Proceedings of the 2nd Workshop on Building Educational Applications Using NLP,
pages 85–92, Ann Arbor, June 2005. c©Association for Computational Linguistics, 2005
Situational language training for hotel receptionists 
 
Frédérique Segond, Thibault Parmentier Roberta Stock, Ran Rosner 
Xerox Research Centre Europe The Marathon Group 
Meylan, 38240, France  Tel Aviv, Israel 
segond@xrce.xerox.com rstock@marathon-group.net 
 
Mariola Usteran Muela 
Grupo gdt. 
Sevilla, Spain 
mustaran@grupogdt.com 
 
 
Abstract 
This paper presents the lessons learned in 
experimenting with Thetis1, an EC project 
focusing on the creation and localization 
of enhanced on-line pedagogical content 
for language learning in tourism industry. 
It is based on a general innovative ap-
proach to language learning that allows 
employees to acquire practical oral and 
written skills while navigating a relevant 
professional scenario. The approach is en-
abled by an underlying platform 
(EXILLS) that integrates virtual reality 
with a set of linguistic, technologies to 
create a new form of dynamic, extensible, 
goal-directed e-content. 
1 Credits 
The work described in this paper has been sup-
ported by the European Commission in the frame 
of the eContent program2. 
2 Introduction 
Thetis focuses on the creation and localization of 
enhanced on-line pedagogical content for language 
learning in tourism industry. It is based on a gen-
eral approach to language learning that allows em-
ployees to acquire practical oral and written skills 
                                                           
1 THETIS – Training for hotel employees to interact in situa-
tions (EDC- 42052). www.thetis-project.org. 
2 http://www.cordis.lu/econtent/ 
while navigating a relevant professional scenario. 
The approach is enabled by an underlying platform 
(EXILLS3) that integrates virtual reality with a set 
of linguistic technologies to create a new form of 
dynamic, extensible, goal-directed e-content 
Thetis has two aims: 
• Test the value of linguistic technologies 
for on-line language learning.  
• Localize and repurpose existing e-content 
for English and vocational blended train-
ing material that was first designed for 
CDROM in order to offer it on-line. The 
material is meant to develop oral compre-
hension and reading skills in order to en-
able end-users to communicate with 
English speaking customers. It will be 
used for the continuous vocational training 
of professionals in this sector through the 
Internet as well as on-site.  
In the following sections we present the following:  
• The Thetis scenario and the technologies 
applied 
• Aspects of content adaptation  
• the results of the users’ evaluation  
• the lessons we have learned both regarding 
the value of the technologies (including the 
linguistics technologies), and the peda-
                                                           
3 See  www.exills.com 
85
gogical value of such an innovative ap-
proach to language learning 
3 Thetis : scenario and linguistic tech-
nologies 
As the general technical architecture behind Thetis 
has already been described in details in (Segond 
and Parmentier 2004) (Brun et Al. 2002) we just 
give an overview of the entire system and concen-
trate below on the description of the Thetis sce-
nario and on the linguistic technologies that have 
been integrated into the system.  
Thetis integrates virtual reality and linguistic tech-
nologies in a web application in order to propose a 
truly e-learning solution that can be used both syn-
chronously and asynchronously. Our motivations 
for applying these two components are the follow-
ing: 
• Virtual reality offers a cognitive context 
and promotes interaction. 
• Linguistic technologies offer autonomy to 
the students by showing them concepts, 
giving them assistance to understand word 
meanings within particular contexts by 
presenting various examples, providing 
feedback on their skills (during chat ses-
sions as well as through exercises or even 
free activities). 
The notion of scenario is central to Thetis. The 
scenario allows the users to act in typical work 
situations such as introducing themselves, reading 
emails, searching for and understanding informa-
tion, ordering a meal, and interacting with col-
leagues and customers. The lessons include  
traditional contents such as grammar rules, exer-
cises, and speech acts.  The users interact either via 
chatting or during the different activities proposed. 
The Thetis scenario has been explicitly designed 
for people working at hotels’ reception desks. 
 
3.1 Scenario 
Students and tutors enter a virtual reality scene 
either all together at the same time (synchronously) 
or whenever they want (asynchronously). They are 
then in a virtual hotel where they are given roles 
that correspond to the different prototypical hotel 
situations listed above. All students have to play 
their everyday job, hotel desk receptionist. They 
interact with customers. The tutors can choose the 
role they play: customers, fellow students or tutors. 
The resulting system can be used either individu-
ally from any place with access to the Internet or 
collectively, all or some students being in the same 
location (open class, hotel etc.) 
The customers are robots and the students are ava-
tars that work at the reception desk. 
The robots are 3D-human representations that have 
been programmed in advance. They can invite us-
ers to chat, react to certain stimuli such as prede-
fined lists of words or movements of others in the 
3D-scene. 
The avatars are non-programmed 3D-human repre-
sentations of users. The users can decide where to 
go in the virtual scene, with whom they want to 
interact, what to say. 
The application is composed of several virtual 
rooms. Each of these rooms corresponds to a dif-
ferent scenario related to the receptionists work 
tasks. 
The students’ avatars interact either with the robot-
customers, or among themselves. The language of 
the interactions is exclusively the language being 
learned i.e. English, (which is automatically 
checked by the system). In order to encourage in-
teractions among the students and to strengthen the 
playful aspect of the course, the students are asked 
to work in groups of two or three. Each group of 
two or three students enter rooms at the same time, 
and interact individually with different robot cus-
tomers. Since they are in the same room they are 
also able to communicate among themselves, in the 
language they learn, when they have difficulties. 
The tutor can also be present. There is a common 
back room where all the students can go any time 
to interact with the rest of the student group, be it 
because they need help or because the tutor asks 
for a meeting. 
In order to call for interaction students and tutors 
can either type text in the chat window or, in some 
86
specific cases (previously defined in the scenario) 
record their answer and send the speech file to the 
tutor.  
The figure below is a snapshot of the different ac-
tivities in the different virtual rooms. Traditional 
linguistic exercises, cards with grammar or speech 
act hints are associated to the activities of each 
room. 
 
 
In the Virtual World students always have access 
to a phone, a fax, a reservation book, computers, 
documents internal to the hotel (regulations, menu, 
price lists, tourist information etc.). 
During the different activities, the students interact 
with robots that play different roles (customers, 
travel agents, taxi driver etc.). Dialogues between 
students and robots work as follows: the students 
type some text in the chat box, and the robots reply 
orally. Robot participations to dialogues are pre-
recorded sound files. The robots are configured to 
reply to stimuli like a new learner arriving in a 
scene asking for a private chat, etc. For instance, 
when a student asks a question, the robot can 
automatically select the most appropriate answer or 
action based on the dialog’s steps or based on 
keywords found in students’ written production. 
Examples:  
“Could you show/give me your identity card?” 
would produce the action of displaying a win-
dow with the image of an identity card. 
“How long will you stay?” would produce an 
answer like “three nights”. 
When robots ask questions, the students’ answers 
can be used (or not) to select the next move in the 
dialogue. 
Examples:  
Robot: “Do all the rooms have an Internet con-
nection?” Student: “No” would imply a specific 
notice on the reservation fax to require an Inter-
net connection. 
3.2 Linguistic Technologies 
In terms of tools and resources available in the vir-
tual world, the students have access to real hotel 
documents (including a list of hotel services, prices 
and regulations, real identity documents from dif-
ferent countries, etc.)as well as to linguistic tech-
nologies. 
The linguistic tools include the following: 
 
• A comprehension help that 
provides students with the 
most appropriate contextual 
translation of any word or ex-
pression. Comprehension help 
is crucial in speeding up both 
the students’ comprehension 
and written production. 
 
• A linguistic tool box that pro-
vides the students with cus-
tomizable services which 
allow them to parse and to tag 
their own production in order 
to check its correctness. 
 
• A morphological analyzer that 
gives the students access to 
conjugations or declinations. 
 
• A language guesser that auto-
matically prevents students 
87
from interacting in another 
language than the one they 
learn i.e. English. 
 
4 Good quality content  
The other interest of Thetis is to create and adapt e-
content to offer it on-line rather than on CD-ROM. 
One of the most difficult issues that e-learning  
faces is global delivery, especially with the chal-
lenges of localization and repackaging content for 
different audiences. Europe has been active in cre-
ating high quality pedagogical content for years 
and is genuinely concerned about languages.  
Over the past few years, multimedia publishers 
have developed electronic content for language 
learning of very high pedagogical value and very 
good quality. With the arrival of e-learning, they 
see their work changing and do not always have 
the necessary expertise and resources to take the 
e-learning route.  
Indeed e-learning revolutionizes the training do-
main as it provides personalized, real-time training 
and communications programs, rather than a “one 
size fits all” approach. To take advantage of the 
potential of e-learning, multimedia publishers need 
to develop expertise in Internet technologies in 
general. To be able to put their content on-line they 
need to work in partnership with companies that 
have Internet expertise and document processing 
technologies. On the other hand, technology pro-
viders in the area of e-learning have given very 
little attention to the quality of the content that will 
be offered, to the pedagogical aspect of e-learning 
as well as to the question of what needs to be 
changed in order to use the Internet to train people. 
Most of the e-learning companies have tried to sell 
platforms to content providers rather than trying to 
work in partnership with them in order to create 
good-quality e-learning solutions. It is certainly not 
enough to integrate content that has been built for 
another purpose (to teach in face to face meetings 
for example, or even for CD-ROM distribution) 
with technologies to make a good training module. 
Pedagogical aspects must be considered and tech-
nologies should support pedagogy rather than be-
ing the main objective. 
 
4.1 eContent adaptation 
Thetis particularly concentrated on defining the 
virtual environment and the content of the training, 
adapted to the needs of the employees in a hotel 
reception. 
The main functions of receptionists are presenting 
a positive image of the hotel and assisting with all 
aspects of guest service; they act as buffers be-
tween the customers and management of the hotel. 
The training needed to take into account their main 
duties, including the following: 
Answering enquiries regarding hotel services 
and registration by letter, telephone and in per-
son 
Making room reservations 
Registering arriving guest and assigning rooms 
Responding to guests’ enquiries, requests and 
complaints 
Using computerized or manual systems to 
compile and check daily record sheets, guest 
accounts, receipts and vouchers 
Presenting statements of charges to departing 
guests and receiving payment 
Additional duties, which may be required by 
small hotels. 
It was also important to consider the skills needed 
by these kinds of workers, even when they need to 
communicate with customers in languages differ-
ent from theirs.  
Good communication skills and a neat appearance 
are essential for hotel receptionists. They need ba-
sic analytical skills and experience with word 
processing equipment and computers. Good judg-
ment is necessary, as well as ability to solve prob-
lems in a bold way and with determination, since 
they may have to deal with difficult people, as well 
as with emergency and security problems.  
88
But one of the best assets is the knowledge of, at 
least one, second language. To communicate flu-
ently with customers in a foreign language, solve 
problems and deal with complaints, turn them into 
opportunities is the aim of this project, together 
with the development of key competences for em-
ployees working at hotel reception desks, such as 
efficiency, courtesy, initiative, capability of work-
ing in team and communication skills, among oth-
ers. 
What is different in Thetis, compared to “tradi-
tional CD-ROM content” building are the scenar-
ios through which the courses will be delivered to 
hotel receptionists. So the content had to be appro-
priate to the scenarios so that it is adapted to the 
learning context, and it had to be localized follow-
ing the target region of the training. 
The pedagogical content is realized on different 
media like texts, audio support, and videotapes. 
For each medium different activities are available 
in order that the students can practice a specific 
grammar point, an idiomatic structure or phrase. 
There are several kinds of exercises each corre-
sponding to a lexical area and/or speech act. This 
means that a student is immersed in a work situa-
tion associated with specific difficulties in gram-
mar and vocabulary and (s)he is provided with 
exercises helping to develop her/his competencies.  
The objective is not to find a typical situation in 
order to learn a specific grammar point or an idio-
matic structure. The process is rather first to list the 
typical work situations the hotel receptionists are 
confronted with and afterwards to identify the lan-
guage difficulties associated with each situation 
and, in a third time, to select the media, the con-
tent, the type of activity that is the best suited to 
help the students to deal with the work situation. 
One other difficulty is to propose a sequence of 
working activities that also match with the lan-
guage and vocabulary level of the student. 
Besides the scenario, the students are also provided 
with some classical content in order to help them 
to meet the needs they could be confronted with 
outside typical training situations. Such classical 
content is listed in the points below: 
• a grammar book containing a list of all 
the sentence patterns and forms that are 
used in the program with explanations 
of their use and form; 
• a lexicon with English headwords, ex-
amples of usage and explanation or 
translation into native language. Stu-
dents also have the option of hearing 
the pronunciation of new items; 
• student records which allow students to 
monitor their own progress, 
• technical help in order to present the 
characteristics of the delivery web ap-
plication. 
This existing content has been adapted to the tour-
ism industry  
Although the basic content will be the same for all 
target audiences, the addition of elements that ap-
peal directly to each target user group makes the 
product attractive and motivating.  
In terms of language practice, it is also possible to 
place an emphasis on those items that cause par-
ticular problems to specific language groups. 
As far as already existing content related to tour-
ism scenarios is concerned, they had to be adapted 
to our purposes: some pedagogical activities had to 
be reversed because they had been developed in 
order to help a customer, a tourist in our case, to 
interact with other people i.e. receptionists, wait-
ers/waitresses, passers-by. The main task was to 
propose mainly the same activities but for training 
receptionists. 
5 Users’ tests  
The content of the course focuses on teaching Eng-
lish to hotel receptionists. Therefore, the users 
were faced with real situations that occur in hotels 
within the context of their everyday work envi-
ronment. The objective then was to test the course 
within this target group in order to analyse its po-
tential use, and possible adaptation to any socio-
professional domain, as well as extension to any 
language.  
The main elements of THETIS that needed to be 
tested are the following: 
• Technical issues: linguistic tools, virtual 
reality 
89
• Pedagogical effectiveness: quality of con-
tent, game aspects, etc. 
• Work related competencies: hotel envi-
ronment and situations, level of interest, 
relevance, etc. 
There were two major groups of users involved in 
these tests: 
• Students 
• Teachers or tutors. 
All students involved (18) are currently students of 
a Hotel Management vocational training program 
organised by the Andalusian Entrepreneurial Con-
federation (CEA).  
The profile of the group was the following: 80% 
were between 20 and 24 years old, 20% were be-
tween 25 and 30 years old. All of them have a uni-
versity degree in Tourism. 80% had previous work 
experience in hotels or tourism related jobs. 50% 
had an intermediate competency level in English, 
17% low intermediate and 33% high intermediate. 
A team of 3 teachers was also involved in this test; 
one of them was present at the location of the test 
activities and the other two were doing participated 
from another country, which shows the flexibility 
that distance learning offers. 
Each session followed the same structure: 
• Presentation of the three partners. 
• Presentation of the THETIS project. 
• Goals and objectives of the session. 
• Explanation of the different elements of 
THETIS: 
o Chat 
o Virtual Reality 
o Dictionary 
o Verb conjugation 
o History 
o Notes 
o Grammar  
o Exercises 
After the initial explanations, the session continued 
by proposing different activities in which the stu-
dents had to do the following: 
• Chatting among students - learning 
games. A reservation form had to be filled 
out. Each form contained some informa-
tion of different customers. We assigned 
one specific guest to each student, so they 
had to discuss and ask each other about the 
missing data. 
• Chatting with the teacher - role playing 
exercises. The teacher acted as a customer 
and started different dialogs with the stu-
dents based on typical hotel situations such 
as making reservations or asking for direc-
tions.  
• Exercises. The students were given time to 
practise four categories of exercises: read-
ing comprehension, grammar, listening 
and vocabulary.  
• Interacting with the robots. The students 
had to take part of the pre-set dialogs with 
the customer-robots at the virtual reality 
scenario and choose the right answer in 
every situation.   
Once the testing activities described were finished, 
the students were asked to write their comments by 
answering a complete questionnaire 
Most of the students agreed that using this program 
continuously as part of complete language training 
would be very beneficial. Some students appreci-
ated most that it helped remember expressions and 
vocabulary they had learnt before, so it was good 
practice. Most of them acknowledged the acquisi-
tion of the specific language used in hotels and the 
expressions needed in situations that occur at the 
front desk.  
Regarding the scenario itself all students consid-
ered that what happens in a hotel and the daily 
functions that a receptionist must carry out are very 
90
well represented. Some typical appreciations are 
the following: 
“It is a very creative way to reflect the different 
situations happening in a hotel” 
“Doing different things at the same time is very 
common at hotels and getting to learn English by 
having the same experience is crucial” 
“It is good that we get to see the nice things that 
normally happen but also the not so nice ones, like 
dealing with upset customers” 
Basically, the main advantages of THETIS appear 
to be how realistic the situations are, the entertain-
ing and fun part of learning and the interaction 
with the teacher and the other students.  
As for the weaknesses, these appear to be related to 
the fact that Thetis is still a pilot product and some 
technical and content details have yet to be final-
ized.  
Regarding the elements the testers found most use-
ful during the learning process, we found a broad 
range of answers and every tool was named. Re-
garding the linguistic technologies students men-
tioned the contextual dictionary look-up as the one 
they made the most use of. However, the elements 
that received a higher appreciation were the inter-
action with robots (virtual reality) and with the 
teacher (chat), because of the entertaining aspect 
they both bring, the realistic feature they portray as 
well as their practical application.  
In order to have a complete and enriched informa-
tion report, we asked for the teachers’ opinion as 
well.  
It was perceived that through this program students 
are given different ways to consolidate what they 
learn which makes it quite efficient. It is also im-
portant to say that there is a clear need to have 
teacher-student interactions to make it more suc-
cessful pedagogically, since this element gives the 
students the opportunity to use free and authentic 
language and to be corrected in real time. 
Since interactions between the teacher and the stu-
dents are allowed through chat - which is some-
thing the students enjoyed a lot - it was remarked 
that the teacher needs to have the possibility to cor-
rect mistakes. So for this tool to be more useful 
and not damaging the role-playing situation, there 
is a need to find a way to send the students that 
feedback, for example, by making it appear in an-
other color. 
It is also important to prepare different possible 
scenarios/exercises for the teachers to develop 
through the chat. Therefore, it appears important to 
prepare a teacher’s guide. 
6 Conclusion: lessons learned  
In general, we found that the creation of content 
for a Thetis like type of concept was not a trivial 
and easy task. Indeed Thetis provides a solution 
that is strongly web oriented in the sense that it 
insists on interaction on the web, personalization 
and information access. As a consequence the type 
of content that fits these requirements is radically 
different from ones that already exist for other sup-
ports such as, for instance, CD-ROMs. We did not 
expect this adaptation to be so time consuming. It 
turns out that while a “traditional” type of CD-
ROM content can be integrated in the form of ex-
ercises within the Thetis solution, it is necessary to 
build new types of content in the form of scenarios. 
Indeed the notion of a scenario, which is central to 
Thetis as well as its strong point, means one that 
covers both learning a language and learning a 
work practice. This scenario should also be attrac-
tive and evolving enough so that it retains students’ 
interest and motivation. The borderline is some-
times difficult: we need to be careful not only to 
propose games where students learn how to do 
their work but not so much how to speak a lan-
guage. Moreover, the scenarios, since they take 
place on the web and in a virtual world, cannot be 
like the ones that are proposed in a face to face 
course. Interaction is different. Robots plays a cen-
tral role in pushing people to interact with each 
other, and so does the tutor. How to make the best 
pedagogical use of these two aspects? What is the 
type of dialogues that work better in terms of 
pedagogical purposes?  
During the testing we had had various comments 
on the specific features offered by the system as 
well as suggestions for improving it; all of them 
would need to be further tested. These suggestions 
91
come from different perspectives: from the tutors 
and from the students. 
As far as the tutors are concerned, while they find 
the idea really interesting pedagogically, they also 
generally consider it difficult to interact with sev-
eral students at the same time. For instance correct-
ing several students in the chat is almost 
impossible. Therefore they ask for tools to help 
them in this process. These tools would be like the 
ones that already exist in current management sys-
tems courses, and they would have to be enriched 
with some specific functions to deal with on-line 
interaction. Teachers usually liked a lot the lan-
guage guesser module as it forces students to inter-
act in the language they learn and relieve them 
from checking that students do not interact in their 
own language. 
On the opposite side, the students liked and used a 
lot the contextual on-line dictionary because they 
really want to play and interact and are forced to 
do so in the foreign language. The students defini-
tively liked the virtual reality and the game aspect. 
The strongest point of Thetis is undoubtedly the 
interactions based on a virtual reality scenario as 
compared to e-mail interactions, chats or forums. 
Probably owing to the virtual world shy people 
tend to participate much more than usual. This is 
also the feedback that we got from tutors concern-
ing their students.  
While students enjoyed the game very much, this 
would need to be tested on the long run as Thetis 
certainly benefits from the “novelty effect”. 
Both teachers and students would like to be able to 
speak rather than just to write. However, it is not 
clear how this type of interaction would be possi-
ble in virtual reality with 15 people not being able 
to see each other. During the project, we identified 
strong points as well as difficulties and gaps. The 
main difficulty is in the type of content that fits 
Thetis’ philosophy. Indeed, building such content 
requires more time than we first expected. While 
part of the content can be just repurposed and lo-
calized from already existing materials, a com-
pletely new type of content needs also to be 
created from scratch. Users made interesting sug-
gestions regarding the integration of new technolo-
gies such as speech processing or tools for the 
tutors. These would require further testing.  

References  
Brun C., Parmentier T., Sandor A., Segond F., 2002, 
Les outils de TAL au service de la e-formation in 
Multilinguisme et le traitement de l'information, Fré-
dérique Segond Ed., Hermès, pp. 223-251  
Frase, L., R. Almond, J. Burstein, K. Kukich, R. 
Mislevy, K. Sheehan, L. Steinberg, and K. Singley 
and M. Chodorow. (2002). In H. O'Neill and R. Perez 
(Eds.) Technology Applications In Education: A 
Learning View. Lawrence Erlbaum 
Kashny M., Les usages des Technologies d'Information 
et de Communication par des enseignants dans un 
dispositif de formation tutorée en langues vivantes 
étrangères. Une approche ergonomique. Thèse Uni-
versité Pierre Mendès France, 2001, Grenoble. 
Kearns M., Satinder Singh C., Litman D., and Howe.J. 
CobotDS, 2002: A Spoken Dialogue System for 
Chat. In Proceedings of the Eighteenth National 
Conference on Artificial Intelligence (AAAI-2002), 
Edmonton, Canada. 
Kindley, R., Scenario-Based E-Learning: A Step Be-
yond Traditional E-Learning, 
http://www.learningcircuits.com/2002/may2002/kind
ley.html. 
Kukich, K , 1992, Technique for automatically correct-
ing words in text, ACM Computing Surveys (CSUR) 
Volume 24 ,  Issue 4,  ( 377 – 439) 
Paulsen, J. (2001-section 4). "Authentic Online Target 
Language Reference Resources." New Era Trends 
and  Technologies in Foreign Language Learning: An 
Annotated Bibliography - Interactive Multimedia 
Electronic Journal of  Computer-Enhanced Learn-
ing, April 2001, Wake Forest University. 
Puren, C. Histoire des méthodologies de l'enseignement 
des langues , Clé International, Nathan, 1988, Paris. 
Segond F, Parmentier T, 2004 NLP serving the cause of 
language learning Proceeding of International Work-
shop on eLearning for Computational Linguistics 
and Computational Linguistics for eLearning 
COLING , Geneva.
