumula- 
tive effect of many heuristics, and not the per- 
fection of each one separately, does the job. 
The choice of certainty factors rests mainly 
on intuition. Some choices are easy; some infer- 
ences, for example, are obviously weaker than 
others. In other cases the values have to be ad- 
justed by trial and error, by processing many ex- 
amples. It is interesting to note that, as our 
corpus of examples increa~s, the certainty fac- 
tors are converging toward apparently stable val- 
ues. Our system currently includes about 20 
heuristic rules and is able to handle the prep- 
ositions "with," "by," "after," and "in." It has 
been tested successfully on about 50 examples so 
far. 
2. ATFACItMENT OF RELATIVE CLAUSES 
A typical problem in attaching relative clauses 
occurs when the clause is separated from the 
noun it modifies by a prepositional phrase: 
(3) I want the book by my uncle that is on the 
shelf. 
In (3), the relative clause "that is on the shelf" 
probably modiffcs "book" and not "uncle." A 
human reader assumes this because of knowing 
that a book is more likely to be on a shelf than 
an uncle is. I Iowcver, syntax alone cannot tell 
us so. A syntactic parser will normally produce 
a trec which shows the relative clau~ modifying 
the closest noun, namely "uncle. ~ (See Appendix 
A, Tree 2.) Note that the parser attaches the 
relative clause (RELCI.) node arbitrarily to the 
closest head noun "uncle," but marks the other 
possible attachment site ("book ~) with a question 
mark. The higher question mark in Tree 2 is for 
the PP attachment. 
The grammar that supports all ol" the parsing discussed here is the PI.NLP ~nglish Grammar (Jensen in preparation, 
Heidorn 1976). 
154 
We have implemented the solution to this 
kind of relative clause ambiguity. Our system 
starts by trying to solve the PP attachment 
problem: does "by my uncle" modify "book" or 
"want'?. Of all possible relationships between the 
various word pain, the AUTHOR relationship 
between "book" and "uncle" will receive by far 
the best ranking. This will happen because it can 
be established, by using the dictionary, that an 
uncle can be a human being (and thus able to 
author a work), and that a book is some kind of 
work. 
The processing of the RELCL attachment 
then begins. Syntax tells us that the relative 
pronoun "that" is the subject of the predicate "be 
on the shelf." One of the properties of the verb 
"to be" is that a prepositional complement qual- 
ifying this verb really qualifies the subject of the 
verb. Applied to Tree 2 of Appendix A, this " 
provides two possible interpretations: 
book on the shelf 
uncle on the shelf 
At this point we can ~ that the relative clause 
attachment in Tree 2 reduces to a prepositional 
phrase attachment, which can be solved easily by 
the PP attachment methods already described. 
Specifically, the dictionary defudtion for "shelf" 
will tell us that a shelf is "to hold objects" or "for 
placing things on," and the word "book" can be 
related to "object" or "thing" much more easily 
than the word "uncle" can be so related. This 
will lead to the preference for -book" as the 
antecedent of the relative clause. 
However, most relative clause attachment 
problems cannot be reduced to PP attachments. 
Consider (4): 
(4) l know the actor in the movie that you met 
last month. 
The parse tree for this sentence (Tree 3 of Ap- 
pendix A) shows question marks in the same 
positions as Tree 2. floweret, because of the 
syntactic structure of the RELCL in (4), we 
know that the relative pronoun this time refer~ 
to the obiect. ~ its main verb "met." Either 
"movie" or "actdr" must be the object of "met." 
No prepositional phrase is involved. 
Now we have to decide which is more likely: 
You met an actor. 
You met a movie. 
Although semantic codes are included in the on- 
line version of LDOCE (i.e., features like HU- 
MAN and ABSTRACT are marked on nouns, 
and subcategoriz~ation codes using these features 
are marked on verbs), the codes do not help with 
problems like this one. According to the 
LDOCE codes, possible objects for the simple 
transitive verb "meet," in its various sub-senses, 
are IIUMAN, ABSTRACT, and (moveable) 
SOLID. No ranking of likelihood or preference 
is given, and of course a syntactic parser would 
not know which sub-sense it is dealing with. 
"Actor" is marked + tlUMAN, and "movie" is 
marked + ABSTRACT. So either object noun 
is equally likely (Mary Neff, personal communi- 
cation). 
Although we have not yet implemented this, 
we believe that the same "approximate reasoning" 
that we implemented for PP-attachments will 
work here, too. The strategy is to formulate 
heuristics that yield "certainty factors," not cate- 
gorial acceptance or rejection of an interpreta- 
tion. These heuristics would propose a solution 
for the stated task by operating on the output of 
the syntactic parser. For the current example, 
the first step would be to parse the LDOCE entry 
for "meet" (shown in Figure I), looking for direct 
objects. 
meet' /mi:t/ p met /mat/ 1 \[TI.IO! to ~m8 
tolleth~ (with), by chancal o¢ anlapmeut: /dr# /or ~mm~Jt_...mtt h/m m the str,~ --coml~ 
WITH Z \[TI J to find o¢ ezponm1~; ~In 
: I met a lot of diff~dtwj m the m~k 3 \[ ! J\] to 
come together or clme: The cars a/man mn 
( --oBe front aSainslt me o01~), bwr dtt, w 
awayanddea~aw 4 \[TI .il)\] to |et to kJ~OW or I~ 
inttodu¢lxl (to) foe the first time: Come to thepm.ty 
and m~.t ~ mtc~rxtmg pcop@.j We mat at Anm'l 
py.n.y, a'~t w. ~q. i don't r,,mm~ ~ ,~ | 
\[lllJ to ,,ola at a rasltentn 8 point: .W~ skirt ~'t 
meet round m y mtdd/G • {\[OI to gather tolether: 
77tewholtschooimeytoheardjespeeck 7 \[TL;I#IIa 
touciL (at if) aetutafly: Them" li~z mtt (at a km).l 
Her#~tdraM,qilf~"emaviolomMow 811"1 (mal)J 
to amtwer, etp. i- oppomtion: Hit cha~eJ wertraa 
w,k cr~, of a.'~er.lAnlrry cry# mtt kU xpceck O 
\[TI \] to be them at the zmvsi of: rll ~ .vow o\[/ 
the tmm.lThe tart wdl n~tt the teamtwdl ~ you 
off" the team 10 \[TI\] to pay: Can you ~ 0m 
~? 11 \[TI\] to sattdy: Does this meet mr 
tmpes?lThit nm~ road meets a Ionf./elt need 12 
..k. ,.,a. m~t to u,,' one's (small amount o(1 
mon¢~ caR(lilly 50 a to all'oN Whlll nile flesh;Is 13 
mee¢ tmntmto's eye also look ~ Im dm eye-- 
to look dill~tly or Steadily &t 5omeone 14 
( lalto ~me~lBmlD ~ meetl Ule eye h,dden fa~u ar 
reas~ms (in oe foe somcthm8) ~ a.tso mini 
I.Od..rWAy 
Figure I. Text of LDOCE entry for the verb 
"meet" 
155 
The sub-definitions are no help, because no ob- 
jects are shown. But the example sentences in 
the entry are a rich source of information about 
typical usage. There are eleven different example 
object nouns: him, lot (of difficulties), people, 
face, speech, you (twice), train, amount, hopes, 
need. Over a third of them can be easily related 
to the word "actor': the word "people," and the 
three occurrences of personal pronouns. (The 
general rule here is that any personal pronoun 
except "it" can be substituted for any word that 
has "person" as the head of one of its def'mitions.) 
None of them can be so easily related to the word 
"movie.* Thus the system concludes that "actor" 
is a more likely object of the verb "meet" than is 
"movie." This conclusion is no accident; 
lexicographers are experts on words, and they 
have incorporated their expertise, in ways both 
obvious and subtle, into standard dictionaries. 
Mother interesting example of the relative 
clause attachment problem is found in the fol- 
lowing sentence from a large data base of busi- 
ness letters: 
(5) There are no agencies within the country 
which would loan money to individuals for es- 
tablishment of boarding homes. 
The choice here is between possible nouns to 
serve as the subject of the predicate "would loan 
money': 
Agencies would loan money. 
Country Would loan money. 
First, the LDOCE definition for "loan" refers 
us to the word "lend." Moving to the entry for 
"lend," we look for cited subiects. The example 
sentences, in this case, are no help: subject words 
arc either personal pronouns or the word "flags"; 
and none of these helps us to choose between 
"agencies" and "country." But one of the sub- 
definitions of "lend" is 
"to give out (money) for profit, esp. as a 
business'. 
The phrase "as ?a" is often used in definitions to 
signal the AGENT that does the action. Then 
we consult the dictionary to see which better 
qualifies as a "business': "agency" or "country." 
The answer comes easily; the first sub-definition 
of "agency" is 
"a bzL~iness that makes its money esp...". 
The two words "country" and "business" cannot 
be connected so easily as "agency" and "business" 
along any path of heuristic searching. Therefore 
we prefer to attach the relative clause to "agen- 
cies" rather than to "country." 
It is important to realize that none of the in- 
formation being cited here is manually coded; 
the English text of the LDOCE entries is being 
used. Our strategy can be considered to be 
making explicit a semantic network that exists 
implicitly in this text. The entry for -lend" shows 
"business" as an AGENT of "lending'; the entry 
for "agency" shows that "agency" is a kind of 
(ISA) "business." This implicit chunk of network 
is shown in Figure 2: 
Figure 2. A semantic path connecting lend" and 
"agency" in LDOCE 
3. RESOLUTION OF PRONOUN REFER- 
ENCE 
Problems of pronoun reference are many and 
varied, and not all of them will yield to this same 
method of solution (llobbs 1986, Sidner 1986). 
But for some, the information in dictionary defi- 
nitions can give important clues. Consider (6) 
and (7): 
(6) We bought the boys apples because they were 
hungry. 
(7) We bought the boys apples because they were 
cheap. 
In the absence of other information, human 
readers assume that "they" probably refers to the 
boys in (6) and to the apples in (7). The com- 
puter needs to follow some inference path that 
will lead to the same tentative assumptions. 
156 
For sentence (6), we need to choose a most 
likely subject noun for the predicate "be hungry" 
-- either: 
Boys were hungry. 
Apples were hungry. 
We would first parse the dictionary defmition for 
• hungry." In LDOCE, there are two example 
sentences with personal pronouns for subjects; 
the word "boys" can be quickly related to all 
personal pronouns. There are no example sen- 
tences with subjects that can be easily related to 
"apples." 
Additional support can be found in two di- 
rections. The first definition for "hungry" in 
LDOCE is "feeling or showing hunger." We 
want to find out what sort of entity can "be 
hungry," so we ask what sort of entity can "feel." 
Of about 30 example sentences for the verb 
"feel," 26 are personal pronouns (excluding "it'). 
Hence we prefer "boys" to "apples" as the subject 
of "be hungry." 
A second direction of search also reinforces 
this interpretation. "Hungry" is defined as "feel- 
ing or showing hunger," and "hunger" is defined 
as "the wish or need for food." Briefly summa- 
rized, we conclude that "food" is the object (or 
goal) of hunger, hence of being hungry. LDOCE 
also tells us that an "apple" is "a hard round fruit" 
and "fruit" is "used for food." ttence apples are 
(used for) food; hence apples can be the object 
of q~'ing hungry." Since the mggested object of 
"being hungry" is the same as the object of the 
main clause (see (6)), it stands to reason that 
"they" probably does no( also refer to "apples." 
The paths that we are tracing are delicate, but 
they exist. A computer program that follows 
these paths extracts, from existing text, some very 
interesting real-world relationships. 
In solving the pronoun reference task of sen- 
tence (7), the program must choose between: 
Boys were cheap. 
Apples were cheap. 
By following paths through the LDOCE entries, 
the conclusion that "apples were cheap* appears 
more likely than that "boys were cheap" (al- 
though the latter is certainly possible). 
157 
4. INTERPRETATION OF DANGLING 
MODIFIERS 
English teachers have long objected to a po- 
tential awkwardness and lack of clarity in con- 
structions with dangling modifiers: 
(8) (While) watching TV, the doorbell rang. 
In .sentences like (8), the attachment problem 
appears in a different guise. There is only one 
noun given for the participial to modify, and that 
is "doorbell." It is not possible to set up an ob- 
vious choice pair in the same manner as before. 
However, we do know that participial modifiers 
are a notorious source of confusion. So we can 
check the dictionary to find out how likely it is 
that a doorbell might watch "IV. 
In LDOCE, the sub-definitions for "watch" 
are no help. But the example sentences, once 
again, offer strong hints. There are 16 such ex- 
amples. Fifteen of them have personal pronouns 
as subjects for the verb "watch." The first ex- 
ample is "Do you often watch TV?" (This situ- 
ation was not contrived; sentence (6) was taken 
from a popular high school English grammar 
book, Warriner 1963, before the dictionary was 
consulted.) With this information in hand, we 
can say that "doorbell" is, at best, an unlikely 
subject for the verb "watch." 
5. CONCI.USIONS 
There are many important sources of infor- 
mation for natural language processing. Syntax, 
logical form, intersentential context, and 
presuppositions about the mental state of the 
speaker and of the intended audience (to name a 
few) all make their cortributions, and have all 
been discussed, to varying extents, in the litera- 
ture. Now it appears that the text portions of 
online dictionary entries can serve as a rich 
source of semantic information and world 
knowledge that can aid during the processing of 
other text. 
ACKNOWLEDGMENTS 
We would like to thank George Heidom for 
his helpful suggestions in the preparation of this 
paper, and Yael Ravin for her continuing insights 
into problems of ambiguity. 

Appendix A. Parse Trees 
DECL PRON* 
VERB* "ate" 
DET 
NOUN* 
? PP 
PUNC "• " 
iliil .ili t tlillil I I Iii I ili. I Ill I i Ili 
'tilt 
ADJ'* "a" 
"fish" 
PREP "with" 
DET ADJ* 
NOUN* "fork" 
ttat' 
Tree I. Parse tree for a syntactically ambiguous PP attachment 
DECL NP 
VERB* 
NP 
PUNC 
'ti" PRON* 
ttwant" 
DET 
NOUN* 
PP 
It t, 
ADJ* 
"book" 
PREP 
DET 
NOUN* 
~LCL 
"the" 
"by" 
ADJ'* 
"uncle" 
NP 
VERB* 
PP 
"my" 
PRON* "that" 
'tlSt' 
PREP "on" 
DET ADJ* 
NOUN* "shelf" 
"the" 
Tree 2. Syntactic parse showing rclativc clause attachment 
DECL NP 
VERB* 
NP 
IIIiIII IIt llilllilllill lli~ liil I I t I tl I i i I i I it i I i I Iil I t t III I I 
ttit' PRON* 
"know" 
DET 
NOUN* 
PP 
tt tt P~C 
ADJ* 
"actor" 
PREP 
DET 
NOUN* 
RELCL 
"the" 
"in" 
ADJ* "the" 
"movie" 
NP PRON* "that" 
NP PRON* "you" 
VERB* "met" 
NP AJP ADJ* 
NOUN* "month" 
"last" 
Tree 3. Another relative clause attachment 
159 

REFERENCES 

Binot, Jean-Louis. 1985. SABA: vers un systeme 
portable d' analyse du francais ecriL Ph.D. disser- 
tation, University of Liege, Liege, Belgium. 

Binot, Jean-Louis and Karen Jensen. 1987. "A 
semantic expert using an online standard dic- 
tionary." Proceedings oflJCAI-87, Milan, Italy, 
August 1987. 

Heidorn, George E. 1975. "Augmented phrase 
structure grammars" in Nash-Webber and 
Schank, eds., Theoretical Issues in Natural Lan- 
guage Processing. Association for Computational 
Linguistics. 

Hobbs, Jerry R. 1986. "Resolving Pronoun Ref- 
erence" in Grosz et al., eds., Readings in Natural 
Language Processing. Morgan Kaufmann Pub- 
fishers, Inc., Los Altos, CA. 

Jensen, Karen. In preparation. "PEG: A broad- 
coverage computational syntax of Engiish." IBM 
Research Report. 

Jensen, Karen and Jean-louis Binot. forthcoming. "Disambiguating prepositional phrase attachments by using on-line dictionary 
definitions." Computational Linguistics, special 
issue on the lexicon. 

Longman Dictionary of Contemporary English. 
1978o Longman Group Ltd., England. 

Markowitz, Judith, Thomas Ahlswede, and 
Martha Evens. 1986. "Semantically significant 
patterns in dictionary definitions." Proceedings 
of the 24th Annual Meeting of the ACL, 
Columbia University, June 1986. 

Shortliffe, E.II. 1976. Computer-based medical 
consultation: MYCIN. Artificial Intelligence Se- 
ries. Elsevier. 

Sidner, Candace L. 1986. "Focusing in the 
Comprehension of Definite Anaphora" in Grosz 
et al., eds., Readings in Natural Language Proc- 
essing. Morgan 'Kaufmann Publishers, Inc., Los 
Altos, CA. 

Warriner, John E. 1963. English Grammar and 
Composition: Complete Course. Harcourt, Brace 
& World, Inc., New York. 

Webster's Seventh New Collegiate Dictionary. 
1963. G. & C. Merriam Co., Springfield, Mass. 
