Managing information at linguistic interfaces* 
Johan Bos and C.J. Rupp 
Computerlinguistik 
Universit~it des Saarlandes 
D-66041 Saarbrticken 
{bos, cj }@toll. uni-sb, de 
Bianka Buschbeck-Wolf and Michael Dorna 
Institut ftir Maschinelle Sprachverarbeitung (LMS) 
Universit~it Stuttgart 
D-70174 Stuttgart 
{bianka, michl}@ims, uni-sl;utl;gart, de 
Abstract 
A large spoken dialogue translation system imposes 
both engineering and linguistic constraints on the 
way in which linguistic information is communi- 
cated between modules. We describe the design and 
use of interface terms, whose formal, functional and 
communicative role has been tested in a sequence of 
integrated systems and which have proven adequate 
to these constraints. 
1 Introduction 
This paper describes the development and usage 
of interface representations for linguistic informa- 
tion within the Vorbmobil project: a large dis- 
tributed project for speech-to-speech translation of 
negotiation dialogues, between English, German 
and Japanese. We take as our reference point the 
Verbmobil Research Prototype (Bub et al., 1997), 
but this is only one of a sequence of fully integrated 
running systems. The functional, formal and com- 
municative role of interface terms within these sys- 
tems, once instigated, has already been maintained 
through changes in the overall architecture and con- 
stitution of the Ve rbmobil consortium. 
There are two aspects to our story: 
• the practical and software engineering con- 
straints imposed by the distributed develop- 
ment of a large natural language system. 
• the linguistic requirements of the translation 
task at the heart of the system. 
° This work was funded by the German Federal Ministry of 
Education, Science, Research and Technology (BMBF) in the 
framework of the Vorbmobil project under grant 01 IV 701 
R4 and 01 IV 701 N3. The responsibility for this article lies 
with the authors. Thanks to our many Vorbmobil colleagues 
who helped to design and develop the results presented here, 
and also to the anonymous reviewers for giving useful hints to 
improve this paper. 
The prominence of the engineering requirements 
is further heightened by the fact that we are dealing 
with a spoken dialogue system which must strive 
towards real time interactions. 
We proceed by describing the requirements of the 
Vorbmobi! Research Prototype, the actual contents 
of the Verbmobil Interface Term (henceforth VIT), 
the semantic formalism encoded within VITs and 
processing aspects. We conclude that VITs fulfill 
the joint goals of a functional interface term and 
an adequate linguistic representation within a single 
data structure. 
1.1 Modularity 
We are concerned here with the interface represen- 
tations exchanged between modules that make use 
of traditional linguistic concepts. Figure 1 shows a 
simplified form of the Verbmobi! Research Proto- 
type architecture I , the modules in the grey shaded 
area make use of VITs as linguistic interface terms 
and are loosely termed the "linguistic" modules of 
the system, in contrast to the other modules shown 
which are chiefly concerned with the processing of 
acoustic signals. The linguistic design criteria for 
VITs derive mainly from the syntactic and seman- 
tic analysis module, labelled SynSem, the generation 
and the transfer modules. 
One point that should be made at the outset is that 
these linguistic modules really are modules, rather 
than, say, organisational constructs in some larger 
constraints system. In practice, these modules are 
developed at different sites, may be implemented 
in different programming languages or use differ- 
ent internal linguistic formalisms, and, indeed, there 
may be interchangeable modules for the same func- 
tion. This level of modularity, in itself, provides suf- 
ficient motivation for a common interface represen- 
~ln that we have excluded the modules that employ alter- 
native techniques and express no interest in linguistic informa- 
tion. 
160 
I Recogrtitionl-~ Prosody "~; SynSem ~: ~::z ~,!::~ :: : ;:':~" ~::: Synthesis I I J 
.............. ~;~ ..... ~2~ ........... ............. ~ .... , ....... 
Figure 1: A Simplified diagram of the Vorbraobil Research Prototype Architecture 
tation among the linguistic modules, allowing the 
definition of a module's functionality in terms of 
its I/O behaviour, but also providing a theory in- 
dependent linguafranca for discussions which may 
involve both linguists and computer scientists. The 
Verbraobil community is actually large enough to 
require such off-line constructs, too. 
1.2 Encoding of Linguistic Information 
A key question in the design of an interface lan- 
guage is what information must be carried and 
to what purpose. The primary definition criterion 
within the linguistic modules has been the transla- 
tion task. The actual translation operation is per- 
formed in the transfer module as a mapping between 
semantic representations of the source and target 
languages, see (Dorna and Emele, 1996). However, 
the information requirements of this process are 
flexible, since information from various levels of 
analysis are used in disambiguation within the trans- 
fer module, including prosody and dialogue struc- 
ture. 
To a large extent the information requirements di- 
vide into two parts: 
• the expressive adequacy of the semantic repre- 
sentation; 
• representing other types of linguistic informa- 
tion so as to meet the disambiguation require- 
ments with the minimum of redundancy. 
The design of the semantic representations en- 
coded within VITs has been guided by an ongoing 
movement in representational semantic formalisms 
which takes as a starting point the fact that certain 
key features of a purely logical semantics are not 
fully defined in natural language utterances and typ- 
ically play no part in translation operations. This 
has been most clearly demonstrated for cases where 
quantifier scope ambiguities are exactly preserved 
under translation. The response to these observa- 
tions is termed underspecification and various such 
underspecified formalisms have been defined. In 
one sense underspecification is, in itself, a form of 
information management, in that only the informa- 
tion that is actually present in an utterance is repre- 
sented, further disambiguation being derived from 
the context. In the absence of such contextual in- 
formation further specificity can only be based on 
guesswork. 
While the management of information in the VIT 
semantics consists of leaving unsaid what cannot be 
adequately specified, the amount of information and 
also the type of information in the other partitions 
of the VIT (see Section 2.1) has been determined 
by the simple principle of providing information on 
justified demand. The information provided is also 
quite varied but the unifying property is that the 
requirements are defined by the specific needs of 
transfer, in distinguishing cases that are truly am- 
biguous under translation or need to be resolved. 
For example: 
(1) Geht das bei ihnen? 
a. Is it possible for you? 
b. Is it possible at your place? 
In (1), the German preposition bei displays an 
ambiguity between the experiencer reading (1 a) and 
the spatial interpretation (lb). The resolution of this 
ambiguity requires in the first instance three pieces 
of information: the type of the verb predicate, the 
sort of the internal argument of bei and the sort 
of the subject. This, in turn, requires the resolution 
of the reference of the anaphor das, where mor- 
phosyntactic constraints come into play. If the ref- 
erent has the sort time then the experiencer read- 
ing (la) can be selected. This is the more usual re- 
sult in the Verbmobil context. Should the referent 
161 
Slot Name Description 
VIT ID 
Index 
Conditions 
Constraints 
Sorts 
Discourse 
Syntax 
Tense and Aspect 
Prosody 
combines a unique tag for the turn segment described by the current VIT and the 
word lattice path used in its linguistic analysis; 
a triple consisting of the entry points for traversing the VIT representation; 
labelled conditions describing the possibly underspecified semantic content of an 
utterance; 
scope and grouping constraints, e.g. used for underspecified quantifier and opera- 
tor scope representation; 
sortal specifications for instance variables introduced in labelled conditions; 
additional semantic and pragmatic information, e.g. discourse roles for individual 
instance; 
morpho-syntactic features, e.g. number and gender of individual instances; 
morpho-syntactic tense combined with aspect and sentence mood information, 
e.g. used for computing surface tense; 
prosodic information such as accenting and sentence mood. 
Table 1: A list of VIT slots. 
be sortally specified as a situation, further infor- 
mation will be required to determine the dialogue 
stage, i.e. whether the time of appointment is being 
negotiated or its place. Only in the latter case is the 
spatial reading (lb) relevant. 
(2) Dann wiirde das doch gehen 
a. Then, it WOULD be possible, after all. 
b. It would be possible, wouldn't it? 
Consider the discourse particle doch in (2) which 
can be disambiguated with prosodic information. 
When doch is stressed and the utterance has falling 
intonation, it functions as a pointer to a previous 
dialogue stage. Something that was impossible be- 
fore turned out to be feasible at the utterance time. 
Then, doch is translated into after all and the auxil- 
iary takes over the accent (2a) 2. If (2) has a rising 
intonation and the particle is not stressed, it signals 
the speaker's expectation of the hearer's approving 
response. In English, this meaning is conveyed by 
a question tag (2b). Lieske et al. (1997) provide a 
more detailed account of the use of prosodic infor- 
mation in Verbmobil. 
In addition to the information that is explicitly 
represented in the specified fields of a VIT, in- 
cluding the surface word order that can be inferred 
from the segment identification, and the resolution 
of underspecified ambiguities in context, transfer 
might require further information, such as domain- 
specific world knowledge, speech act or discourse 
2We indicate prosodic accent with SMALL CAPITALS. 
stage information. This information can be obtained 
on demand from the resolution component (see Fig- 
ure 1). This flexible approach to the information 
required for transfer is termed cascaded disam- 
biguation (Buschbeck-Wolf, 1997) and is balanced 
against the fact that each level of escalation implies 
a correspondingly greater share of the permissible 
runtime. 
2 The Verbmobil Interface Term 
The VIT encodes various pieces of information pro- 
duced and used in the linguistic modules. The con- 
tent of a VIT corresponds to a single segment (or 
utterance) in a dialog turn. This partitioning of turns 
enables the linguistic components to work incre- 
mentally. 
2.1 Multiple Levels of Information 
A VIT is a record-like data structure whose fields 
are filled with semantic, scopal, sortal, morpho- 
syntactic, prosodic, discourse and other information 
(see Table 1). These slots can be seen as analysis 
layers collecting different types of linguistic infor- 
mation that is produced by several modules. The in- 
formation within and between the layers is linked 
together using constant symbols, called "labels", 
"instances" and "holes". These constants could be 
interpreted as skolemized logical variables which 
each denote a node in a graph. Besides purely lin- 
guistic information, a VIT contains a unique seg- 
ment identifier that encodes the time span of the an- 
alyzed speech input, the analyzed path of the orig- 
inal word lattice, the producer of the VIT, which 
language is represented, etc. This identifier is used, 
162 
for example, to synchronize the processing of anal- 
yses from different parsers. For processing aspects 
of VITs see Section 3. 
A concrete example of a VIT is given in Figure 2 
in a Prolog notation where the slots are also marked. 
This example is further discussed in Section 2.2. 
2.2 VIT Semantics 
The core semantic content of the VIT is contained 
in the two slots: Conditions and Constraints. 
The conditions represent the predicates of the se- 
mantic content and the constraints the semantic 
dependency structure over those predicates. This 
partitioning between semantic content and seman- 
tic structure is modelled on the kind of represen- 
tational metalanguage employed in UDRS seman- 
tics (Reyle, 1993) to express underspecification. 
The semantic representation is, thus, a metalan- 
guage expression containing metavariables, termed 
labels, that may be assigned to object language con- 
structs. Moreover, such a metalanguage is minimally 
recursive 3, in that recursive structure is expunged 
from the surface level by the use of metavariables 
over the recursive constituents of the object lan- 
guage. 
In UDRSs quantifier dependencies and other 
scope information are underspecified because the 
constraints provide incomplete information about 
the assignment of object language structures to la- 
bels. However, a constraint set may be monoton- 
ically extended to provide a complete resolution. 
VIT semantics follows a similar strategy but some- 
what extends the expressivity of the metalanguage. 
There are two constructs in the VIT semantic 
metalanguage which provide for an extension in 
expressivity relative to UDRSs. These have both 
been adopted from immediate precursors within the 
project, such as Bos et al. (1996), and further re- 
fined. The first of these is the mechanism of holes 
and plugging which originates in the hole seman- 
tics of Bos (1996). This requires a distinction be- 
tween two types of metavariable employed: labels 
and holes. Labels denote the instantiated structures, 
primarily the individual predicates. Holes, on the 
other hand, mark the underspecified argument po- 
sitions of propositional arguments and scope do- 
mains. Resolution consists of an assignment of la- 
3We owe the term minimal recursion to Copestake 
et al. (1995), but the mechanism they describe was already in 
use in UDRSs. 
vit( 
X Segment ID 
vitID(sid(ll6.a.ge.2.181.2.ge.y.syntaxger). 
X WHG Strin E 
\[word(jedes.24.\[i5\]). 
word(treffen.25.\[19\]). 
word(mit.26.\[112\]). 
word(ihnen.27.\[114\]). 
word(hat.28.\[ll\]). 
,ord(ein.29.\[117\]). 
word(interessantes.30.\[121\]). 
word(thema°31.\[122\])\]). 
X Index 
index(12.11.i3). 
Conditions 
\[decl(12.h23). 
jed(1S,i6,18,h7), 
treffen(19,i6), 
mit(112,i6,i13), 
pron(l14.i13). 
haben(11.i3). 
arg3(ll,i3,i6), 
arg2(ll,i3,i16), 
ein(llT.il6.120.hl9). 
interessant(121.i16). 
thema(122.i16)\]. 
Constraints 
\[in_g(122.120). 
in_g(121.120). 
in_g(114.18). 
in_g(112.18). 
in_g(19.18). 
leq(15,h23), 
leq(ll.h7). 
leq(117.h23). 
leq(11.h19)\]. 
Z Sorts 
\[s_sort(i6,meeting_sit), 
s_sort(i13,human), 
s_sorr(il6,info_content)\], 
X Discourse 
\[prontype(i13,he,std), 
dir(112.no)\]. 
Syntax 
\[num(i6.sg). 
pers(i6,3), 
pers(il3.1). 
num(i13.pl). 
num(i16.sg). 
pets(J16.3)\]. 
Tense and Aspect 
\[ta_mood(i3.ind). 
ta_tense(i3.pres). 
ta_perf(i3.nonperf)\]. 
Prosody 
\[pros_accent(15). 
pros_accent(121). 
pros_mood(12.decl)\] 
) 
Figure 2: A Verbmobil Interface Term 
163 
bels to holes, a plugging. In general, the constraint 
set contains partial constraints on such a plugging, 
expressed by the "less than or equal" relation (leq) 
but no actual equations, leq(L ,H) should be inter- 
preted as a disjunction that the label is either subor- 
dinate to the hole or equal to it. Alternatively, this 
relation can be seen as imposing a partial ordering. 
A valid plugging must observe all such constraints. 
The other extension in expressivity was the in- 
troduction of additional constraints between labels. 
These form another type of abstraction away from 
logical structure, in that conjunctive structure is also 
represented by constraints. The purpose of this fur- 
ther abstraction is to allow lexical predicates to be 
linked into additional structures, beyond that re- 
quired for a logical semantics. For example, focus 
structure may group predicates that belong to dif- 
ferent scope domains. More immediately, prosodic 
information, expressed at the lexical level, can be 
linked into the VIT via lexical predicates. The con- 
straints expressing conjunctive structure relate basic 
predicate labels and group labels which correspond 
to a conjunctive structure in the object language, 
such as a DRS, intuitively a single box. Grouping 
constraints take the form in_g (L, G), where the "in 
group" relation (in_g) denotes that the label L is a 
member of the group G. The content of a group is, 
thus, defined by the set of such grouping constraints. 
n 
g = A li such that in_g(//, g) 6 Constraints 
i=1 
These two forms of abstraction from logical struc- 
ture have a tendency to expand the printed form of 
a VIT, relative to a recursive term structure, but in 
one respect this is clearly a more compact represen- 
tation, since lexical predicates occur only once but 
may engage in various types of structure. 
2.3 An Example 
Figure 2 shows the VIT produced by the analysis 
of the German sentence: Jedes Treffen mit lhnen hat 
ein interessantes Thema ("Every meeting with you 
has an interesting topic"). The instances which cor- 
respond to object language variables are represented 
by the sequence {il, i2, ...}, holes by {hi, 
h2, ...} and labels, including group and predi- 
cate labels, by {ll, 12 .... }. The base label of a 
predicate appears in its first argument position. The 
predicates haben, arg2 and arg3 share the same 
label because they form the representation of a sin- 
gle predication, in so-called neo-Davidsonian nota- 
decl(12,h23) 
jed(1f,i6,18,h7) ein(llT,il6,120,hl9) 
haben(ll,i3) 
arg3(ll,i3,i6) 
arg2(ll,i3,il6) 
Figure 3: A Graphical Representation of the Scop- 
ing Constraints 
tion (e.g. (Parsons, 1991)). The two groups 120 and 
18 form the restrictions of the existential quantifier, 
ein, and the universal, j ed, respectively. Two of the 
scoping constraints place the quantifiers' labels be- 
low the top hole, the argument of the mood operator 
(decl). The other two link the quantifiers respec- 
tive scopes to the bottom label, in this case the main 
verb, but no constraints are imposed on the relative 
scope of the quantifiers. The whole structure is best 
viewed as a (partial) subordination hierarchy, as in 
Figure 3. 
A complete resolution would result from an as- 
signment of the labels {11, 15, 117} to the three 
holes {h23, hi9, h7}. Taking into account the 
implicit constraint that any argument to a predicate 
is automatically subordinate to its label, there are in 
fact only two possibilities: the pluggings pl and p2 
given below, 
II I h231h191 h7 II 
p2 117 15 11 
corresponding to the two relative scopings of the 
quantifiers: 
VX(TREFFEN(X)&MIT(X,Z)-+ 3y(THEMA(y) 
&INTERESSANT(y)&HAB EN(X,y))) 
and 
164 
3y(THEMA(y)&INTERES SANT(y)& 
Vx(TREFFEN(X)&M IT(X,Z)-+ HABEN(X,y))) 
A full resolution is, however, rarely necessary. 
Where transfer requires more specific scoping con- 
straints these can be provided in a pairwise fash- 
ion, based on default heuristics from word order or 
prosodic cues, or at the cost of an additional call to 
the resolution component. 
3 VIT Processing 
3.1 Lists as Data Structures 
Almost all fields in a VIT record are represented 
as lists. 4 In general, list elements do not introduce 
any further recursive embedding, i.e. the elements 
are like fixed records with fields containing con- 
stants. The non-recursive nature of these list ele- 
ments makes the access and manipulation of in- 
formation (e.g. addition, deletion, refinement, etc.) 
very convenient. Moreover, the number of list el- 
ements can be kept small by distributing informa- 
tion over different slots according to its linguistic 
nature (see Section 3.2 below) or the producer of 
the specific kind of data (see Section 2.1). The kind 
of structuring adopted and the relative shortness of 
the lists make for rapid access to and operations on 
VITs. 
It is significant that efficient information access is 
dependent not only on an appropriate data structure, 
but also on the representational formalism imple- 
mented in the individual data items. This property 
has been presented as an independent motivation for 
minimally recursive representations from the Ma- 
chine Translation point of view (Copestake et al., 
1995), and has been most thoroughly explored in 
the context of the substitution operations required 
for transfer. We believe we have taken this argu- 
ment to its logical conclusion, in implementing a 
non-recursive semantic metalanguage in an appro- 
priate data structure. This, in itself, provides suf- 
ficient motivation for opting for such representa- 
tions rather than, say, feature structures or the recur- 
sive QLFs (Alshawi et al., 1991) of CLE (Alshawi, 
1992). 
3.2 ADT Package 
In general, linguistic analysis components are very 
sensitive to changes in input data caused by modifi- 
4In typical AI languages, such as Lisp and Prolog, lists are 
built-in, and they can be ported easily to other programming 
languages. 
cations of analyses or by increasing coverage. Ob- 
viously, there is a need for some kind of robust- 
ness at the interface level, especially in large dis- 
tributed software projects like Verbmobil with par- 
allel development of different components. There- 
fore, components that communicate with each other 
should abstract over data types used at their inter- 
faces. This is really a further projection of standard 
software engineering practice into the implementa- 
tion of linguistic modules. 
In this spirit, all access to and manipulation of 
the information in a VIT is mediated by an abstract 
data type (ADT) package (Doma, 1996). The ADT 
package can be used to build a new VIT, to fill it 
with information, to copy and delete information 
within a VIT, to check the contents (see Section 3.3 
below), to get specific information, to print a VIT, 
etc. To give an example of abstraction, there is no 
need to know where specific information is stored 
for later lookup. This is done by the ADT package 
that manages the adding of a piece of information 
to the appropriate slot. This means that the external 
treatment of the VIT as an interface term is entirely 
independent of the internal implementation and data 
structure within any of the modules and vice versa. 5 
3.3 Consistency Checking 
As a side effect of adopting an extensive ADT pack- 
age we were able to provide a variety of check- 
ing and quality control functions. They are espe- 
cially useful at interfaces between linguistic mod- 
ules to check format and content errors. At the for- 
mat checking level language-specific on-line dic- 
tionaries are used to ensure compatibility between 
the components. A content checker is used to test 
language-independent structural properties, such as 
missing or wrongly bound variables, missing or in- 
consistent information, and cyclicity. 
As far as we are aware, this is the first time that 
the results of linguistic components dealing with se- 
mantics can be systematically checked at module 
interfaces. It has been shown that this form of test- 
ing is well-suited for error detection in components 
with rapidly growing linguistic coverage. It is worth 
noting that the source language lexical coverage in 
the Verbmobil Research Prototype is around 2500 
words, rising to 10K at the end of the second phase 6. 
Furthermore, the complex information produced by 
5The kind of data structure used by the communication ar- 
chitecture (Amtrup and Benra, 1996) is, similarly, transparent 
to the modules. 
6In the year 2000. 
165 
linguistic components even makes automatic output 
control necessary. 
The same checking can be used to define a quality 
rating, e.g. for correctness, interpretability, etc. of 
the content of a VIT. Such results are much better 
and more productive in improving a system than 
common, purely quantitative, measures based on 
failure or success rates. 
4 Conclusion 
We have described the interface terms used to carry 
linguistic information in the Verbmobil system. 
The amount and type of information that they carry 
are adapted to the various constraints that arose dur- 
ing the distributed development of such a large sys- 
tem over a periods of several years. The VIT format 
has been integrated into several running systems, 
including the most successful, the Verbmobil Re- 
search Prototype. Subsequent modifications in the 
second phase of Verbmobil have been possible in 
a systematic and monotonic fashion. Similarly, the 
adaption of several new syntactic and semantic anal- 
ysis modules has not presented any major or un- 
seen problems. Among the tasks currently under de- 
velopment is the use of VIT representations in the 
generation of dialogue protocols, in the language of 
each participant. 
We feel that VITs have proved themselves ad- 
equate to the considerable system requirements 
through their continued, efffective use. We also find 
it reassuring that, despite the priority given to the 
engineering requirement, we have been able to em- 
bed within this interface language representations 
that are, at least, equivalent to the state of the art 
in Computational Semantics. 

References 
Hiyan Alshawi, David M. Carter, Bj6ru Gamback, 
and Manny Rayner. 1991. Translation by Quasi 
Logical Form Transfer. In Proceedings of the 
29th Annual Meeting of the Association for Com- 
putational Linguistics (ACL'91), pages 161-168, 
Berkeley, CA. 
Hiyan Alshawi, editor. 1992. The Core Language 
Engine. ACL-MIT Press Series in Natural Lan- 
guages Processing. MIT Press, Cambridge, MA. 
Jan W. Amtrup and J6rg Benra. 1996. Communica- 
tion in Large distributed AI Systems for Natural 
Language Processing. In Proceedings of the 16th 
International Conference on Computational Lin- 
guistics (Coling'96), pages 35-40, Copenhagen, 
Denmark. 
J. Bos, B. Gamb~ick, C. Lieske, Y. Mori, M. Pinkal, 
and K. Worm. 1996. Compositional Semantics 
in Verbmobil. In Proceedings of the 16th Interna- 
tional Conference on Computational Linguistics 
(Coling'96), pages 131-136, Copenhagen, Den- 
mark. 
Johan Bos. 1996. Predicate Logic Unplugged. In 
Paul Dekker and Martin Stokhof, editors, Pro- 
ceedings of the Tenth Amsterdam Colloquium, 
pages 133-143, ILLC/Department of Philosophy, 
University of Amsterdam. 
T. Bub, W. Wahlster, and A. Waibel. 1997. 
Vorbmobil: The Combination of Deep and Shal- 
low Processing for Spontaneous Speech Trans- 
lation. In Proceedings of the 22nd Interna- 
tional Conference on Acoustics, Speech, and Sig- 
nal Processing (ICASSP'97), Munich, Germany, 
April. 
Bianka Buschbeck-Wolf. 1997. Resolution on De- 
mand. Verbmobil Report 196, IMS, Universit~it 
Stuttgart, Germany. 
A. Copestake, D. Flickinger, R. Malouf, S. Riehe- 
mann, and I. Sag. 1995. Translation using Mini- 
mal Recursion Semantics. In Proceedings of the 
6th International Conference on Theoretical and 
Methodological Issues in Machine Translation 
(TMI'95), Leuven, Belgium. 
Michael Dorna and Martin C. Emele. 1996. 
Semantic-based Transfer. In Proceedings of the 
16th International Conference on Computational 
Linguistics (Coling'96), Copenhagen, Denmark. 
Michael Doma. 1996. The ADT-Package for the 
Verbmobil Interface Term. Verbmobil Report 
104, IMS, Universitat Stuttgart, Germany. 
C. Lieske, J. Bos, B. Gamb~ick M. Emele, and 
C.J. Rupp. 1997. Giving prosody a meaning. 
In Proceedings of the 5th European Conference 
on Speech Communication alut Technology (Eu- 
roSpeech'97), Rhodes, Greece, September. 
T. Parsons. 1991. Events in the Semantics of En- 
glish. MIT Press, Cambridge, Mass. 
Uwe Reyle. 1993. Dealing with Ambiguities 
by Underspecification: Construction, Represen- 
tation and Deduction. Journal of Semantics, 
10(2):123-179. 
