On Representing the Temporal Structure of a Natural Language Text* 
Kurt Eberle 
Institut fiir Maschinelle Sprachverarbeitung 
Universit£t Stuttgart 
Azenbergstrafle 12 
7000 Stuttgart 
Germany 
Net address: kurt@adler.philosophie.uni-stuttgart.de 
Abstract 
A proposal to deal with tenses in the framework of 
Discourse Representation Theory is presented, ms it 
has been implemented for a fragment at the IMS for 
the project LILOG. It is based on the theory of tenses 
of H. Kamp and Ch. Rohrer. The system uses tile 
tense and aspect information, the information about 
the temporal discourse structure of tile preceding text 
stored in a specific list of possible reference times, and 
background knowledge. These types of information 
interact in order to choose a suited temporal anchor 
for the event of a new sentence. 
With respect to extended texts, choosing the right 
reference time for a new event is a problem which has 
been largely neglected in the literature. 
1 Introduction 
In Reichenbach (1947) the tenses are analyzed into 
different orderlngs of the relevant temporal parame- 
ters which are: the time of the event talked about, 
the speech time (time of utterance) and the reference 
time. Simple past, for instance, identifies event- and 
reference-time, which are both before the speech time. 
In the case of past perfect the event time precedes the 
reference time which precedes the speech time. 
Kanlp attd Rohrer (1983,1985) were led by text- 
phenomena to split up the Reichenbachian reference 
time into two contextually given coordinates: the ref- 
erence time (in the narrow sense) and the temporal 
perspective time. 
(1) Harry handed the book about giraffes to Jane (el). 
Two hours ago he had descovered it in Alfred's book- 
shop (e2). He had bought it (es), had inserted a dedi- 
cation (e4) and had it wrapped up (e~). Jane took the 
book with a smile (e~) 
In (1), the events e2 - e5 are ordered consecutively, 
*This research has been done in the context of LILOG. 
Thanks on this way to the members of the LILOG-researeh 
group of IBM for helpfull comments and especially for having 
made the integration of the suggested system in the LILOG- 
prototype possible. 
each taking its predecessor as reference time, with 
the exception that the context does not provide au 
event which precedes e2 and which, therefore, could 
play the role of reference time with respect to e2. 
Tile whole sequence e2 - es is seen from a vantage 
point in the past, el, which, therefore, plays the role 
of perspective time with regard to the structural inte- 
gration of e2 - c5. The perspective from which e 1 and 
es are seen is the speech tinle, e t serves as reference 
time for c~. In the approach of Kamp and Rohrer the 
role that the reference time plays in Reichenbach's 
system is taken over by the perspective time with re- 
spect to the meaning of the tenses in single sentence 
discourses: The meaning (one of tile meanings) of past 
perfect is that the event tittle precedes the perspec- 
tive time which precedes the speech time. The mean- 
ing of simple past is that the event time precedes 
the speech time which coincides with tile perspec- 
tive time. However, in addition to a system like Re- 
ichenbach's there is another parameter which comes 
exclusively from the text, the reference time proper, 
which is part of tile impact of the tenses on the tem- 
poral reconstruction of a narrative, titus completing 
the sentential meaning of the tenses by a textual as- 
pect. Here, the idea, that we take over from Kamp 
and P~ohrer, is that in a sequence of simple past sen- 
tences introducing events (not states) normally the 
event of a new senteuce follows the reference time 
which is provided by the event of the preceding sen- 
tence within the sequence. By transposition the same 
should hold for sequences of past perfect sentences. 
Here, the perspective is shifted from the speech time 
to a (contextually given) point in the past. There is 
an additional feature however with respect to tem- 
poral parameters. Adverbials can provide a location 
time for events and states. For instance, in (1), the 1o. 
cation time of e2 is the time pointed to by two hours 
ago. 
On the basis of the Reichenbachian approach us- 
ing three temporal parameters the problem of inte- 
grating a new event in the event structure of tile 
preceding text has been attacked by Hinrichs (1986), 
Partee(1984) and others. More independant of this 
approach are suggestions made by Moens and Steed- 
malt (1988} aatd by Webber(1988). 
AcrEs DE COLING-92, NANTES, 23-28 nOt\]T 1992 2 8 8 PRO<:. OF COLING-92, NANTES, AU6.23-28, 1992 
Since we tllink tltat it correct anMysis of narra- 
tives eamlot do without a fine grained distinction be- 
tween the roles played by temporal anchors, our ap- 
lu'oach uses the coordinates suggested by Kamp and 
Ptnllrer. Such coordinates arc stored in a structured 
list., called T-list, which rettects the discourse struc- 
ture of the preceding text. The l)rol)lem of choosing 
tim right anchor for the new event froln the preceding 
text is largely neglected in thc literature. Very often 
there, tile preceding text consists of only one sentence 
or tile problem is restricted to thc intra-sentential one 
presented by temporal conjunctions. How do we de- 
cide in (1) that the reference time for ca is not tile ear- 
lier introduced el, but e2 and that the reference tinle 
for e(t is not the most recent reference time es, but 
el': hi our system we use tile structure of the T-list 
aluong other things to get a hierachy of the topicality 
of tile rcfi~'renee times. This is very ch)se to tim idea 
of accessibility in Discourse Rel)resentatiou Theory 
(Kamp (1981)) and to tile notion of focus in the work 
of Grosz and Sidner (1985). Ill or(ler to tackle tile task 
of event-integration, which we call "temporM resolu- 
tion" - slaking Mhtsion to the very similar probteln 
of the NP-resohltion, where anaphoric links for the 
c~me of nominals have to be established - we struct~.lrc 
the T-list along the lines of different discourse rcla- 
tions such ms continuation, elaboration, flashback etc. 
Fla-sllback, for instance, holds in the example above 
between e 1 alld e 2 and between el alld e3. Tile re- 
lation between e2 and c3 is contimmtion. As we will 
set' ill the section after next, often, fi'om the purely 
linguistic l)oint of view timre are alternatives with 
resl)ect to the decision about the discourse relation 
which directs the iutegration of a new event or state. 
Ill our system, the criteria to decide about the 
relevant discourse relation are provided by tile tense 
fi)rlus of the reference event(s) and of the new event 
respectively, by tile existence aairl the elmracteristics 
of temporal adverbiMs, by the ,aspect of the events - is 
it all event proper or a state - and by internal se.luantic 
properties of the events whicil are made explicit by 
means t)f rules stored ill the background knowledge. 
Of course, in addition, tile hierarchically structured 
T-list inflnences tile cltoice and helps to update the 
system correctly, where updating means adding a new 
telnporal condition to the representation of the text 
and a(lding a new refereltee time to tim T-list at, a 
specific l)htee. 
In tlm next section we will briefly introduce tile 
kind of discourse representation structures that we 
use. Then we wilt say something about the discourse 
relations deidt with and we l)resent tile syntax of tile 
T-list. After tltat we sketch the relevant background 
knowledge and finally, we give some of the resolution 
rules of tilt' system. 
The structuring of tile T-list extents a suggestion 
of Eberle and Kasper (1989) and comprises elenlents 
of a proposal by Reyle (1986). 
Tile algorithm suggested is part of the LILOG- 
prototype witich is used for text understanding. In 
Affl\]~s DE COLING-92, NANTES, 23-28 AOrdr 1992 
particular it uses kaowledge provided by tile knowl- 
edge base and the inference component of the proto- 
type. 
2 DR;F and Events 
Witidn tile framework of Discourse Representation 
Theory (DRT) (Kamp(1981)) a Discourse Represen- 
tation Structure (D1LS) is a pair < U,K > consisting 
of a set U of discourse referents (DRFs) and a met 
K of conditions. DRSs are interpreted modeltheoret- 
ically. DRT uses a variant of tim Davidsonian method 
of talking about events: timy are treated as a kind of 
objects. 
In LILOG we use DRSs for text representiLtions. 
But we deviate front the cla.~sicM Dl~F-style in us- 
ing one-placc event predicates and thematic roles in- 
stead of n-place ew;nt predicates. A second difference 
is given \]W the fact that in L1LOG DllSs come with 
all indcx. The index is the list of reference tinles avail- 
able with respect to the DRS ill question. For a simple 
sentence it will bc a list consisting only of one element 
- the DRF of the event introduced - for a text it will 
lie a COmlllcx list, tit(: T-list. 
So, for instance, the LIL()G-DILS for cominy o\]x 
can be illustrated i~s follows: 
a(le~tt(e) = x 
For the \])ITS to be valid there must exist an em- 
bedding function which nmps e onto an event of the 
model structure such that e satisfies tile conditions as 
described ill the I)RS. 
3 Discourse Relations 
It is clear that the idea about tile order of events 
introduced by sequences of sentences that wc have 
sketched in the introduction is an oversilnpfification. 
Iu the folh)wing we will restrict ourselves to the 
tenses prevailing in narrativcs, silnple past and past 
perfect, and to only some of the text pllenomena re- 
lated to these tenses (as we have done ill tile LILOG 
ilnl)lelnentation with respect to tile German counter- 
parts of these tenses). 
}3csi(les tile disl:ourse relations used ill CXalnple 
(1), i.e. tl~shback and continuation, there are at least 
two nlore, which we call elabo~vltlon and backgTound. 
Conlpare the two fi)\]lowing exalllples: 
(2) Mouday Hans went to Par~s. Tuesday he met 
Gabt. 
(3) Monday Hans went to Paris. At the border he had 
some trouble. 
(2) dcscril)es a continuation. This can lie deduced 
from the characteristics of the location times. 
2 8 9 PRec. oY COLING-92, NANTES, AUG. 23-28, 1992 
In contrast to that, (3) describes an elaboration. 
We call the relation between a new event e2 and its 
reference event el an elaboration, if e2 describes et 
on a nmre fine-grained level (which gives rise to the 
temporal condition of inclusion (C_) between the new 
event and tile reference event within the representa- 
tion of the text). 
In (3) the relation of elaboration can be deduced 
only if there is some input from the knowledge base 
equipping the resolution component with facts about 
journeys to foreign countries. Notice that, here, the 
temporal inclusion is due to the hommnorphic local 
inclusion of tile border into the path of the travel. 
This illustrates that an intelligent temporal resolutimt 
component cannot do without having access to local 
reasoning. 
Now compare the following pair of examples, 
taken from Dowty (1986) and from Hinrichs (1986) 
respectively: 
(4) John entered the president's office. The clock on 
the wall ticked loudly. 
(5) Jameson entered the room, shut the door carefully 
and switched off the hght. It was pitch-dark around 
hint because the venetian blinds were closed. 
hi both (4) and (5) the last sentence introduces not 
all event proper, but a state. As widely accepted, the 
normal case of relating a new state e2 to an exist- 
ing reference event el seems to be that of inelusiml 
(e, C_ e2). l 
However, as exaalples like (5) make clear, tills de- 
fault can be overwritten. Here ton, it seems that back- 
grmmd knowledge gives the decisive indication about 
when the default has to be upset. 
We call the discourse relation of (4) background 
and that of (5) causally introduced background. For 
cases like (5) we introduce tile new state e~ a.s fol- 
lowing the actual reference event el (el -< e2) and 
as including a dummy reference time t (t C e:) with 
respect to which e2 introduces a backgrouud, ht the 
filrther course of processing the text, t has to be in- 
stautiated by a new event which is understood as fol- 
lowing el. 
The T-list makes use of these discourse relations 
a-s structuring elements. For instance, processing the 
samlfle text (6), which integrates tile text phenomena 
reflected by the examples (1) - (4), should result in a 
representation which has as index the T-list (7): 
{6) John left Paris by the Porte de 5 ~ Cloud late in 
the evening (el). He went to Frankfort (e2). 
At the border he was stopped (e3). It was pretty dark 
(eJ. Only a .few lights were burning (es). He had to 
open hz.~ boot (es). In Paris he had bought five cartons 
of e:garettes (e7) and had hidden them in the car (es). 
The customs officers missed them (eg}. 
IThis parallels, for instance, the insights of Partee(1984), 
H inrichs( 1986l, Kamp and Rohrer(1985) and others. 
Early in the morning John entered a bar near the 
mess tower in Frankfort (elo), ordered a beer (el1) 
and lighted a cigarette (el2). He was happy (eta). 
(7) 
el .... s2 .... el0 -- ell -- el2 
/\ /\ 
/ el \ / bg \ 
/ \ / \ 
/ \ e13 
e3 ....... e6 .... e9 
/\ /\ 
/ bg \ / :fb \ 
e4 , aS e7 , e8 
Ill (7) the structure of (6) is stored in the following 
way: el, c2, sis, el 1, el2 (the leaving of Paris, the going 
to Frankfort, the entering tile bar, the ordering of beer 
and the lighting the cigarette) on tile one hand and 
ea,eo, e0 (tile stoppiug, the opening of the boot, the 
failure of the officers) on the other form continuation 
chains (tied up in the picture by -- ). The second 
of these chains elaborates an event of the first, e2, 
tile transition to Prankfort (el/\). e4 and e~ serve as 
background to ea, cla is the background of e12 (bg/\). 
Finally, the continuation chain c7 - ea is a flashback 
seen from the perspective es. 
We define tile accessibility of all event as reference 
event in such a way that, having processed the whole 
text, the only accessible event will be elz, whereas, in 
the case of having processed the text till the integra- 
tion of es, the accessible events will be es, es and e2 
(with decreasing prominettce). In this situation es is 
the so called actual reference event. When integrat- 
ing a new sentence, one will try to expand the node 
represented by the actual referent first. But, finally, 
what level has to be chosen and what kind of anchor- 
ing has to be stored in the next step is decided by 
the system rules. So, for instance, in the case of (6), 
by purely linguistic reasons, ea cannot be related to 
tile actual reference event e8, because there is a tense 
switch from past perfect to simple past between es 
and es which indicates the termination of the flash- 
back. This results in tile re-initialization of the level 
ttle flashback started from as the level providing the 
new actual referent. In the case of (6) this is cs which 
is a member of an elaboration. Actual elaborations 
may be terminated by knowledge indicating that ex- 
panding the structure at the actual level is not plau- 
sible or even not possible. In this case the system 
will try to relate the new event with respect to tile 
(accessible) events of higher levels of the elaboration 
hierarchy. Since, with respect to (6), eg is compatible 
with e2 the system has no reason to terminate the 
level marked by e0. We think that the data confirm 
the concept of a hierarchy of reference events as used 
in the system. The exact defiuition of the accessible 
reference events of a T-list resides in the definition of 
the specific access-functions for T-lists that we sketch 
ACTE.S DE COLING-92, NANTES, 23-28 AOt~r 1992 2 9 0 PRec. OF COLING-92, NANTES, AUG. 23-28, 1992 
in section 5. 2 
In fact, ill tile LILOG implementation, tile items 
of the index, i.e. of the. T-list, do not consist of the 
pure events. They are terms with different slots filled 
hy the relevant temporal information such ms tile 
tense form which introduces the corresponding event, 
the ,xspect, the actual speech tittle etc. Skiplfing such 
technical details, the simplified syntax of the T-list is 
the following: 
T-list := \[\] ; 
\[E(D-list)JT-List\] 
where 
E(D-List): list item 
with 
E: the discourse referent for the event, and 
D-List: tile list of the items depending on the 
event, 
where 
D-List := \[\]; 
\[bg(\[E(\[\])lT'-List\])lD-List\]; 
\[fl)(\[E(D-List)lW-List\]) \[D-List\]; 
\[el(\[E(D-List )lT-ListD ID-List\]; 
Here, of course, embeddings which are given by the di- 
mensions by, fb, elin turn stand for background, flash- 
back and elaboration. T'-List is a T-list where each 
D-list is the empty list. We use T'-List for the items ill 
bg-lists, because we do Ilot think that a background 
give rise to further embeddings. The neighborhood of 
T-list items of the same level signifies continuation, 
except the case where such a list is fl'onted by bg. In 
this case, in the corresponding DtLS the elements of 
the list include the same event and, therefore, overlap 
each other. Continuation is reflected in the reversed 
order, because we use T-list as a stack. For (6), ac- 
cording to (7), we write (8). Here mid in the following 
we abbreviate E(\[\]) by E: 
(8) \[el2(Ibg(\[elaD\]), 
e11, 
elO, 
e2(\[el(\[e9,e6(\[fb(\[eS,eT\])l),e3(\[bg(\[e5,e4l)\])\]), ell 
Of course, there are other discourse relations ~ 
those mentioned. Especially, if one tries to analyse 
the structure of texts on a more fine-grained level 
than the purely temporal one, the relations described 
are not sufficient, others like explanation, contrast etc. 
are needed, s For lack of space, here, we can nothing 
say about these. The relations described are those 
2Anchoring new evetlts to non-accessible reference events 
clearly is possible, provided there is additional information to 
motivate this clmice, such as, for instance, definite descrip- 
tions using event nomlnalizations in order to establish cc- 
referentiallty, But, at present, such possibilities are not im- 
plemented. 
3 Partly for other purposes discourse or rhetorical relations 
are tlsed, for instance~ in Grosz and Sidner(1985}, Thompson 
and Mann (1987}. 
dealt with ill tile LILOG system. 
4 Background Knowledge 
The LILOG knowledge base provides us with a sort 
hierarchy for discourse referents which is related 
with respect to events to tile classification given in 
Vendler(1967). This sort hierarchy is used as one of 
several information packages within the temporal res- 
olution. With respect to the hierarchy we mainly use 
calendar knowledge and incompatibilty-statements 
about event types. For instance, all event introduced 
with the location time 1987 (or with subintervals of 
1987) cannot overlap with an event with location time 
1988. Exactly the same is true when the new event has 
an event type which is incompatible with the type of 
the reference event. A SWIMMING-event is incom- 
patible with a WRITING-event if tile relevant the- 
rustic roles are the same, i.e., here, the agent-role. Of 
course, when fornlulating such incompatibilities one 
tries to make use of the sort hierarchy. One will define 
the incompatibility for very general sorts (if possible) 
so that it is inherited by subsorts. 
Knowledge of this kind is hard information so to 
speak. In addition to this, the system uses default 
knowledge about event types which is similar to what 
is cldled event f~nmes (cf. Miusky (1975), Schank and 
Abelson (1977), Bartsch (1987) and others). Typical 
(simplified) examples of the two kinds of knowledge 
are (9) and (10): 
(9) VEI C MOVE, L E LOC 
(L c path(El) -~ 3E2 e PLACED-AT 
(theme(El) = theme(E.~) ^ E~ C_ E1 )) 
(10) VE~ E TURN, E.2 e BEING-LIGHTED 
(object(El) E L1GHT-SWITCIt A E~meetsE1 --*o 
3E3 e BEING-DARK (ElmeetsE:l)) 
(9) regulates the localization of the theme of a move- 
meat. From this rule we infer the existence of dif- 
ferent PLACED-AT-states for the thente within the 
movement. (10) predicts that normally (D fl)r de- 
fault) turning a switch when it is lighted results in 
darkness. 
5 Temporal Resolution 
The construction of tile seulantic represeutation for a 
discourse proceeds in several stages. I shall skip here 
the steps leading to the DRS of a sentence. But it has 
to be stressed that tile result of sentence processing is 
not a completed DRS. There may be variables which 
have to be instantiated in the process of integrating 
the sentence-DRS into the text-DRS. 
The sentence-DRS is understood as the input of 
the aspect-calculus of the prototype which has to 
contpute the aspect of the events of the DRS. There- 
fore it uses tile inform~ttion provided by the thematic 
roles. The starting point is the aspect-entry for the 
ACTES DE COLING-92, NANTES, 23-28 ^OI3T 1992 2 9 1 PROC. OV COLING-92, NA~CrES, AUO. 23-28, 1992 
verb in the lexicon. It can be overwritten by specific 
role information and morphological information with 
regard to aspect. For the sake of simplicity in the fol- 
lowing we assume that there are just two contrasting 
aspects corresponding to the main sorts of the upper 
structure of the event sort hierarchy, the heteroge- 
neous events proper (ev) and the honmgeneous states 
(st). The aspect calculus used is based on ideas fi'om 
Kritka (1987)2So, for Peter wrote a book we will get 
the aspect value ev, whereas for Peter wrote books we 
will get st. For the example (6) we get that e4, es, el3 
are states, the rest will be analyzed as events proper. 
After tile computation of the aspect we come up with 
an indexed DRS its descrihed iu the second and third 
section with the value of the aspect feature instanti- 
ated. 
Within the process of anchoring we use several 
access-functions with respect to the T-list. Among 
other things these are act-tel(T-list) which gives us 
the most topicalized item in the list, the actual 
referent, el-embedder( T-list, E), fb-embedder( T-list,E) 
which givo us the points elaborations and flashbacks 
containing E start from. It has to be stressed that ref- 
erence events have to be events proper. They caunot 
be states. The actual referent is the first element of 
the stack, provided that this element does not contMn 
fl~hback- or elaboration-embcddings. If it does we 
start the recursivc call of the search procedure with 
respect to the first of these embedded lists, tense{El, 
aspect(El, and tp(E) give us the valnes of tile tense 
and aspect features (sp, ppf, ev, st: simple past, past 
perfect, event and state) and the temporal perspec- 
tive from which E is seen. 
In order to decide the anchoring of new events, 
the rules of the system make use of three filters: 
(F1) tile tense/aspect filter which decides whether 
a particular discourse relation is possible with 
respect to tense and aspect, 
(F2) the "consistency" filter, which decides whether 
tile choice of a particular discourse relation is 
logically compatible with the information of the 
preceding text against the background of tile 
information from the inference component (this 
consists mainly of testing sort subsumptions ill 
the context of teml)oral incompatibilities be- 
tween calendar units and event sorts), and 
(F3) the evidence filter which tries to support the 
relation tested by means of rules like (9) and 
(10). 
To illustrate the system we restrict ourselves to 
some simple eases (compare figure 1). We suppose 
that the new sentence introduces just one event anti 
we only sketch simplified versions of tile rules for 
continuation, 1.), for elaboration, 2.), for flashback, 
4 For a detailed description of the calculus refer to Eberle {1991}. 
3.), for causally introduced background, 4.), for back- 
ground, 5.), for termination of elaborations, 6.), and 
for the termination of flashbacks, 7.). We skip rules 
which deal with specific subcases of these relations. 
For instance, we skip rules which deal with dummy 
referents t introduced by the causal background and 
rules which allow for more precise temporal localiza- 
tions of tile new events. We use rewrite rules which 
depict the amalgamation of a sentence DRS with a 
text DRS. 
In rule 1.) of figure 1, (*) reflects the particular 
case where e,.ef is ntember of all elaboration. If it is 
not, tile corresponding condition in 1(,+1 is omitted. 
Ill tile cases 1.) - 3.) (F1) requires that es is 
heterogeneous (aspect(es) = ev). Tile tense forms 
of reference- and new event have to be identical 
(tense(es) = tense(e,.,l) ). In addition, (F1) accepts 
3.), if lense(ere/) = sp and tense(es) = ppf). This 
arrangement allows also for continuation and elabora- 
tion on flashback levels and for iteration of flashbacks. 
For 4.) and 5.) tile tense forms must be identical. In 
contrast to 1.) - 3.), it is required that aspect(es) = st. 
(F2) should be clear in all cases. For acceptance, 
the tested solutious have to be consistent. 
With respect to tile cases 1.) - 4.), (F3) accepts 
the relation tested, if some evidence for this relation 
can be inferred from the representation of the pre- 
ceding text on the be~sis of the background knowl- 
edge. Ill addition, in tile cases 1.) - 3.) (F3) also ac- 
cepts the relation, if evidence for the coml)eting rela- 
tions cannot be entailed. Ill tile case of tet~e(es) = 
tense(e,.el) = ppf, flashback, elaboration and contin- 
uation are pairwise competing relations. Ill the case 
of tense(es) = tense(e,,~I) = sp, fl,'~shbaek is ruled 
out as competing relation with respect to elaboration 
and eontinuatiolL In the case of 5.), background, and 
3.) with tense(es) ¢ tense(e,.eI) (F3) is eml)ty. 
Evidence for a particular relation X can be in- 
ferred, if there exists an event sort P which is more 
general than the event description from DRSs with 
respect to tile event sort hierarchy, such that tile 
knowledge base (KB) together with the represen- 
tation of the l)receding text (DRS,) predicts an 
event e auf type P temporally related to e,.~y in a 
way significant for X, and that KB, DRS, do not 
entail the corresponding statement with respect to 
the competing relations of X. Formally, for instance 
with respect to X="eontinuation", X is evident, if 
for P with Ve (ORS s\[es/e\] ---* P(e)), we can in- 
fer that 3e (P(e) A e,~j" -< el, but neither that 
~e (P(e)AeCe,.~l)nor3e (P(e)Ae<e,ef). 
Tile whole system is implemented in Prolog. So 
tile order of the rules is important. Thus, we get a 
preference relation over the allowed readings. Contin- 
uation is preferred to elaboration which is preferred 
to flashback (ill the case that all alternatives seem 
possible and that they are supported by (F3)). With 
respect to 6.) and 7.), we see that local level expan- 
sion is preferred to the anchoring of the new event at 
AOTES DE COLING-92. NANTES, 23-28 AO't~'r 1992 2 9 2 PROC. OF COLING-92. NAN'I'ES. AUG. 23-28. 1992 
Text .+ l Text, Sentencen+i 
< T,,+bDRS.+~ > ~ < T.,DRS,, > < Ts,DRSs > 
1. Continuation: 
U,,+, = U,, O u~ 
K,,+l = I(,, U Ks 
\[ e,.~! -< e~ 
O e s -~ e,,, I 
e,s c e,,.~/(*)~ 
T,,+I .... \[ .... s .... I .... 1"" 
(F1) (F2) (F3) 
akt-re\](7~) = e,.~/ 
tv(e,.. D = elp 
el-embedder(T,),e.~ I) = eet~eI(*) 
(F1) (F2) (F3) akt-ref(T,,) ..... ! 
2. Elal)oration: 
g,,+l = U,,UUs 
K.+l = K,, U Ks O 
T.+I = .\[.e,.,/(\[et(fe~l),.\]),.l. 
-,3. Flashback: 
U,,+I = U,, U Us 
/(.+l = K,,UKsU 
:/'.+1 = .\[.erel(\[fb(\[e s l\],~.l\], .1' 
(F1) (F2) (F3) akt-ref(T,,) ..... S 
(rl) (F:9) (r'3) akt-re/(T,,) ...... S T~ = le~l 
(Vl) (F2) (F3) akt-reJ:(T,,) ..... ! Ts = \[es\] 
4. Causally introduced Background: 
g,,+l = V. UUs 
/(,,+l = I(,,UI(.~. 
_O r,,~( e .~l), e,.ot, ._~,. l . 
5. Background: 
U.+I = u,,uus K,,+l = K,,UKs U ~ 
_ T,,+I = .{.c~/(\[ba(le~\]),.l),.I. 
6./7. Return from elaborations/flashbacks: 
restart the system 
with e,.,s', 
instead of e,.~I 
(F1)(F2)(F3) 
akt-,'4(Y.) = e,,,+ 
6.) el-embedder( Z, , e,.e f ) = ere l' 
7.) Sb-embedder( T. , e,.~ f ) = e.. s, 
Figure l: resolution rules 
T,s =-\[es\] 
higher levels of the hierarchy. This strategy seems to 
be validated by the text phenomena. 
For illustration, we briefly sketch the effects of the 
rules when applied to the sample text (6). Leaving 
Paris- and Going to Frankfort-events cannot over- 
lap. Therefore, (F2) rules out the alternative "elabo- 
ration" for e~ with respect to ca and the remMning 
competing "continuation" is chosen. Further supl)ort 
by (F3) is not needed in this c~se. For travels e to h)r- 
eign countries wc can infer the existence of PLA CED- 
AT~states e' fin" the agent or theme of the travelling 
event and for the borders contained in the path of 
the travelling such that e ~ C e (compare the KB rule 
(9)). Since the description of e3, the stop at the bof 
der. is subsumed by the more general PLACED-AT: 
sort, (F3) - ia the absence of corresl)onding evidences 
for the competing relation of continuation - SUl)ports 
"elaboration". Since for the states e4,cr,,e13 the al- 
teHlative of "causally introduced background" is not 
evident, they are anchored by means of the rule 5.). 
In the case of e¢ and e9, the fact that evidence for 
"elaboration" is missing selects for "contimiation'. 
ell) which is situated in Fraukfort cannot lie part of 
the travel. For this reason anchoring el0 with respect 
to the level of e,.,l, which actually is eg, is not pos- 
sil)le. Therefore, with rule 6.), we have to climb up 
the elaboration hierarchy. The anchoring of eT, es, el 1 
and el2 should be clear on the basis of what we have 
said so far. 
In order to arrive at alternative readings of a text 
we use a specific backtracking routine which can sup- 
press the impact of (F3) for the resolution. 
6 Conclusion 
The described system complltes the event stru(:ture of 
a discourse. Siuce we think that the semantics of tense 
and aspect is not sufficient to establish tile temporal 
dlsfollrse llleallillg, we use backgl'ollnd kllowledge to 
¢lisambiguate between different readings. 
The system is incomplete at present. The frag- 
ment it deals with doesn't allow for relative clauses. In 
addition the interactioa between NP-resolutlon and 
temporal resolution is not available. Some temporally 
relevant discourse relations are not treated. The logic 
tmderlying the preference relation is only rudimen- 
tary sketched. The approach to temporal resolution 
AcrEs DE COLIN'G-92, NAUrES, 23-28 AOU'T 1992 2 9 3 PROC. OF COLING-92, NANTES, AUG. 23-28, 1992 
presented here is a condensed version of that sug- 
gested in Eberle (1991). The corresponding imple- 
mentation for the LILOG prototype dates back to 
1989. Meanwhile, independent of this approach, sim- 
ilar suggestions have been made (cLAsher and Las- 
carides (1991)). In the work of Asher and Lascarides 
the underlying theory of non-monotonic reasoning is 
MASH (cf. Asher and Morreau (1991)). It is projected 
to extract a suited proof theory from MASH for the 
purpose of refining the temporal resolution compo- 
nent suggested, s 
ACIES DE COLING-92, NANTES, 23-28 AOI3T 1992 2 9 4 PROC. OF COLING.92, NANTES, AUO. 23-28, 1992 

References 

\[1\] Asher, N., Lazcarides. A. (1991): Discourse Rela- 
tions and De.feasible Knowledge. In: Proceedings 
of ACL,1991 

\[2\] Asher, N., Morreau. M. (1991): Common Sense 
Entailment: A Modal Theory of Non-monotonic 
Reasoning. In: Proceedings of IJCAIL, Sydney 
1991 

\[3\] Bartscb, R.(1987): Frame Representations and 
Discourse Representations. ITLI Prepublication 
Series 87-02. University of Amsterdam. 

\[4\] Dowty, D.(1986): The Effects of Aspectual Class 
on the Temporal Structure of Di.~course: Seman- 
ties or Pragmatics ?. In: Linguistics and Philoso- 
phy Vol.9,No.1 (1986) pp.37-62 

\[5\] Eberle, K. (1991): Ereignisse: Ihre Logik und On. 
tologie au.~ textsemantiseher Sicht. Dissertation 
Universit£t Stuttgart, 1991 

\[6\] Eberle, K., Kasper, W.(1989): Tenses as 
Anaphora. In: Proceedings of E-ACL, Manch- 
ester 1989 

\[7\] Grosz, B., Sidner, C.(1985): The Structure of 
Discourse Structure SRI-technical note 369, 1985 

\[8\] Hinrichs, E.(1986): Temporal Anaphora in Dis- 
courses of English. In: Linguistics and Philoso- 
phy Vol.9,No.1 (1986) pp.63-82 

\[9\] Kamp, H.(1981): A Theory of Truth and Se- 
mantic Representation. In: Groenendeijk ct al 
(eds.) Formal Methods in the Study of Language. 
Mathematical Centre Tract, Amsterdam 

\[10\] Kamp, H., Rohrer, C. (1983): Tense in Texts. In: 
B£uerle. R.,Schwarze, C.,von Stechow, A. (eds.), 
Meaning, Use and Interpretation of Language, 
Berlin. de Gruyter 

\[11\] Kamp. H., Rohrer, C. (1985): Temporal Re\]er. 
ence :n French. Ms. Stuttgart 
s.c, ince M. Morreau is member of our institute and N. Asher visiting professor in Stuttgart, this can be done in close con- 
nection with the authors of MASH. 

\[12\] Kritka, M. (1987): Nominalreferenz und geitkon- 
stitution. Znr Semantik yon Massentermen, PIu- 
raltermen und Aspektklassen. Dissertation Uni- 
versidi.t Miinchen, 1987 

\[13\] Minsky, M.(1975): A Framework for Represent- 
ing Knowledge. In: Winston,P. (ed.) The Psy- 
chology of Computer Vision. McGraw-Hill, New 
York. 

\[14\] Moens, M., Steedman,/vl. (1986): The Temporal 
Dimension in Information Modelling and Natu- 
ral Language Processing, Acord Deliverable 2.5, 
Edinburgh,1986 

\[15\] Moens, M., Steedman, M. (1988): Teraporal On- 
tology and Temporal Reference, in: Computa- 
tional Linguistics 14(2), 1988 

\[16\] Partee, B. (1984): Nominal and Tem- 
poral Anaphora, in: Linguistics and Philosophy 
Vo1.7,No.3 (1984) pp.243-287 

\[17\] Reichenbach, H. (1947): Elements of Symbolic 
Logic. New York, 1947 

\[18\] Reyle, U. (1986): Zeit und Aspekt bet der Verar- 
beitung natarlicher Sprachen. Dissertation Insti- 
tut fiir Linguistik tier Universit£t Stuttgart,1986 

\[19\] Schank, R., Abelson, R. (1977): Scripts, Plans, 
Goals and Understanding. Hillsdale, N.J. LEA. 

\[20\] Thompson, S.,Mann, W.C. (1987): Rhetorical 
Structure Theory: A Framework for the Analy- 
sis of Texts. In: International Pragmatics Asso- 
ciation Papers in Pragmatics, Vol.1, pp.79-105, 
1987 

\[21\] Vendler, Z. (1967): Linguistics in Philosophy, 
Cornell University Press, Ithaca, New York 

\[22\] Webber, B.L. (1988): Tense as Discourse 
Anaphor, in: Computational Linguistics 14(2), 
1988 
