Coreference resolution in dialogues in English and Portuguese 
Marco Rocha 
Universidade Federal de Santa Catarma 
Departamento de Lingua e Literatura Vemficulas 
Centro de Comunicac,~o e Expressao 
88040-900 Florian6polis - Brazil 
marcor@cce.ufsc.br 
Abstract 
This paper introduces a methodology to 
analyse and resolve cases of coreference in 
dialogues in English and Portuguese. A 
four-attribute annotation to analyse cases of 
anaphora was used to analyse a sample of 
around three thousand cases in each 
language collected in dialogue corpora. The 
information thus gathered was analysed by 
means of exploratory and model-building 
statistical procedures. A probabilistic model 
was then built on the basis of aggregate 
combinations of categories across the four 
attributes. This model, in combination with 
direct observation of cases, was used to 
build an antecedentqikelihood theory, which 
is at present being organised as a decision 
tree for the purpose of testing with a view 
for automatic annotation and subsequent 
resolution of coreference cases in dialogues 
in both languages. It is thought that the 
findings could be extended to Spanish, 
Italian and possibly French. 
Introduction 
The problem of anaphora resolution has received 
a great deal of attention in theoretical linguistics, 
psycholinguistics and also in natural language 
processing. Perhaps as an inevitable 
consequence of such a large body of work 
related to the subject, the term anaphora has 
been used to mean a varying range of 
phenomena. 
Approaches that build on the concept of 
cohesion ties (Halliday and Hasan 1976) analyse 
anaphoric relations within a broad framework of 
discourse or textual cohesion. As a result, the 
notion of anaphora, which had been initially 
linked quite closely to the older concept of 
pronominalisation, has been expanded to include 
all referring expressions with some form of 
antecedent either explicitly introduced in the text 
or inferable from it. 
In an earlier study, Webber (1979) had already 
widened the scope of anaphoric relations, by 
including nonpronominal noun phrases which 
refer back to antecedents in the discourse; the 
so-called one-anaphora; and verb-phrase 
deletions. Gradually, the distinction between 
anaphoric and coreference relations became less 
and less relevant in approaches concerned with 
robust implementation of systems with a 
capacity for anaphora resolution. The present 
study follows the same sort of approach. 
Therefore, the term coreference in the present 
study is used to refer to all pronominal forms, 
anaphoric nonpronominal noun phrases, one 
anaphora, numerals when used as heads of noun 
phrases, prepositional phrases used as responses 
to questions or statements, responses to 
questions in general (including yes, no and short 
answers using auxiliaries), so anaphora, do- 
phrase anaphora and whatever other elements in 
dialogues were thought to be referring 
expressions with an identifiable antecedent. 
The next section describes the annotation 
scheme used to analyse the coreference cases. 
The third section presents the antecedent- 
likelihood (henceforth, AL) theory, which is the 
way information collected by means of the 
annotation was organised so as to be used to 
resolve new cases of coreference in other 
dialogues. The decision trees which are to be 
built on the basis of the AL theory are explained 
in the subsequent section, whereas the final 
53 
section concludes with a discussion of results 
obtained so far and an analysis of future 
developments. 
2 The annotation of coreference cases 
The operational routine of data collection was 
simply to search manually for tokens which 
coreferred in dialogue samples. Samples 
consisted of full dialogues for Portuguese but 
not for English, as a result of the sampling 
technique used in the English corpus (the 
London-Lund). Whenever a case of coreference 
was found, it was classified, according to four 
attributes, namely: type of anaphor; type of 
antecedent; topical role of the antecedent; and 
processing strategy. 
The first a~ibute refers to the word or phrase 
which triggers the anaphoric link, that is, the 
visible item that requires the retrieval of another 
element in the text for semantic interpretation. 
Concepts such as zero pronouns or empty 
categories are not used in the classification. 
Thus, the anaphor is invariably a phonetically 
realised item, and a verb without a phonetically 
realised subject is classified as an anaphoric 
verb. Although such verbal forms are rare in 
English, they are fairly common in spoken 
Portuguese. The same approach is used for 
transitive verbs without a phonetically realised 
object, which are also frequent. 
The type of antecedent concerns primarily the 
implicit/explicit dichotomy. Typically anaphoric 
words, such as it and that, may occur in 
nonreferential uses - for instance, the 'prop' it 
(Quirk et al. 1985). Thus, a third category, 
nonreferentiai, was used to classify these cases. 
Although these are not cases of coreference 
strictu sensu, it was thought important to include 
them, so that they could be identified when it 
came to implementation. Some tokens of 
pronouns with a vague antecedent identifiable 
by means of inference based on discourse 
information were classified as discourse 
implicit antecedents. 
The attribute named as the topical role of the 
antecedent classifies the antecedent of a given 
coreference case according to categories which 
assign a saliency status to discourse entities 
(typically noun phrases) in a dialogue. These 
categories include a discourse topic for the 
dialogue; a segment topic for every stretch of 
dialogue in which the topic is considered to he 
the same, according to specific procedures; a 
subsegment topic, if further division within a 
segment is needed for the appropriate modelling 
of topicality; and both global and local thematic 
elements, which are salient discourse entities 
related to the topics above mentioned. As 
antecedents may also be discourse chunks of 
varying length, these same categories were used 
to classify such antecedents as predicates of a 
given topical role thought to be the dominant 
entity within the discourse chunk. 
The aim of this attribute is to use the often 
mentioned relationship between topicality and 
coreference (see Grosz and Sidner 1986) for 
operational purposes. This classification does 
not claim to be the actual key for the modelling 
of topicality in dialogues from a 
psycholinguistic point of view. It does claim, 
however, to be a useful tool for the resolution of 
particularly hard cases of coreference, in which 
the antecedent is not the nearest syntactically 
appropriate candidate, as will be shown in 
section 3. The topical roles are assigned on the 
basis of frequency, distribution and order of 
appearance. This information is used in 
conjunction with an adaptation for dialogues of 
Hoey's method (Hoey 1991) to establish 
patterns of lexis. Procedures were thus defined 
for the assignment of the topical roles above 
mentioned to the various discourse entities in a 
dialogue. 
The fourth attribute is the processing strategy, 
which is an attempt to classify the resolution 
path according to informational demands seen as 
the most essential for the processing at hand. 
The processing strategy was included in the 
annotation scheme as a way of enriching the 
classification model, uncovering distinctions 
which, might remain unnoticed if only the type 
of anaphor were to be specified. The plain 
assignment of a type of anaphor based on word 
classes would ignore distinctions in the 
processing required for the resolution of 
anaphors of the same type. On the other hand, 
54 
subsuming processing information in the 
classification used for the type of anaphor would 
disrupt the intended link of the latter to 
phonetically realised forms in a strict way. 
The annotation is entered between brackets in 
the order previously presented, beginning with 
the type of anaphor and ending with the 
processing strategy. The code for each one &the 
properties is delimited by semicolons. An 
example of annotated text is shown below. 
(i) 
B: well I think probably 
what Captain Kay (ENP; 
ex 222; dthel; LR;) 
must have said was a will is 
legal if it's ~P; ex 224; 
dthel; FtC;) witnessed on the 
back of an envelope 
The first token of coreference is the anaphoric 
nonpronominal noun phrase Captain Kay, which 
has been previously introduced in the dialogue. 
The type of anaphor is classified as FNP, for 
full noun phrase; the next slot defines the type 
of antecedent as explicit (ex__) and assigns a 
number for the referent according to order of 
appearance in the dialogue (222). The topical 
role of the antecedent is considered to be of a 
discourse thematic element. This means, thus, 
that Captain Kay is a fairly frequent discourse 
entity not only in a specific stretch of discourse, 
but throughout the dialogue, being, therefore, 
closely associated to the discourse topic. As the 
reference to Captain Kay is identified by means 
of verbatim repetition of the noun form under 
which it appeared for the first time in the 
dialogue, the processing strategy is defined as 
lexical repetition (LR). 
The subsequent anaphoric it refers to the first 
syntactically appropriate candidate looking 
backwards. Having Hobbs' (1986) naive 
algorithm as a reference, a primary first- 
candidate processing strategy was established 
under the code FtC. An extension of this 
primary strategy is the first-candidate chain 
(FtCCh), for cases in which Hohbs' naive 
algorithm finds another anaphor for antecedent. 
This sort of chain is crucially important in 
dialogues, as demonstrated by Biber (1992). An 
example is given below 1. 
(2) 
B: and I went down this 
morning to talk to the 
American Embassy on the off 
chance that the State 
Department might be you know 
able to finance a bit of 
travelling in the States and 
they can't they've (SP; ex_13; 
st; FtCCh;) got priority on 
vice-chancellors and uh 
English schoolteachers 
The second token of they refers to the first 
one, which, eventually, links both anaphors 
to the referent State Department. The two 
first-candidate processing strategies, 
together with resolutions relying on 
syntactic parallelism, were grouped under 
the umbrella category named syntactic 
processes. 
As the analysis of anaphora cases found in 
the corpus proceeded, a number of other 
categories for the classification of the 
processing strategy came up. These 
included, for instance, coilocational 
knowledge (CK), for cases in which the 
basic information required for processing 
was thought to derive from the use of 
anaphors within crystallised phrases, such as 
that is to say. Example (3) is one of those 
cases. 
(3) 
B:the bibliography has gone 
about as far as I can take 
it on my own that (De; 
ex_lO; p_st; CK;) is to say 
er in order to complete it I 
will have to visit the major 
resources in the United 
States and uh several in 
Europe 
t Annotation for other cases of coreferencc is 
omitted. 
55 
By collecting these phrases in association 
with each type of anaphor, a collocation list 
of anaphoric terms was built for each one of 
the types, with a resolution procedure 
attached, which was designed on the basis of 
corpus data observation. This list was 
subsequently used as an ancillary routine in 
the AL theory, as will be shown later. 
Several forms of lexicai knowledge, 
assigned to cases in which the antecedents 
were identified chiefly by means of semantic 
information contained in the anaphor, were 
also identified, such as part-whole 
relationshps. In example (4), monies refers 
to finances by means of information 
conveyed by the lexieal semantics contained 
in the lexical item itself, but not by means of 
plain repetition. Thus, the classification used 
is lexical signalling (LS), one of the 
categories within the umbrella category 
iexical knowledge, along with lexical 
repetition. 
(4) 
B: 
A" 
B: 
A: 
and uh - you know my own 
personal finances are 
well sure 
it's just out 
but you have applied er 
for monies (FNP; im_12; st; 
LS;) I keep hearing 
wherever I go 
Finally, a category named as discourse 
knowledge was used to classify cases in 
which the resolution required full processing 
of combined bits of discourse information. 
These four broader categories, including the 
essentially syntactic information required for 
the first-candidate strategies, grouped more 
fine-grained subclassifications in all cases, 
except for collocational knowledge. Thus, 
the umbrella categories were used to 
perform a statistical analysis using the data 
collected by means of manual annotation. 
However, the more detailed classification 
was retained in the actual annotation of the 
sample. The same approach was used in the 
other attributes. 
Frequencies for each category were then 
used in three different statistical procedures: 
a chi-square test; a measure of association; 
and the model-building variety of loglinear 
analysis. Chi-square tests with the attributes 
considered two by two showed statistical 
significance in all measurements (p < 
0.00005) in both languages. The Goodman 
and Kruskal tau was used to measure 
association between attributes two by two. 
Association was shown to be high (over 
0.30) between the processing strategy and 
the other three attributes, but relatively low 
(under 0.30) between these three attributes 
measured two by two. The loglinear analysis 
revealed that interactions considering three 
of the attributes were significant whenever 
the processing strategy was one of the three. 
The opposite was true when it was not. 
These results were true for both languages 
with minor variations. 
The statistical analysis showed thus that the 
classification model was adequate to represent 
the anaphora world. Moreover, it became clear 
that the attribute named as processing strategy 
yielded the highest information gain, acting as a 
link between the type of anaphor and the other 
two attributes which classify the antecedent. 
Therefore, the type of anaphor in itself, which 
could be mapped from POS tags or, in some 
cases, skeleton parsing (see Mitkov 1997), only 
became truly useful information for the 
resolution of the anaphoric reference when 
associated to the definition of a processing 
strategy. This made of course psycholinguistic 
sense, as it is not difficult to infer from corpus 
data that the same anaphor (such as it or that) 
may appear in contexts that lead to distinct 
processing demands for their resolution. 
3 The antecedent-likelihood theory 
The AL theory is made up of a series of entries 
for each type of anaphor. Entries contain 
56 
instructions organised in an algorithm-like form 
to check the applicability of all possible 
processing strategies, relying on information 
taken from the training set. The initial 
information considered is the probability of 
occurrence for each processing strategy and the 
two other attributes. As a result, some categories 
included in the general classification model are 
never checked because there are no tokens in the 
training set associating them to the type of 
anaphor in question. The subject pronoun entry 
is shown below. 
Subject pronoun 
global probability = 0.247 
Category probabilities 
process, strat, type antec, topical role 
FtCCh = 0.458 explicit=0.886 dthel= 0.341 
FtC = 0.232 implicit=0.058 st= 0.188 
CK = 0.155 NR= 0.049 sst = 0.156 
DK = 0.090 dim=0.008 dt = 0.055 
SeRf = 0.047 thel = 0.110 
PI = 0.012 fdv = 0.056 
Dx ' = 0.003 p_st = 0.038 
SK = 0.001 p_sst= 0.024 
SetMb = 0.001 p_dthel=0.014 
SetCr = 0.001 p_dt = 0.005 
p_thel= 0.005 
sithel= 0.004 
The table with the category probabilities 2 
defines the likelihood of categories in the three 
other variables being assigned to tokens of the 
anaphor type described in the entry, having the 
total number of tokens for the type of anaphor - 
not the full sample - as a reference. The first 
column specifies the probabilities for the 
categories which define the processing strategy, 
while the second column shows the figures for 
the type of antecedent, and the third column lists 
the topical roles of antecedents with the 
respective numbers. In order to make the table 
visually compact, most of the categories are 
listed using the code specified for the annotation 
of the sample. 
2 Categories cannot be fully described in this paper 
for reasons of space. The essential features have been 
presented though. 
Some processing information can be directly 
derived from the table of category probabilities. 
Categories which are not listed in the columns of 
the variables they belong to were not used to 
classify any tokens of the anaphor type, and thus 
can be left out of the processing. This may 
mean, for instance, that the processing need not 
be concerned with implicit antecedents for a 
given type of anaphor, because there are no 
tokens classified as such. Another possibility is 
that no tokens have been classified as being 
processed on the basis of collocational 
knowledge, and thus there is no point in 
checking the collocation list in search of 
matches. 
The header in AL theory entries is followed by a 
set of instructions organised in algorithm-like 
form. These instructions rely on the taxonomy 
employed to analyse processing strategies. The 
choice is based both on the results of the 
loglinear analysis and on the nature of the 
variable, which is in fact a description of the 
way a given anaphor token is resolved. The 
typical instruction appears as check ps, ps being 
any category included in the list of possible 
classifications of processing strategy for the type 
of anaphor. This means that the processing 
towards resolution of an anaphor of the type 
described in the entry should check, at this point, 
whether the processing strategy specified is a 
possible way to identify the correct antecedent. 
The typical check ps instruction is usually 
followed by a set of attached probabilities 
specific to the processing strategy being 
checked. These probabilities concern categories 
in the remaining two variables. Other 
information, such as the probability of predicate 
topical roles, may be added whenever this is felt 
to be useful. The subsequent items in a typical 
check ps instruction are recognition and 
resolution path. The first item contains 
information about features of the token itself and 
the immediate context in which it occurs, based 
on the observation of corpus data. The purpose 
is to guide the processing in the attempt to 
recognise the need for a certain type of 
processing strategy in order to resolve the 
anaphoric reference. The second item contains 
57 
information related to the actual identification of 
the correct antecedent. 
The amount and complexity of information 
included in each one of the items varies with the 
type of anaphor and the processing strategy. In 
some cases, the recognition requires careful 
analysis, involving a number of details and 
check-ups. In other cases, recognising that a 
certain processing strategy is the adequate one is 
not as difficult as identifying the antecedent, as 
in some cases of discourse-implicit antecedents. 
The AL theory is built so as to permit the 
expansion or reduction of guidelines included as 
instructions or items within instructions. 
In case a given processing strategy presents 
sufficient diversity of recognition and/or 
resolution patterns, the instructions may be 
divided into subtypes of recognition and 
resolution. This approach to the form of entries 
applies generally but not always, that is, there 
may be check ps instructions which do not 
include one or more of the items described 
above. There may also be instructions which 
specify actions of an unique nature for the type 
of anaphor or processing strategy under scrutiny. 
The extract of the subject pronoun entry shown 
below illustrates this flexibility. The header 
shown above is followed by two instructions 
which break with the general check ps norm, 
only to return to it in the third instruction, as 
shown below. 
check ifPOS tag is Q-tag item 
)~ if not, go to instruction 2; if yes 
go to tag-question entry in collocation 
list 
)~ follow resolution path in entry 
identify pronoun 
pronoun is he, she or they 
)" go to instruction 5 
pronoun is it 
)" go to instruction 4 
pronoun is first or second person 
go to instruction 3 
check secondary reference 
• attached probabilities 
• type of antecedent 
• explicit = 0.889 
• implicit = 0.111 
implicit antecedents are in a chain 
ultimate resolution by shared knowledge 
topical roles 
• dthel = 0.750 
• st = 0.250 
recognition 
• separate from endophoric usage 
• previous move 
• verbs say; ask; answer; explain 
• subject a third person pronoun or 
personal name 
• simultaneous tense and 
shift between utterances 
• if it is a second person pronoun 
• check identifying vocative 
utterance 
resolution path 
person 
in the 
select first human candidate searching 
backwards 
• check lexical clues 
if there is an identifying vocative 
• select it as the antecedent 
verbs say; ask; answer; explain 
• subject a third person pronoun or 
personal name 
• simultaneous tense and person shift 
between utterances 
• check collocation list 
The AL theory was manually tested on a 
previously analysed dialogue used as a test bed. 
There were 804 cases of anaphora in the testing 
set for English. The AL theory predicted the 
correct antecedent in 98.4% of the cases, which 
is evidently a satisfactory result. Results were 
also satisfactory, although not quite as good 
(93.5%), for Portuguese. However, the score 
was only obtained on the assumption that the 
dialogue had been POS-tagged, parsed and 
segmented according to topicality, using the 
procedures defined for each category in the 
attribute named as topical role of the 
antecedent. These are not minor assumptions, 
particularly if it is taken into account that, in 
real-life processing situations, these tasks would 
58 
have to be carried out during an ongoing 
conversation. 
Nevertheless, the approach seems worth 
pursuing as a promising way to solve a difficult 
problem in the actual implementation of 
dialogue interfaces and in NLP in general. Thus, 
the attempt to transform the AL theory into an 
automatic procedure may be a useful way 
forward. 
4 The decision trees for coreference 
resolution 
The general procedure for the resolution of any 
anaphora case is then to cheek the processing 
strategy with the highest probability first. If 
anaphors classified as determinative 
possessives in the English sample are taken as 
an example, this strategy would be the one 
named as first-candidate chain, in which the 
first appropriate candidate - in syntactic terms - 
searching backwards is selected, although it is 
also an anaphor. It may be safely assumed that 
this anaphor has already been dealt with, as it 
precedes the one being resolved. 
Checking a processing strategy for adequacy 
involves a recognition procedure specified in the 
entry, which, in the example considered above, 
would be to check the appropriateness of the 
first candidate. However, the probabilities 
indicate that there were cases in the training set 
in which this type of anaphor was resolved by 
means of discourse knowledge. This means that 
there were tokens in which the use of syntactic 
information only - as in Hobbs' "naive" 
algorithm - would lead to the identification of 
an incorrect antecedent. 
Therefore, ways of checking whether the first 
appropriate candidate is actually the correct 
antecedent had to be devised. Two basic routines 
were used: selectional restrictions and 
association history. As formalised in Katz and 
Fodor (1963), selectional restrictions are 
semantic constraints which the sense of a given 
word imposes on those syntactically related to it. 
Thus, whenever an anaphor is linked to a verb as 
a complement, it is useful to check if a candidate 
antecedent is a good fit by using selectional 
restrictions. 
There were cases in the training set, however, in 
which selectional restrictions would not detect 
the incorrectness of a syntactically appropriate 
candidate. A second kind of lexical clue was 
then included as a checking routine: the 
association history. It is unusual - although not 
impossible of course - that pronoun reference is 
used in the first instance of an association 
between a verb and a referent. This is even less 
likely in situations in which there is an 
established competitor with a record of tokens 
repeatedly associated to the verb in question. 
These checking routines may signal that it is 
advisable to consider bypassing the first 
candidate on the basis of discourse information. 
Checking the possibility of a resolution by 
means of discourse knowledge usually involves 
a recognition procedure, which relies on 
topica!ity information. If the alternative 
candidate selected is one of the highly salient 
discourse entities, the chances that the speaker 
felt the listener would successfully process the 
reference are much higher, making the bypass of 
the first candidate far more likely. 
The entry for determinative possessives is a 
relatively simple one, however, if compared to 
those for subject pronouns or anaphorie 
demonstratives in English or anaphoric verbs 
in Portuguese. Moreover, entries for other types 
of anaphor may require various forms of 
checking routines, which are specific to the type 
of anaphor in question. In spite of this highly 
complex and broad set of required information, 
it seems possible to organise it into decision 
trees for operational use. The notion of decision 
tree (as in Quinlan 1993) may have to be 
somewhat expanded in order to accommodate 
the various bits of specific information related to 
each type of anaphor. 
At present, several different algorithms and 
adaptations of these algorithms are being tested 
in order to establish their adequacy to the task, 
including the well-known C4.5. A hybrid 
approach, in which an example-based altemative 
process would choose the most closely related 
59 
case in the training set and use it to resolve a 
new case of anaphora, is also being considered, 
having the TiMBL package (Tilbury 1999) as a 
primary reference. It is expected that initial tests 
will be run soon, yielding results which will be 
then used to gradually improve the approach and 
its implementation. The GATE structure 
(Cunnigham et al. 1995) is likely to be used as a 
way to organise the various required elements of 
linguistic information as an integrated system. 
At the present stage, however, the software 
mentioned are quoted as reference rather than 
firm choice. 
5. Conclusion and future developments 
The process of building solutions for natural 
language processing on the basis of corpus 
information may rely simply on a classification 
model of any kind that would enable decision 
trees to be created inductively. However, the 
direct observation of corpus tokens allows the 
sort of refinement that may prove crucial for the 
actual operational success of the model in real- 
life processing situations. The approach 
described in this paper is an attempt to find an 
appropriate balance between the practicality of 
automatically inducing decision trees out of a 
training set and the thoroughness that the 
contrastive analysis of the various cases in the 
corpus is likely to accomplish. 
The systematisation of observed regularities in 
combination with statistical evidence proved 
very successful in dealing with the testing set of 
cases previously analysed for the purpose. It is 
also true, nevertheless, that the complexity 
introduced by the inclusion of a large amount of 
information to be taken into account during the 
processing makes actual implementation 
extremely hard. Therefore, the high score of the 
manual test must be seen cautiously. Future 
developments of the. approach described in the 
present paper aim at testing the actual gain of 
dealing with a thorough account of anaphoric 
relations in dialogues as compared to the 
increased difficulty of implementation, of which 
the inclusion of topicality and segmentation in 
the model are obvious examples. It is expected 
that the above mentioned balance will be 
eventually reached, preserving the satisfactory 
results to an extent that offsets the undesirable 
processing complexity. 

References 
Biber, D. (1992)Using computer-based text to 
analyse the referential strategies of spoken and 
written texts. In: Jan Svartvik, editor, Directions in 
corpus linguistics, pages 215-252, Mouton de 
Gruyter, Berlin. 
Cunningham, H., Gaizauskas, R. and Wilks, Y. 
(1995). A General Architecture for Text Engineering 
(GATE) - a new approach to Language Engineering 
R & D. Department of Computer Science, University 
of Sheffield, CS-95-21. 
Daelemans, W., Zavrel, J., van der Sleet, K., and van 
den Bosch, K. (1999). TiMBL: TilburgMemory 
Based Learner, version 2.0, Reference Guide. ILK 
Technical Report 99-01. Tilburg University. 
Grosz, B. and Sidner, C. (1986). Attentions, 
intentions and the structure of discourse. 
Computational Linguistics, 12:175-204. 
HaUiday, M.K. and Hasan, R. (1976). Cohesion in 
English. Longman, London. 
Hobbs, J. (1986). Resolving pronoun references. In 
B.L. Webber, Barbara Grosz, and K. Jones, editors, 
Readings in natural language processing. Morgan 
Kaufmann, Pale Alto, CA. 
Hoey, M. (1991). Pattexns of lexis in text. Oxford 
University Press, Oxford. 
Katz, J.J. and Fodor, J.A. (1963). The structure of a 
semantic theory. Language, 39:170-210. 
Mitkov, R. (1997) How far are we from (semi-) 
automatic annotation of anaphoric links in corpora ? 
In: Ruslan Mitkov and Branimir Boguraev, editors, 
Workshop on operational factors in practical, robust 
anaphora resolution for unrestricted texts, pages 
110-115, Madrid, July 1997, ACL/EACL. 
Quinlan, JR (1987). C4. 5 : Programs for Machine 
Learning. Morgan Kaufmaun, Palo Alto, CA. 
Quirk, R., Greenbaum, S., Svarvik, J. and I_~ech, G. 
(1985) A comprehensive grammar of the English 
language. Longman, London. 
Webbcr, B.L. (1979). A formal approach to discourse 
anaphora. Academic Press, New York. 
