PROJECTED GOVERNMENT NEEDS IN HUMAN LANGUAGE 
TECHNOLOGY AND THE ROLE OF RESEARCHERS IN MEETING 
THEM 
Helen M. Gigley, Ph.D. 
Head, Human Computer Interaction Laboratory 
Naval Research Laboratory 
CODE 5530 
Washington, D.C. 20375-5337 
ABSTRACT 1 
After a brief discussion of current government uses and needs 
for human language technology, this paper will discuss the 
uses which will probably survive in the future as well as 
attempt to define some anticipated ones. Business uses, 
military uses, and education/training receive the most focus 
today. Current state of the art technologies are providing 
initial resources to enable better management of the above. 
Suggested critical efforts to make the technologies more 
appropriate for the tasks at hand in the government will be 
discussed. A tentative view of the role of researchers, in 
providing the basis to obtain adequate critical natural 
language technologies, will be presented. 
functional needs. Inter- and intra-agency coordination and 
information dissemination are critical to efficient and 
proper functioning of our government. When is natural 
language technology a suitable solution to maximize the 
effectiveness in these efforts? Some examples will shed 
light on the nature of our current human language 
technology capabilities and provide guidance on what still 
is needed even within our current government operations. 
Information Management Functions of the 
Government: 
• Access to personnel. 
• Dissemination of information. 
1. CURRENT GOVERNMENT NEEDS 
IN HUMAN LANGUAGE 
TECHNOLOGY 
• Acquisition of information. 
• Protection of access to information. 
The functions of government fall into three general 
classes: 
1. business, both external and internal; 
2. military, beyond the business uses; and 
3. educational/training responsibilities. 
One very important question related to determining current 
human language technology needs is to look at these three 
broad areas and to assess what technology exists today and 
how effectively it can be applied. Another equally 
important question which must be considered is, "Does 
technology help or hinder the functionality of the 
government?" 
1.1 Human Language Technology and 
Current Government Business Needs 
Government business needs are coextensive with 
operational business needs of the United States. Access to 
information, interpersonal communication, and joint work 
environments within government operations are all critical 
1 The opinions and assertions in this paper are those of 
the author and are not to be construed as official or 
reflecting the views of the Department of the Navy. 
Communication Functions of the Government: 
• External Communication Functions -- 
need to communicate with the general 
public and provide suitable responses not 
only in content but in presentation style. 
Internal Communication Functions -- 
communication of protocols, procedural 
modifications, relevant day to day 
operational changes. 
* Multi-language interaction facility where 
the interaction may be in any of several 
languages as well as translation 
capabilities. 
* Joint work environments where multiple 
decision making is enhanced via multi- 
tasking environments at distributed 
locations. 
287 
Education/Training 
Government: 
Functions of the greatly depend on how education needs for the entire U.S. 
are met. 
• Required to maintain up-to-date knowledge of 
system operations and procedures. 
• New personnel must be brought up to speed 
on how to function. 
The bottom line for government-needed human language 
technology is that it should provide alternative resources 
for communication which increase the functional precision 
and response to enhance productivity. 
1.2 Human Language Technology and 
Current Military Needs 
Military human language technology needs include many 
of the logistical and business-type needs just mentioned. 
In addition, military needs include factors of real-time 
performance and secure access which are not absolute 
within business ones. The functions of military systems 
assume a different purpose. Presently, they also insist on 
a role for humans which many business functions attempt 
to minimize; these systems are Human-in-the-Loop 
Systems. Military systems enhance decision making, 
include dissemination of information having various 
forms, and often must work in real-time response arenas, 
under secure conditions. How to maximize the 
information capacity within Human-in-the-Loop decision 
support systems is a critical need. Human language 
technology has a vital part to play in providing this 
capability. Language capabilities need to function 
precisely, correctly, and efficiently in multiple modalities. 
Military Decisions require real-time language 
facilities. Military decisions require integration of 
many types of information which exist in disparate forms. 
Language technologies provide a possible addition to 
current methods to increase the information available in 
stressful, short-response-time decision situations. A 
primary concern is to determine what language 
technologies to employ and under what conditions. 
Security implications for human language 
technology appear chiefly as constraints on its 
use. Speech is available to all within range of its 
audible production. Its internal form may also be 
accessible. What language technologies maintain security 
and how might language technologies provide added secure 
access constraints? Voice recognition can be a key, but is 
it sufficient? 
1.3 Human Language Technology and 
Education / Training Needs 
Education and training needs in government function do 
not appear to be diminishing. Even though the work force 
might shrink, its demography is constantly changing. 
Meeting the challenge of developing suitably capable 
individuals from the current and future work force will 
Currently, demands on the government work force beyond 
physical plant maintenance and related in-house 
infrastructure function, minimally require language 
fluency in English. All positions require literacy in 
English. Second language fluency and literacy also have 
become a staffing criteria for some positions. All of these 
facts present problems for many individual's qualifications 
to work for the government. 
Lack of language fluency and literacy will create a 
possible problem in the near future as older workers retire. 
Many of the tasks government workers perform are 
language based, whether accessing information, providing 
it, or collecting it. An issue is can we use technology to 
replace them? Do we want to? And if not, how will we 
guarantee that they can do the job? Training systems for 
language skills can be one viable solution. These 
systems will be multimodal and require all aspects of 
language function. 
As an additional need for training and education, we will 
need to continually update the skills and facility of 
personnel on the job. This is a critical need in the 
military. Having language technology available to 
develop the training methodologies is critical. What will 
documents be like? How will one use the available 
language tools? Answers to these questions are unknown 
but are hinted at in the current multimode capabilities 
which can now be integrated on our workstations. 
When considering education and training for the 
government, it should be the case that developed 
technology is also relevant and available to education 
systems throughout the country. 
2. CURRENT TECHNOLOGIES--ARE 
THEY HELPING OR HINDERING? 
As previously mentioned, successful application of 
technology can only be measured by its efficiency and 
precision in use. Otherwise we are in a state of 
technology only for technology's sake. Let's look at 
some currently available technologies which have been 
introduced into government. What can we learn from their 
introduction? 
Immediately, one is struck by the fact that more often 
than not a technology is introduced and deployed publicly 
without any study as to its impact. We need to consider 
the impact of such adoptions before rather than after they 
get put into place. One only has to keep in mind the 
work of Gray, John, and Atwood (1) which demonstrated 
through analysis and data evaluation that introducing 
technology does not necessarily increase effectiveness. 
We are in that situation now with our language 
technologies. Caution is advised and careful scrutiny of 
the effects should be considered when adopting any new 
288 
technological invention. A case in point is voice mail. 
My personal opinion is that it is one of the greatest cogs 
in U.S. competitiveness today. 
Imagine yourself as a foreign business person calling a 
U.S. top company to place an order and having to sit for 
two minutes or more to reach the appropriate ordering 
department while menu options are read sequentially 
Often, in voice mail, one is not even sure the menu item 
selected is the appropriate one. The language technology 
relies on hardware selection which is sequential and 
numeric, but is this necessary? 
We need to develop methods for studying the impact of 
potential technologies and their effect on function. We 
need to consider the problem from two viewpoints, the 
function being served within the government and equally 
important, the functional demands it places on non- 
government or citizen users of the technology even 
without their consent. Language technology more than 
other types of technology subtly conveys an institution's 
attitude toward its clientele. This interaction for many 
government functions is vital. 
Functional transaction speed is an important factor of any 
technology that should always be considered. We need to 
minimize the impact of the technology on the efficiency 
of human functioning, and still be able to get necessary 
things done. 
3. ROLE OF RESEARCHERS IN 
MEETING LANGUAGE TECHNOLOGY 
NEEDS 
Researchers' primary focus continues to be studying basic 
scientific principles and using them to expand the 
capabilities of language technology regardless of modality. 
Evaluation and determination of constraints on 
applicability of the technology also need to be developed. 
Furthermore, researchers need not only to consider their 
contribution as a basic scientific result but must be aware 
of the potential use of the result within our society, both 
favorably and adversely. 
Suppose many access functions are completely assigned to 
machines and that as a user, you need to obtain some vital 
information. When the machine responds that you are not 
permitted access even though you know you are, how will 
you be able to get to a responsible agent to correct the 
problem if you are talking to a machine? Will the 
responsible agent also be a machine? How will you find 
out it is a machine if all interactions are via language 
technology? Ultimately, with automatic language 
processes in place, how would you determine how to 
correct the situation? This is bordering on the question of 
reality and virtual environments and their detectability if 
and when we really produce language technologies that are 
indeterminable from real speakers. Is this possible? 
Even now we can produce speech that sounds like a given 
individual based on sampling of the person~s actual 
speech. Where will we draw the limits of such use? 
Where will be draw the limits of appropriateness for 
language technology usage in general? There are vital 
research issues to be addressed regarding these concerns. 
4. CONCLUSION 
Government needs for language technology encompass 
those of general business as well as more demanding 
military constraints. The technologies we currently have 
in place have not always been introduced expeditiously and 
beneficially. This does not mean that we should abandon 
all efforts to develop language, but instead points to a 
more cautionary view of accepting language technology as 
a solution. The basic efforts now underway will advance 
the role of language technology and will enhance 
functionality of government. We as researchers need to 
accept some of the responsibility for making sure the 
contributions to government function in the future remain 
as positive enhancements rather than harmful ones. 
5. REFERENCES 
1. Gray, Wayne D.; John, Bonnie E.; and Atwood, 
Michael E.; "The Precis of Project Ernestine or An 
Overview of a Validation of GOMS," CHI 92 Conference 
Proceedings, Monterey, CA, pp. 307-312, 1992. 
289 
