ADAM: An Architecture for xml-based Dialogue Annotation 
on Multiple levels 
Claudia Soria 
ILC - CNR 
Pisa, Italy 
soria@ilc.pi.cnr.it 
Roldano Cattoni 
ITC-irst 
Trento, Italy 
cattoni@itc.it 
Morena Danieli 
CSELT 
Torino, Italy 
Morena.Danieli@cselt.it 
Abstract 
In this paper annotation modular- 
ity and use of annotation meta- 
schemes are identified as basic re- 
quirements for achieving actual cor- 
pora reusability. We discuss these 
concepts and the way they are 
implemented in the architectural 
framework of the ADAM corpus, 
which is a corpus of 450 Italian spon- 
taneous dialogues. The design of 
ADAM architecture is compatible 
with as many practices of dialogue 
annotation as possible, as well as ap- 
proaches to annotation at different 
levels. 
1 Introduction 
In this paper we describe the methodological 
assumptions and general architectural frame- 
work underlying the ADAM Corpus, which 
is currently being developed as part of the 
Italian national project SI-TAL (Autori vari, 
2000) 1. Annotated dialogue corpora are of 
crucial importance for the development of vo- 
cal applications. Because of their cost, how- 
ever, it is essential that their acquisition and 
annotation be designed in order to maximize 
their reusability as much as possible. The 
main assumption behind the design of the 
ADAM corpus is that actual reusability of 
corpora cruciMly depends on the strategic 
choices concerning the architectural design of 
the corpus, i.e. the way in which annotation 
tThe final version of the ADAM Corpus will be 
released by the end of 2001. A pilot set, consisting of 
30 dialogues, is currently available. 
is organized and structured, and the way in 
which this is represented in a given physical 
format 2. Corpus reusability is traditionally 
seen as mainly a by-product of either corpus 
representativeness and degree of standardiza- 
tion of annotation schemes. We claim that 
two other requirements should be taken into 
account when designing and building an anno- 
tated corpus that aims at reusability, namely 
modularity of annotation, and use of annota- 
tion meta-schemes. The meaning of these no- 
tions, as well as the way in which they are 
realized by the ADAM Corpus will be de- 
scribed in detail throughout the paper. Sec- 
tion 2 introduces the corpus; section 3 focuses 
on ADAM: the motivations underlying its ar- 
chitecture are presented and the adoption of 
XML as mark-up  is discussed. Fi- 
nally, the various annotation schemes are il- 
lustrated in section 4. 
2 Corpus Description 
The ADAM Corpus, which is currently be- 
ing developed, will consist of 450 Italian 
dialogues, both human-human and human- 
machine, belonging to the tourist domain. 
The human-machine component of the corpus 
is represented by human-machine dialogues 
over the phone, which are recordings of ac- 
tual interactions occurring between customers 
and the Italian national railway information 
system (FS-Informa, (Bahia et al., 2000)). 
The human-human component is represented 
by task-oriented dialogues between a person 
playing the role of a travel agent and another 
playing the role of a customer. Each dia- 
logue in the corpus is represented by an or- 
2For a similar view see (Ide and Brew, 2000). 
9 
thographic transcription, recording the words 
uttered by the speakers plus any other non 
linguistic sound. The transcription is linked 
to an audio file in PCM format. In addi- 
tion, each dialogue is annotated according to 
five annotation levels, namely prosody, mor- 
phosyntax, syntax, semantics and pragmatics 
(see below). 
3 ADAM: Architectural Principles 
3.1 Reusability Requirements 
The ADAM approach is mainly driven by the 
need of meeting the requirements of poten- 
tial users of annotated corpora, with a par- 
ticular emphasis on corpora reusability. An 
annotated corpus is reusable as far as it com- 
plies with several requirements, such as cor- 
pus representativeness and use of standard- 
ized annotation schemes (see some widely 
renoWned standardization efforts such as EA- 
GLES, MATE, DRI, ATLAS, etc.). The 
physical format or mark-up  for cor- 
pus encoding is another crucial issue, as it 
will be argued in section 3.4. In addition to 
this, we claim that an annotated corpus is 
useful beyond the immediate particular ap- 
plication aims only to the extent to which it 
is designed so as to meet two other important 
needs. A fundamental requirement appears 
to revolve around the way annotation is orga- 
nized, structured and represented in a corpus. 
In short, it is essential that the annotation, 
i.e. the linguistic information added to the 
data, should be easily and quickly modifiable 
at a moderately low cost by subsequent users 
of the corpus. We can think of at least two 
possible scenarios, referring to two orthogo- 
nal dimensions of customization operations. 
First, it might be the case that a user wishes 
to reuse a corpus which is annotated for sev- 
eral types of linguistic information, but lacks 
of a particular annotation type; the poten- 
tial user could nevertheless be interested in 
the existing annotations, and would like to 
supplement them with a new one. On the 
other hand, it might be the case that a user 
is interested in some annotation only (e.g., 
pos-tagging or syntactic structure) and s/he 
might want to leave aside other annotation 
types. Reusability of an annotated corpus 
can thus be thought of as a function of the 
extent to which new levels of linguistic infor- 
mation can be added, or uninteresting ones 
can be removed. This is what we call the ver- 
tical dimension of customization in annotated 
corpora. Second, for each level of linguistic 
analysis, an annotated corpus is likely to be 
reused depending on the extent to which ex- 
isting annotation can be changed, so as to ac- 
commodate different annotation practices. It 
is often the case that a corpus which is anno- 
tated with a given annotation scheme "hard- 
wires" the annotation so as it is impossible 
to replace the annotation without reverting 
to the raw text and rebuilding the annotation 
from scratch, which is enormously expensive. 
This is what we call the horizontal dimension 
off customization of an annoted corpus. 
The extent to which an annotated corpus 
can be compliant with these two requirements 
clearly depends on the architectural choices 
made at the design level: if, for instance, all 
types of annotation are flattened onto a single 
representation level, it is clear that the cus- 
tomizing operations above become hardly fea- 
sible. We claim that the vertical and horizon- 
tal customization requirements can be easily 
achieved on the one hand by appealing to the 
two related notions of modularity of annota- 
tion and use of annotation meta-schemes, and 
on the other by exploiting a physical format 
of encoding that fully supports them. In the 
next three sections we illustrate the two con- 
cepts as well as the way they are implemented 
in the ADAM Corpus. 
3.2 Annotation Modularity 
In an annotated corpus, several different 
types of annotation or linguistic information 
may be present in relation to the same in- 
put data. These types of information can be 
thought of as independent, yet related, levels 
or dimensions of linguistic description . We 
thus can think of a level of prosodic analy- 
sis, another of pos-tagging, another of seman- 
tic analysis, etc. By annotation modular- 
ity we mean that the different layers of an- 
10 
notation are to be kept independent one of 
another. In the ADAM Corpus we provide 
five layers of annotation: prosodic, morpho- 
syntactic (pos tagging), syntactic, conceptual- 
semantic, and pragmatic. Each dialogue in 
the corpus is annotated at the above five lev- 
els of analysis; synchronization among the dif- 
ferent analyses and between these and the 
speech signal is ensured by the different anno- 
tations (stored as separate files) making ref- 
erence to the same input file. This file, con- 
taining the transcription of the dialogue, is in 
turn linked to the audio file in PCM (a-low 
or u-low) format. As it will be argued in sec- 
tion 3.3, support for this structure is provided 
by the use of XML as mark-up . By 
adopting this structure, annotation layers are 
linguistically heterogeneous and mutually or- 
thogonal, so that changing one of them af- 
fects others only to a limited extent; layers 
are nevertheless indirectly related through a) 
their hinging on a common reference file (the 
"raw" text represented by the transcription 
file); b) the indirect correlation of the lin- 
guistic information they convey. This verti- 
cal modularity of the ADAM approach has 
interesting consequences for the purposes of 
reusability. A potential user of the ADAM 
Corpus is left free to select, among the pro- 
posed levels of annotation, those which best 
reflect his/her theoretical and practical inter- 
ests. (S)he can also feel the need for adding 
a new layer of information, not contemplated 
in today's ADAM realization. By the way, 
level modularity is also of theoretical inter- 
est, since most annotation schemes we know 
differ mainly in the way pieces of linguistic in- 
formation categorized, rather than in the in- 
trinsic nature of these levels. Moreover, level 
modularity seems to have a useful impact on 
our theoretical understanding of the linguistic 
phenomena at stake, since it is capable of ex- 
pressing correlations between layers, and ulti- 
mately between dimensions of linguistic anal- 
ysis. 
3.3 Annotation meta-schemes 
Horizontal customization in annotated cor- 
pora can be enhanced by implementing the 
concept of annotation meta-schemes. The 
different layers of linguistic description ira- 
pried by the concept of annotation modular- 
ity presuppose as many annotation schemes. 
As it will be made clear in section 4, for 
each of the five annotation layers envisaged 
for the ADAM Corpus, a particular annota- 
tion scheme has been designed and applied. 
However, it should be emphasized how the 
ADAM specifications do not merely amount 
to another set of ready-made, off-the-shelf an- 
notation schemes. Rather, we would like to 
focus the attention on what we call an an- 
notation meta-scheme, and on the implica- 
tions of this choice. According to our view, 
an annotation meta-scheme is a general de- 
scriptive framework in which different annota- 
tion schemes can be accommodated. In many 
cases the same unit of linguistic information 
can be annotated in different, arguably mutu- 
ally incompatible ways, which are nonetheless 
all compatible with the recommended vertical 
modularity described above: so it is better 
to provide the potential user with the pos- 
sibility of adopting any arbitrary annotation 
scheme without being forced to re-build the 
annotation from scratch or to forcefully com- 
ply with some other annotation scheme, no 
matter how standardized. To do so, it is nec- 
essary to have a representation format for the 
annotation that is general enough for com- 
peting schemes to be mutually substitutable. 
In ADAM we achieve this aim by building 
a general scheme where those features that 
are common to several competing schemes be- 
come slots or descriptive element tags to be 
associated with linguistic elements; the val- 
ues of these attributes can be any arbitrary 
set of tags. Let's consider, for instance, the 
case of pragmatic annotation. The main dif- 
ference between annotation schemes for this 
level of analysis lies in the particular types of 
dialogue act chosen rather than in the notion 
of dialogue act itself, which appears to be un- 
controversial. If, however, we adopt a scheme 
where the basic descriptive element of any ar- 
bitrarily long set of words is the general tag 
<dialogue act>, further described by an at- 
tribute type, different schemes can be applied 
11 
to the same corpus without totally discarding 
the existing annotation: a substitution in the 
set of values will be enough. It is our belief 
that enforcing this practice in the design of 
annotation schemes will bring us to more ef- 
fective corpora exchange and reuse 3 
3.4 XML-based Annotation 
In fact, actual corpus reusability also cru- 
cially depends on the physical \]ormat or mark- 
up  used for corpus encoding. The 
mark-up  used for the ,encoding of the 
ADAM Corpus is XML. XML proved to be 
the ideal candidate for a number of reasons, 
all related to corpus reusability. First, it is 
an emerging and widespread standard, which 
ensures a good degree of corpus reusability 
in the times to come. Second, because of 
its platform-independence it enhances the po- 
tential for wide circulation of the annotated 
material, together with a considerable flexi- 
bility of use. More crucially, however, XML 
proved essential for implementation of the ar- 
chitectural choices described above. Anno- 
tation modularity is supported via extensive 
use of Xlink elements (DeRose et al., 2000). 
Each XML element in the annotation files is 
actually an hypertextual link which refers to 
an element (or set of elements) in the tran- 
scription file. All annotations for each di- 
alogue are thus connected to the same in- 
put reference source (the transcription), thus 
ensuring synchronization of the different an- 
notations and still preserving their indepen- 
dence. On the other hand, the concept of 
annotation meta-scheme is implemented by 
making the XML translation of the different 
annotation schemes content-independent. In 
other words, a general preference was given 
towards representing the different ~nnotation 
tags as values of generic, scheme-independent 
attributes of XML elements. In this way the 
different annotation schemes (represented as 
different DTDs) are represented in a generic 
enough way, so that a future user of the cor- 
pus will only need to change the values of 
Sin addition, the meta-scheme can be seen as a 
tool for effective compariso n of alternative annotation 
schemes. 
the different attributes for the entire annota- 
tion scheme to be changed. We believe that 
this approach represents a further value of the 
ADAM Corpus. 
3.5 Previous and related work 
Our work builds on some important standard- 
ization efforts which were going on during the 
past few years in the field of dialogue an- 
notation (DRI, EAGLES, and MATE). We 
are also indebted to the experience gained 
in other projects using stand-off XML an- 
notation, and in particular to the MATE 
project. The multi-level markup framework 
adopted in ADAM closely reflects the MATE 
approach (Dybkjaer et al., 1998). In addi- 
tion, in our project we are using the MATE 
workbench (Dybkjaer and Bernsen, 2000) for 
visualization and information extraction pur- 
poses. However, at the best of our knowl- 
edge ADAM is the first corpus being archi- 
tecturally designed by explicitly adopting the 
concept of annotation modularity and meta- 
scheme at different levels. A recent standard- 
ization project in the annotation field is con- 
stituted by the ATLAS (Bird et al., 2000) con- 
sortium, including NIST, LDC and MITRE. 
The ATLAS architecture is based on a formal 
model for annotating linguistic data (Bird 
and Liberman, 1999). ATLAS offers a three- 
layers solution to the problem of integrating 
different data storage formats by providing 
a logical level which consists of the  
formalism and the API. The architecture we 
are proposing for the ADAM corpus is not 
a software architecture such as the one im- 
plemented by ATLAS. While the latter one 
meets the requirement of flexible and dynamic 
extension of the sofware modules, the ADAM 
architecture mainly refers to functional orga- 
nization of the different annotation layers. In 
ADAM the flexibility requirement is about 
the possible extensions of those layers. In ad- 
dition, the ATLAS architecture covers a large 
variety of possibly annotated data (not only 
linguistic data, but visual data of different 
kinds too), while ADAM is only focused on 
linguistic and speech annotations. 
12 
4 Levels of Annotation 
The ADAM's five levels of annotation were 
mainly chosen in consideration of their inter- 
est for practical applications of the annotated 
material. In spite of the number of levels con- 
sidered, and their sometimes conflicting re- 
quirements, we tried to develop a coherent, 
unitary approach to design and application 
of annotation schemes. In particular, in de- 
veloping the different annotation schemes for 
the five levels envisaged, attention was paid 
to be consistent with criteria of robustness, 
wide coverage and compliance with existing 
standards. 
4.1 The prosodic level 
For the annotation of the prosodic phenomena 
of dialogue we are adopting the meta-scheme 
for prosody annotation developed by Quazza 
and Garrido (Klein et al., 1998) within the 
MATE project. This meta-scheme allows to 
annotate the prosodic phenomena of natural 
dialogue by distinguishing the following four 
sub-levels of prosodic annotation: 
• phonetic transcription 
• phonetic representation of intonation 
• phonological representation of intonation 
• prosodic phrasing 
The four levels do not represent a fixed hi- 
erarchy. The two phonetic levels are directly 
aligned with the speech signal and in this 
sense may be considered as base levels. The 
two phonological levels keep a natural rela- 
tionship both with the base prosodic levels 
and with other linguistic units. In the actual 
use of the scheme, the levels and their links 
can be fully or partially specified. In a lin- 
guistic text-oriented analysis, prosody could 
be considered in its function, leaving out the 
details of its realization. In this case, the sole 
phonological levels may be filled and linked 
to the orthographic level of words. Com- 
plex schemes like ToBI could be used in this 
way, or simpler schemes providing labels to 
distinguish types of accents, associated with 
words, and types of intonation boundaries. In 
a speech technology context, a more signal- 
oriented approach could be adopted. In or- 
der to recognize or synthesize prosodic pat- 
terns, detailed phonetic descriptions are nec- 
essary, requiring both phonetic segmentation 
and phonetic representation of intonation - in 
terms of pitch movements or target f0 levels. 
The ADAM prosodic annotation is done at 
the level of the prosodic phrasing. The third 
section of Appendix B reports an excerpt 
of the prosodic annotation file of a human- 
human dialogue of the corpus. These are four 
dialogue turns whose text is translated in the 
first section of the Appendix. The element 
breakindex allows to encode the ToBI labels 
which constitute the values of the attribute 
type. For example, the brkndx_O07 is an- 
notated with type label 3p to mark-up an 
hesitation pause. 
4.2 The Morpho~Syntactic and 
Syntactic Levels 
The ADAM proposal for the morphosyntac- 
tic level is a two-layer annotation structure, 
containing respectively information on word 
category and morphosyntactic features (pos 
tagging), and non recursive phrasal nuclei 
(called chunks). Robustness and coverage 
were a crucial aspect in the development of 
the two schemes, in particular for what con- 
cerns i) syntactic constructions specific of spo- 
ken dialogues (ellipses, anacolutha, non ver- 
bal predicative sentences etc.), and ii) dis- 
fluencies (repetitions, false starts, trailing off 
etc.). The morphosyntactic annotation level 
encodes the following information: a) iden- 
tification of morphological words and linking 
to their corresponding orthographic counter- 
parts; b) annotation of their pos-category; c) 
annotation of morphosyntactic features (such 
as number, gender, person, tense, etc.); d) an- 
notation of their corresponding lemma. The 
partic~ar tag set, though adapted to repre- 
sentation of Italian, is compliant with EA- 
GLES recommendations (Gibbon, 1999). In 
addition, the tag set is structured into a core 
scheme, supplying basic means for annotating 
morphological information, and a periphery 
tag set, which serves the purpose of making 
13 
provision for further linguistic annotation to 
be added to obligatory information. The syn- 
tactic annotation level is built on top of the 
previous one and consists in identification of 
non-recursive phrasal nuclei (called chunks) 
and annotation of their category (Mengel et 
al., 1999). The preference given to shallow 
parsing over, e.g., phrase structure trees is 
chiefly motivated by the locality of the anal- 
ysis offered by this approach, a useful feature 
if one wants to prevent a local parsing failure 
from backfiring and causing the entire parse 
of an utterance to fail. This is particularly 
desirable when dealing with particularly noisy 
and fragmented input such as spoken dialogue 
transcripts. For an illustration of morphosyn- 
tactic and syntactic annotation, see examples 
4 and 5 in Appendix B. 
4.3 The Conceptual Level 
The annotation scheme for the conceptual 
level has been designed on the following re- 
quirements and assumptions: 
• portability: although most of concepts 
encode strictly domain-dependent infor- 
mation, the annotation scheme should be 
domain-independent as much as possible; 
• expressiveness: the scheme should al- 
low the representation of the content of 
complex dialogues; 
• minimality: a turn should be annotated 
in a unique way; 
• simplicity: the syntactical complexity 
of the concept is to be minlmized; 
• locality: the annotation should not take 
in account the history of the dialogue. 
The proposed annotation scheme takes inspi- 
ration from the so called "Frame-based De- 
scription Languages" (Cattoni and Franconi, 
1990), a well established framework in the 
field of the Knowledge Representation. In 
our annotation scheme a concept is encoded 
like a "frame", a typed structure with "slots". 
Slots represent the properties of the concept 
and its relations with Other concepts. Slots 
are encoded with the couple <slot-name, slot- 
value>: the former contains the name of a 
property, the latter either a simple value or 
a reference to another concept. This recur- 
sion allows the encoding of complex and struc- 
tured semantics information. Concepts are 
typed: different types of concepts (e.g. "trip", 
"room") encode different contents to be rep- 
resented. 
For example given the sentence to be an- 
notated "the train leaves from rome at eigth 
o'clock of monday fifteen", its conceptual an- 
notation is: 
<concept id="c_O01" ctype="trip"> 
<slot shame= "transport at ion-t ype" 
svalue =" train"/> 
<slot sname="origin" svalue="rome"/> 
<slot sname="departure-time" svalue="*c_O02"/> 
</concept> 
<concept id="c_O02" ctype="time"> 
<slot sname="hour" svalue="8:00"/> 
<slot sname="week-day" svalue="monday"/> 
<slot sname="month-day " svalue="md15"/> 
</concept> 
where the concept c_001 of type trip 
has tre slots; the slot representing the 
departure-time encodes a reference (intro- 
duced by the character '*') to the other con- 
cept c_002 of type time. 
The annotation scheme is domain indepen- 
den/: the tag set does not change when the 
domain changes since the domain-dependent 
information is encoded in the values of the 
attributes. The user is free to adopt the pre- 
ferred ontology, although a good reference are 
the symbols adopted by the C-STAR consor- 
tium (Waibel, 1996) for the inter-lingua: they 
have been developed on the basis of the ex- 
perience on six different (Asiatic and Euro- 
pean) s and this appears to guaran- 
tee a good portability inter-lingua. 
4.4 The Pragmatic Level 
For annotating the pragmatic level of dia- 
logue, we base our work on the concept of 
dialogue act. Informally speaking, a dia- 
logue act tag is a label belonging to a tag 
set which refers to a given iUocutionary di- 
mension that may be performed by uttering 
a sentence. A dialogue utterance may be an- 
notated with a dialogue act label for repre- 
senting the discourse function it plays in the 
14 
dialogue. The annotation scheme used for the 
pragmatic level of the ADAM corpus is an 
extension of both DAMSL (Core and Allen, 
1997) and SWITCHBOARD-DAMSL (Juraf- 
sky et al., 1997). The extension was not mo- 
tivated by domain, rather by the dependency 
on the dialogue type. Actually, most of dia- 
logue acts encode information that is strictly 
dependent on whether the communication is 
task-oriented, familiar, formal, and so on. So 
the inventory of dialogue acts labei should be 
sufficiently wide to cover different types of di- 
alogues, and sufficiently open to add new dia- 
logue act labels for different annotation tasks. 
For the design of the extended tag set we have 
identified the following requirements and as- 
sumptious: 
• minimality: an utterance should be 
tagged with an unique dialogue act label; 
• context-sensitiveness: each turn is 
managed by considering the previous 
turns, that is the annotation should take 
into account the history of the dialogue. 
The tag set used in the corpus is reported in 
Table 1 (see the Appendix). In Appendix B, 
Section 7 the four dialogue turns translated at 
the beginning of the Appendix are annotated 
by using the ADAM pragmatic tag set. Each 
dialogue turn is annotated as a whole by tag- 
ging the communicative level of the turn itself 
(if it is about the task or about managing the 
task, for example). Within the turn the differ- 
ent communicative intentions are labeled on 
the basis of the dialogue act tag set. 
5 Conclusions 
In this paper we have identified two ba- 
sic requirements for achieving actual cor- 
pora reusability, namely annotation modu- 
larity and use of annotation meta-schemes , 
and described how they are addressed in the 
ADAM Corpus. We claim that, for effective 
circulation and re-use of corpora, it is essen- 
tial to make provision for as many practices 
of dialogue annotation as possible, as well as 
approaches to annotation at different levels, 
instead of providing fixed levels and schemes 
of analysis, no matter how standardized. Cor- 
pora will have a chance to be reused as far as 
it will be easy and relatively inexpensive to 
adapt them to different needs and application 
purposes. Use of XML as mark-up  
is a further step toward this end. 

References 
Autori Vari 2000. SITAL: Manuale Operativo. 
Deliverable 1.1, Italy. 
Baggia, P., Castagneri, G. and M. Danieli. 
2000. Field Trials of the Italian ARISE Train 
Timetable System. Speech Communication, 31, 
pages 355-367. 
Bird, S. and M. Liberman. 1999. A Formal 
Framework for Linguistic Annotation. Techni- 
cal Report MS-CIS-99-01, Department of Com- 
puter and Information Science, University of 
Pennsylvania. 
Bird, S., Day, D., Garofolo, J., Henderson, J., 
Laprun, C. and M. Liberman. 2000. ATLAS: 
A Flexible and Extensible Architecture for Lin- 
guistic Annotation. In Proceedings of LREC 
2000, Athens, Greece. 
Cattoni, R. and E. Franconi. 1990. Walking 
through the Semantics of Frame-Based Descrip- 
tion Languages: A Case Study. In Proceed- 
ings of the Fifth International Symposium IS- 
MIS '90, pages 234--241, Knoxville, TN. 
Core, M. and J. Allen. 1997. Coding Dia- 
logues with the DAMSL Annotation Scheme. 
In Working Notes of the AAAI Fall Symposium 
on Communicative Actions in Humans and Ma- 
chines, pages 28-35. Cambridge, MA. 
DeRose, S., Maler, E., Orchard, D. and B. 
Trafford. 2000. XML Linking Language 
(Xlink). W3C Working Draft, 21 February 
2000. http ://www. w3. org/TR/xlink/. 
Dybkjaer, L., Berusen, N. O., Dybkjaer, H., McK- 
elvie, D., and A. Mengel. 1998. The MATE 
Markup Framework. MATE Deliverable 1.2. 
http://mate, hiS. sdu. dk. 
Dybkjaer, L. and N.-O. Bernsen. 2000. The 
MATE Workbench. In Proceedings of LREC 
P000, Athens, Greece. 
Gibbon, D. (ed.). 1999. Handbook of Standards 
and Resources for Spoken Language Systems. 
First supplement, EAGLES LE3-4244, Spoken 
Language Working Group. 
Ide, N. and C. Brew. 2000. Requirements, Tools, 
and Architectures for Annotwted Corpora. In 
Proceedings o/LREC 2000, Athens, Greece. 
Jura?sky, D., Shriberg, E. and D. Briasca. 1997. 
Switchboard DAMSL Labelhlg Project Coder's 
Manual. Technical Report 97-02, University of 
Colorado, Institute of Cognitive Science, Boul- 
der, CO. http://www, colorado, edu/ling/~ 
juraf sky/manual, august I. html. 
Klein, M., Bernsen, N. 0., Davi~, S., Dybkjaer, 
L., Garrido, J., Kasch, H., Mengel, A., Pirrelli, 
V., Poesio, M., Quazza, S. and\[ C. Sofia. 1998. 
Supported Coding Schemes. MATE Deliver- 
able i.I. http://mate .his. sdu.d.k. 
Mengel, A., Dybkjaer, L., Garrido, J., Heid, U., 
Klein, M., Pirrelli, V., Poesio, M., Quazza, S., 
Schiffrin, A. and C. Sofia. 1999. MATE Dia- 
logue Annotation Guidelines. MATE Deliver- 
able 2.1. http://mate, nis. sdu. dk. 
Waibel, A. 1996. Interactive Translation of Con- 
versational Speech. Computer, 29(7):41-48. 
