ON-LINE GENERATION OF TERMINOLOGICAL DIGESTS IN LANGUAGE TRANSLATION: AN AID IN TERMINOLOGY PROCESSING

• Document Type: CREST
• Collection: General CIA Records
• Document Number (FOIA) /ESDN (CREST): CIA-RDP83M00171R001800120013-1
• Release Decision: RIFPUB
• Original Classification: K
• Document Page Count: 49
• Document Creation Date: January 4, 2017
• Document Release Date: December 14, 2001
• Sequence Number: 13
• Publication Date: September 15, 1975
• Content Type: REPORT

Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
Research
RC 5624
(#24297)
9/15/75
Communications
29 pages
ON-LINE GENERATION OF TERMINOLOGICAL DIGESTS
IN LANGUAGE TRANSLATION:
AN AID IN TERMINOLOGY PROCESSING
E. O. Lippmann
IBM Thomas J. Watson Research Center
Yorktown Heights, N. Y. 10598
Yorktown Heights, New York
San Jose, California
Zurich, Switzerland
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
RC 5624 ON-LINE GENERATION OF TERMINOLOGICAL DIGESTS
(#24297) IN LANGUAGE TRANSLATION:
9 / 15 / 7 5 AN AID IN TERMINOLOGY PROCESSING
Communications
29 pages
IBM Thomas J. Watson Research Center
Yorktown Heights, N. Y. 10598
ABSTRACT: Evaluation of technical and scientific
translations dealing with complex subject matter has
shown that a) the majority of errors made by the
translators involve terminology, and b) the translators
spend a great deal of their time searching for the
correct equivalents of technical-scientific terms to be
translated. This paper describes a technique of
generating terminological digests speedily on terminals
connected to a coaputer in order to overcome these
impediments and aid the translator in streamlining the
translation production process. A terminological
digest represents the glossarial framework of a
translation, a unique dictionary constructed
automatically for the text to be translated. The user
can produce a terminological digest by invoking the
appropriate program on his terminal, entering on the
keyboard the terms he wishes to have looked up. All
terms entered are immediately retrieved from an
up-to-date scientific-technical dictionary and provided
with target language equivalents and other pertinent
information. At the user's option, the dictionary
entries may be presented singly, as a list in the order
of entering the terms (e.g., the order in which they
occur in the text to be translated), or as an
alphabetically-sortel list. These lists may be
displayed, typed oit, or printed and saved as
"minidictionaries" for a particular field.
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
This report has been submitted for publication elsewhere and
has been issued as a Research Report for early dissemination
of its contents. As a courtesy to the intended publisher, it
should not be widely distributed until after the date of outside
publication.
Copies may be requested from:
IBM Thomas J. Watson Research Center
Post Office Box 218
Yorktown Heights, New York 10598
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
rAl7z I
As has been emphasized in a variety of works on
translation [1,2,3,4,5,6], mere knowledge of the general
vocabulary and the grammars of two or more languages do not
necessarily enable an educated person to translate
scientific, economic, technical or legal literature. To
translate specialized material of this type correctly, an
acquaintance with the subject matter and a command of the
proper terminology is required. For the professional
translator, one of the most exasperating aspects of
translating modern texts on a variety of subjects is the
inordinate amount of searching and learning time he has to
expend to evaluate special terms employed by the authors of
source language documents. It has been claimed that up to
60% of a conscientious translator's work time is consumed in
tracking down proper terminological information (7].
Not only is the volume of new terminology increasing
rapidly, but also the looseness of using the technical
vocabulary is growing among various specialists in the field
[8]. Added to the proliferation of unabbreviated specialized
terminology must be the ever-enlarging usage of acronyms and
initialisms, particularly in information processing, which
frequently cannot be decoded even by experts in a given
field without the aid of a proper dictionary [9]. The
number of acronyms formally collected in the U.S. has now
swelled to nearly 1C3,000 terms [10] and is continuing to
grow [11,12,13,14]. However, even the most up-to-date
printed dictionary cannot maintain a rate of speed parallel
with the burgeoning growth of specialized terminology [15].
That dictionaries are obsolete is especially evident in
computer systems terminology [16], which is proliferating at
a rate which could perhaps be compared to the proliferation
of higher-level languages in the programming field.
It is, of course, true that as the use of higher-level
languages can effectively assist non-data-processing
professionals in communicating with a computer in their own
professional jargon [17], so special terminology can make
communication for a scientist or technologist more efficient
and convenient when interacting with fellow specialists
[18]. And just as the development of programming languages
is likely to go forward [19], one would expect a continual
growth of new terminology, given the' rate at which
technological concepts are developed. In fact, Jean Sammet?s
statement concerning one of the major reasons for the
proliferation of programming languages can be applied
virtually unchanged to the terminological explosion
[17: 310]: "Some of the causes and motivations behind the
development of these languages rest in quirks of human
nature rather than technological progress or lack thereof.
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83MOO171 R001800120013-1
PAGE 2
Thus, as long as people find it fun to develop languages, as
long as they want something which is specifically tailored
exactly to their needs, and as long as they are going to
find picayune faults with the existing languages, there is
very little that technical progress can do to reduce the
number of languages".*
Thus, although the translator may be familiar with the
scientific or technical subject matter and its fundamental
terminology, the number of terms which take on different
shades of meanings, cover new concepts when used by various
authors, or are outright neologisms, may tend to confound
the translator in his attempts to conscientiously determine
the correct translation 120,21,22]. Compounding this
confusion is the fact that at the rate at which new terms
are coined to communicate technological development, even
the best technical and scientific bilingual or multilingual
dictionaries are out of date by the time they appear in
print [15,22,23,24,25,26].
To overcome the impediments in scientific and technical
translation arising from the dispersion and inaccessibility
of terminology, a variety of suggestions have been made
[7,8,27,28,29,30,31,32] which may be summarized as follows:
1. Constructive efforts should be made by closer
coordination between the terminology boards of the
various disciplines, technical associations and
laboratory information services to standardize jargon
appearing in printed form.
2. A coherent approach should be established to ensure that
all meanings of a scientific/technical term are recorded
in the dictionary with reference to the field to which
the meaning applies.
3. The rate at which special terns are to be added to
existing monolingual glossaries and bilingual or
multilingual dictionaries should be increased, eventually
contributing to a reduction in the translator's overall
search time for specialized terminology.
---------------
* Of the large number of synonymous computer technology
terms, only one example is given for "the function to
combine object modules to produce a single program," as used
by the data processing community: linkage editor, link
editor, builder, winder, loader, linking loader, relocatable
loader, linkage loader, linking (relocatable) loader,
collector, job loader [ 16 ].
Approved For Release 2008/03/03: CIA-RDP83MOO171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
4. The latest dictionaries and reference manuals should be
at the translator's immediate disposal so that the proper
terms can be looked up speedily and conveniently.
5. The translator should be able to rapidly obtain a
terminological digest of a text to be translated, i.e., a
list of searched-for terms extracted from the dictionary
in order of textual occurrence and/or alphabetical
occurrence.
This paper does not address itself to objectives 1 and
2, viz. closer coordination between the parties to
standardize jargon and to ensure that all meanings of
technical terms are documented. Such endeavors are the
domain of the technical experts and technical writers.
Obviously, the mere availability of computers does not
constitute a remedy for controlling the jargon explosion or
for solving human communications problems. However, the
opportunity to interact with computers on-line in a
terminal-oriented environment does provide the potential for
finding effective solutions to objectives 3 through 5
without undue emphasis on the willingness of the
technologists to document and standardize their specialized
terminology expeditiously.
Approaches to meeting objectives 3 and 4 have been
described in detail elsewhere [33,34,35,36,37]. In summary,
objective 3 can be met by giving the user access to a
time-shared computer system supporting data bases and
dictionary maintenance programs to allow bilingual or
multilingual dictionary generation and updating. Objective 4
involves dictionary lookup and browse programs for
presenting dictionary entries in hardcopy or video format on
appropriate terminal devices or on high-speed printers
capable of producing high quality copy. Zoncomitant
requirements would entail (1) context editing, making it
easy both to change stored translation texts and
dictionaries and to input new ones, and (2) formatting, to
produce a variety of professional-looking translation and
dictionary layouts.
The approach which is described on the following pages
involves objective 5 and is oriented toward reducing the
manual searching and sifting time which the translator
requires to determine the proper terminology in a
translation, and thereby toward increasing his productivity.
The approach deals with the semi-automatic generation of
terminological digests of texts to be translated, a method
whose basic ideas can be traced back to the work on
text-related glossaries by the West German Bundessprachenamt
[38].
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
r n%3 L'+ +
The Bundessprachenamt (1bersetzerdienst der Bundeswehr)
analyzed the types of errors in technical-scientific
translations and found, among other things, that the rate of
mistranslated and untranslated specialized terminology
increased proportionately to the increase in the technical
difficulty level of source texts. on the greatest difficulty
level, terminological errors accounted for 62.1% of the
translation error total (where the range of error types
included such categories as orthographical mistakes,
punctuation, wrong German preposition, inflectional errors,
English word order, text inaccuracies and omission of
information). Concurrent time and productivity studies were
conducted indicating that translators using dictionary lists
with special source/target language terms which were
exclusively related to the technical-scientific texts to be
translated could reduce the error rate by approximately 40%
and increase their productivity by over 50% as compared to
their colleagues who had well-equipped conventional
technical-scientific libraries and the consultation of their
colleagues at their disposal [391.
As indicated above, a terminological digest is a list
of terms extracted from a main dictionary in the order of
textual occurrence or alphabetical occurrence, i.e., the
desired terminological framework of a text to be translated.
The text to be processed may be of arbitrary length. Figure
1 represents an example of a text portion to be translated.
Figure 2 shows the automatically-produced terminological
digest of this text in the order in which the desired terms
occur, and Figure 3 shows the terminological digest in
alphabetic order.
The user has at his disposal a terminal, either a
typewriter or a video display unit (Figure 4), which may be
connected over regular telephone lines to a computer. After
having turned on the terminal and, if required, dialed-up
the computer and made the connection, production of a
terminological digest is achieved by first invoking the
appropriate program and then entering the terms one wishes
to have looked up at the keyboard.
If the source text is stored in machine-readable format
in the system, as is the case for text produced by a variety
of text-processing systems, it may also be displayed and
examined by rolling it up or down ("scrolling") on the
surface of a display screen. However, depending on the
translation job, material to be translated may only be
obtainable in non-machine-processable format, in which case
it cannot be stored in the machine. Moreover, even though
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
CA %3.D .,
source text may be available in machine processable format,
it may not be accessible on a particular computer system
with dictionary files, because of lack of storage space or
because of computer installation policy. In fact,
secretarial copy aids for holding manuscripts may make
manual page flipping competitive with, and perhaps even more
economical than, scrolling of machine-readable source text.
Temporary lack of a terminal, or assignment
considerations between translators and typists, may call for
separation of the task of identifying special terms for
terminological digest production and of entering these terms
in the system. For example, the translator may wish to
encircle the desired terminology in the source text and
submit it to a typist who may input these terms on-line or
off-line, by terminal or typewriter (e.g., an MC/ST or
MT/ST).
All terms entered are immediately looked up by the
program in a comprehensive dictionary, which may be oriented
toward a special subject area, and provided with target
language equivalents and other germane information. At the
user's option, the terms may be presented singly (i.e.,
immediate display of a term in dictionary context as soon as
the term is entered or selected), as a list in the order of
entering (normally the order in which they occur in the text
to be translated), or as an alphabetically-sorted list. The
terminological digests may also be saved and used
repetitively, automatically edited as any other translation
document [33], directed to offline output devices for
high-speed printing or punching, or transmitted by
telecommunications to other terminals or computers.
The dictionary storage organization is designed to cope
with the growth potential of multilingual dictionaries,
where an extremely large number of entries may eventually be
accumulated. Details of this organization and of the
associated dictionary lookup procedure are described
elsewhere (33]. During execution of a single dictionary
lookup, the desired source language term is compared with a
table of source language terms whose corresponding
dictionary entries occur at fixed intervals in the
dictionary file. Dependent upon a high/low/equal compare
result, appropriate routines are called to move the desired
entry into a storage buffer for a video display unit or a
typewriter terminal. Unless the buffer is completely
occupied, adjacent entries are retrieved and packed into it
until it is filled. This dictionary excerpt is then flashed
onto the display screen. If a video display unit is not
used, the entry is printed out on the typewriter.
For a terminological digest consisting of a list of
entries, the same dictionary lookup procedure is used except
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
rA%3.U G
that the system delays the search until all terms to be
looked up have been entered. At this point the lookup
procedure is applied iteratively to each term of the entered
collection in order to obtain the corresponding dictionary
entry. When this process has been completed, the resulting
entries are displayed, typed out or printed as a group.
Depending upon the translation task or the size of the
digest, the user can cause immediate digest display,
type-out on his terminal, or printout on a high-speed
printer, or he can save it as a file for future translation
work and possible additional termiaological and statistical
investigations. At his option, he may also cause the system
to generate an alphabetically-sorted digest.* This could be
especially helpful if more than one translator works on one
large text, each concentrating on sections for which
terminological digests in text order and a global
alphabetically-sorted digest encompassing the entire text
(Figure 5) are generated. Although different sections of the
same text are translated by different translators,
consistency of terminology is maintained by usage of the
machine dictionary, ensuring that the same terms are always
translated in the same way [40]. A specialist in an
editorial function may decide that some editing of the
terminological digest is required before it is used by the
individual translators. For example, a manual on unit
record machines in electronic data processing might contain,
among other things, the terms "card stacker" (English/German
dictionary equivalents: Kartenablage, Ablagefach), "vertical
line" (English/German dictionary equivalents: Senkrechte,
Vertikale), and "level" (English/German dictionary
equivalents: Ebene, Ordnung, Stufe, Pegel, Niveau). All of
those translations may be valid within the same text.
However, the specialist may conclude
a) that "Ablagefach" must be edited out of the
terminological digest to maintain uniformity in equipment
terminology,
b) that although either "Senkrechte" or "Vertikale"
could be edited out, they may be used interchangeably by the
translators because of their complete unambiguity in German,
and
c) that "Ebene, Ordnung, Stufe, Pegel, Niveau" must be
left untouched, since each translation may have to be used
even within a small stretch of text, so that the choice must
therefore be left to the discretion of the translator. By
using the automatic editing facility [33], changes to the
terminological digests can be made extremely rapidly.
* see Appendix` for memory layout and sort mechanism in
terminological digest generation.
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
-_y r
I For purposes of protection and recording of
references and changes, main storage is
divided into blocks of 2,048 bytes, each
block having an address that is a multiple
of 2,048. A control field, called "key in
storage", is associated with each block of
storage.
The key in storage has the following
format:
~ACC.IFIRICI
The bit positions in the key are allo-
cated as follows:
Access-Control Bits (ACC): Bits 0-3 are
matched against the four-bit protection key
whenever information is stored, or whenever
information is fetched from a location that
is protected against fetching.
Fetch-Protection Bit (F): Bit 4 controls
whether protection applies to fetch-type
references: a zero indicates that only
store-type references are monitored and
that fetching with any protection key is
permitted; a one indicates that protection
applies to both fetching and storing. No
distinction is made between the fetching of
instructions and of operands.
improper stor;
storing and f
improper fetchir,
When protection
access, the key in s
the protection key
request for storage
permitted only when
matches the protecti
said to match when the
of the key in storage
protection key or whi
is zero. A fetch is
keys match or when
storage is zero. The
summarized in the tat
tion Action."
r----------------T
I Conditions
F-------- T--------
IBit 4 ofl
(Key in I Key
(Storage IRelat?
--------+-----
0 I Mat
0 IMisr
I 1 I Ma
I 1 (Mir
F------- _J
fExalanatior
I Reference Bit (R): Bit 5 normally is set (Match Th
to one each time a location in the corres- I th
ponding storage block is referred to either I tc
for storing or for fetching of information. I Pr
This bit is associated with dynamic address I
Change Bit (C): Bit 6 is set to one each INO Acc,
time information is stored into the corres- fetc
ponding storage block. This bit is asso- not
ciated with dynamic address translation. I progi
I contei
The key in storage is not part of I tion a.
I addressable storage. The program can L---------------
I explictly place information in all seven
I bits of the key by SET STORAGE KEY, and the Summary of Protec
contents of the key can be inspected by
I INSERT STORAGE KEY. Additionally, the When the access
instruction RESET REFERENCE BIT provides a by the CPU, and
I means of inspecting the reference and protection key of
change bits and of setting the reference as the comparand
bit to zero. the CPU occupies b
PSW. When the
channel, and prote
I PROTECTION tection key assoc
tion is used as t!
I The protection facility is provided to tion key for an I
protect the contents of main storage from n bit positions
I destruction or misuse caused by erroneous word (CAW) and i
1 or unauthorized storing or fetching by the 0-3 of the chanr
program. It provides protection against as a result of
Figure 1: Example of a text portion to be translated.
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
PAGE S
key in storage: Speicherschluessel [5/370] [SYST]
protection: Schutz, Protektion, Beschuetzung; Schutzzoll [LEG]
reference:: Hinweis, Bezugnahme Nachschlagen; nit Verweisungen versehen;
durch Verweisungen finden
main storage: Hauptspeicher [SYST]
block: Block (SYST); Satzblock [SYST]; blocken (SYST]; blockieren;
in Bloecke formes
addressr Adresse; Ansprache; adressieren; anreden
multiple: Vielfaches; vielfach
control field; Kontrollfeld (5/370]? Sortierfeld [SYST]
key in storage: Speicherschluessel (5/370] [SYST]
storage: Speicher [SYSTI; Speicherung; Lagerung; Lagermiete [COM1
bit: Bit [SYST], Binaerziffer [SYST]; Bohropitze (MECH); Schluesselbart;
kieines Stueckchen
allocate: zuordnen, zuteilen, anweisen
access-control bits Zugriffa-Steuerungsbit [S/3701
match: abqleichen [SYST]; verbinden, paaren, paarwaise verbinden;
passend verbindes [MECH]; Gleiche(r, s), Zusammenbringen
protection key: Schutzschluessel [S/370] (SYST)
fetch: Abruf (SYST); abruten [SYST]; abholen
fetch-protection bit: Abrufaschutzbit [5/370) (SYST)
apply: zutreffen; anwenden, verwinden; auftragen (MECH]; bewerben
fetch-type
?s? FEHLT IN W EBTERBUCH (1ST ABFRAGE PALSCH BUCHSTABIERT7) ???
indicate: anseigen. angeben, andeuten
store-type
??? FEHLT IN 1OERTERBUCH (IST ABPRAGE PALSCH BUCHSTABIERT7)???
fetch: Abruf [SYST); abrufen (SYST); abholen
store: speichern [SYST]; lagern, aufepeichern; Speicher [SYST];
Lager, Magazin
instruction: Instruktion [SYST); Anweisung, Belehrung; Lehr*
operand: Operand (Parameter in einer Inetruktion);
operand specification - Spezifikation fuer einen Operanden
reference bit: Hinweisbit (5/370) [SYST)
refer: hinweisen, verweisen, Bich beziehen; sich wenden
associate: in Verbindunq stehen; assoziieren, vereinigen
dynamic address translation: dynamieche Adressumsetzung (8/370] [SYST]
change bit: Veraenderungsbit [S/370] [SYST]
dynamic address translation: dynamische Adressumsetzung [8/370) (SYST)
addressable: adressierbar
storage: Speicher (SYST); Speicherung; Lagerung; Lagermiete (CON)
SET STORAGE KEY
?s? FEHLT IN WOERTERBUCH (IST ABFRAGE FALSCH BUCHSTABIERT7) ?s?
contents: Inhalt
inspect: pruefen, untersuchen, beaufsichtigen
INSERT STORAGE KEY
??? FEHLT IN WOERTERBUCH (IST ABFRAGE FALSCH BUCHSTABIERT7) ??s
RESET REFERENCE BIT
??? FEHLT IM WOERTERBUCH (IST ABFRAGE FALSCH BUCHSTABIERT7) ???
protection: Schutz, Protektion, Beschuetzunq; Schutzzoll (LEG]
main storage; Hauptspeicher [SYST]
store: speichern [SYST]; lagern, aufspeichern; Speicher [SYST];
Laqer, Magazin
fetch: Abruf (SYST]: abrufen [SYST]; abholen
protection: Schutz, Protektion, Beschuetzunq; Schutzzoll [LEG]
Figure 2: Terminological digest of a text portion to be
translated into German.
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
PAGE; y
access-control bit: Zugriffs-Steuerungabit [S/370]
address: Adrease; Ansprache; adressieren; anreden
addressable: adressierbar
allocate: zuordnen, zuteilen, anweisen
apply: zutref fen; anwenden, verwenden; auftragen [MECH]; bewerben
associate: in Verbindung stehen; assoziieren, vereinigen
bit: Bit [SYST], Binaerziffer [SYST]; Bohrspitze [MECH]; Schluesselbart;
kleines Stueckchen
block: Block [SYST); Satzblock [SYST]; blocken (SYST); blockieren;
in Bloecke formen
change bit: Veraenderungsbit [5/370] [SYST]
contents: Inhalt
control field: Kontrollfeld [5/370]; Sortierfeld [SYST]
dynamic address translation: dynamische Adressumsetzung (S/370] (SYST)
fetch: Abruf (SYST); abrufen [SYST]; abholen
fetch-protection bit; Abrufsschutzbit (S/370] [SYST]
fetch-type
*** FEHLT IM WOERTERBUCH (IST ABFRAGE FALSCH BUCHSTABIERT?) ???
indicate: anzeigen, angeben, andeuten
INSERT STORAGE KEY
*** FEHLT IN WOERTERBUCH (IST ABFRAGE FALSCH BUCHSTABIERT?) ?**
inspect: pruefen, untersuchen, beaufsichtigen
instruction: Instruktion (SYST); Anweiaung, Belehrung; Lehre
key in storage: Speicherschluessel (5/370) [SYST]
main storage: Hauptspeicher (SYST)
match: abgleichen [SYST]; verbinden, paaren, paarweise verbinden;
passend verbinden [MECH]; Gleiche(r, s), Zusammenbringen
multiple: Vielfaches; vielfach
operand: Operand (Parameter in einer Instruktion);
operand specification - Spezifikation fuer einen Operanden
protection: Schutz, Protektion, Beschuetzung; Schutzzoll [LEG]
protection key: Schutzschluessel [5/370] [SYST]
refer: hinweisen, verweisen, sich beziehen; sich wenden
reference: Hinweis, Bezugnahme Nachachlagen; mit Verweisungen veraehen;
durch Verweisungen finden
RESET REFERENCE BIT
*** FEHLT IM WOERTERBUCH (IST ABFRAGE FALSCH BUCHSTABIERT?)
SET STORAGE KEY
??? FEHLT IM WOERTERBUCH (IST ABFRAGE FALSCHBUCHSTABIERT?)
storage: Speicher [SYST]; Speicherung; Lagerung; Lagermiete [COM]
store: speichern [SYST]; lagern, aufspeichern; Speicher [SYST];
Lager, Magazin
store-type
*** FEHLT IM WOERTERBUCH (IST ABFRAGE FALSCH BUCHSTABIERT?) **?
Figure 3: Alphabetically-sorted terminological digest of a
text portion to be translated into German.
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
rtivZ I V
Figure 4: Basic types of typewriter and video display terminals.
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
PAGE 11
TERMINOLOGICAL DIGEST WITH TERMS IN
SEQUENCE OF OCCURRENCE IN DOCUMENT PARTS
TRANSLATOR A
T.D. I
termk termcT
term, termlT
termr termrT
term, termrT
T.D. II
term, term,T
termI term lT
term, termlT
term term cT
termI term rT
termm termmT
T.D. III
termI term,
T
termb termbT
terms termcT
termr termrT
termd termdT
T. D. IY
term, termrT
term, termnT
termz termzT
term, termaT
termx termzT
ALPHABETICALLY-SORTED
GLOBAL TERMINOLOGICAL DIGEST
term, term,T
termb termbT
terms termcT
termd termdT
termz termzT
PART I
termk.... term,.... termr
....term, ...............
..........................
....term, ...............
PART II
term.. .term I . . term,... term .. .
. term I ..................
..........................
termm ............ .
PART III
......term ..............
. termb...... term,....... .
termr ..................
...............termd .....
PARTLY
..............term,....
term, .....................
termz .....................
term, ......... termx ..... .
NOTE: Subscript T denotes the translation of a term; e.g.,termkT is the translation of termk
Figure 5: Team of translators working cooperatively on a large
translation task, using terminological digests.
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
PAGE 1Z
Producing a terminological digest by computer online
should be maximally simple in order to adapt to the needs of
an inexperienced computer user, such as the translator,
interpreter, terminologist, lexicographer, editor, or
typist. Figure 6 illustrates how a user can employ a video
display terminal to generate a terminological digest. The
system allows the user various options as to how to proceed,
prompting him to enter his choice prior to terminological
digest generation. In Figure 6, after having selected a
sorted version of the terminological digest, the user has
begun entering the desired terms, working from a printed
text document.
If the source text is available in machine-readable
format, it can also be automatically retrieved and displayed
on the screen. Figure 7 shows how a portion of the English
source text of a manual, which was originally produced by
automatic typesetting, is displayed on the video screen
(first 22 lines of display). The user can move the entire
text of the manual up or down oa the screen, in fact viewing
the text as through a window, looking for terms whose
translation he wishes to know. Wheaever such a term occurs,
the user eaters it via ,the keyboard, at which time it is
displayed at the left of the bottom line of the screen and
the term within the text is brightened (i.e., displayed in
double intensity, verifying to the user that it has been
selected for terminological digest generation). The user
can continue entering as many terms as desired, thereby
creating the terminological skeleton of the text to be
translated. Hitting the ENTER key of the keyboard an extra
time causes production of the terminological digest, which
can then be displayed (Figure 8k) or printed on an attached
typewriter (Figure 8B) or a high-speed printer (Figure 8C).
Of course, the user need not display the text to be
translated on the video screen, if he prefers to work from
its printed version. Moreover, a text may not be available
in machine-readable format, or the user may employ a
typewriter terminal where it may be inefficient to print out
portions of the text intermittently and then enter the terms
to be looked up.
During terminological digest generation the user may
flip back and forth between his input and the
already-entered terminological digest terms, e.g., to scan
for possible similarities or redundancies of terms. He may
also delete terms already selected.
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 system-->
system-->
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
Figure 6: Snapshot of a video display screen during on-line
generation of a terminological digest (current system
environment is indicated to the user in double
brightness on right-Fan. bottom line).
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
of tines to be inserted,,- st,
number is not spenfied..,~Gr
between I ;;L
and the i Of ih,
then tone that, follow may be speclfied;.~
and n.# you must specify, the filetype
Vs encountered, the message EOF REACHED is typed* The tist I
inserted becomes- the new current line. if the record tet%th of
records in the f ite containing the data to be insirtod excteds : that
~fhe file being edited, it error message is typedi and the GETFILE is
executed. if shotter, the records are padded to the record It ert9th
th~ffle being edited and inserted it the file,
Figure 7: Display of a text portion of a computer manual with
term "parameter list" being entered.
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
Figure 8A: Display cf a portion of an English-German terminological
digest showing terms in order of textual occurrence.
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
edit:-zum Druck aufbereiten [SYST], aufbereiten [SYST]; AUFBEREITEN
ZUM DRUCKEN (Instruktion); redigieren, revidieren, edieren
omit: weglassen, unterlassen; uebersehen; versaeumen
parameter list: Parameterliste [SYST]
positional: stellenbedingt, stellenabhaengig, positionsbedingt
filetype: Datei-Typ [VM/370]
filemode: Datei-Modus [VM/370]; Datenbestandsart
verification: Pruefung [SYST], Bestaetigung [SYST]; Beglaubigung,
Bescheinigung, Beurkundung
record length: Satzlaenge [SYST], Datensatzlaenge [SYST];
Laenge eines Dokumentes
pad: auffuellen, polstern; Polster, Kissen; Puffer [MECH]
Figure 8B: Portion of an English-German terminological digest
shcwing terms in order of textual occurrence,
typed out on CMC/ST.
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
edit: zum Druck aufbereiten , aufbereiten ; AUFBEREITEN
ZUM DRUCKEN (Instruktion); redigieren, revidieren, edieren
omit: weglassen, unterlassen; ueberseben; versaeumen
parameter list: Parameterliste
positional: stellenbedingt, stellenabhaengig, positionsbedingt
filetype: Datei-Typ
f ilemode: Datei-Modus ; Datenbestandsart
verification: Pruefung , Bestaetigung ; Beglaubigung,
Bescheinigung, Beurkundung
record length: Satzlaenge , Datensatzlaenge ;
Laenge eines Dokumentes
pad: auffuellea, Fclstern; Polster, Kissen; Puffer
Figure 8C: Portion of an English-German terminological digest
showing terms in order of textual occurrence,
printed on an IBM 1403 Printer.
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83MOO171 R001 800120013-1
YAUZ 10
A SPECIAL
TERM?
LOOK
AT TERM_ DIGEST
LIST UP TO THIS
TERM. DIGEST
,GENERATION?
"RETRIEVE" TEXT
WITH SPECIAL TERM
"DELETE" TERM
"TRANSFER" TO TERM.
DIGEST LIST
"EXIT 11
NOTES:
1. Instructions in quotes may be
entered by pressing a program
function key.
2. On exit, a terminological
digest will be produced if
at least one term has been
selected.
Figure 9: Flow of control of a terminological digest generation
process.
'SELECT'
TERM FOR
LOOK
AT MORE
TEXT?
LOOK
AT PRECEDING
TEXT ? ,
Approved For Release 2008/03/03: CIA-RDP83MOO171 R001 800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
If he has at his disposal a display terminal with
program function keys, which allow communicating
instructions to the system in a most simple way by merely
pressing the key, the user may depress one of six keys for:
1. Moving text upward on the screen;
2. Moving text downward on the screen;
3. Flipping screen "pages";
4. Retrieving a screen "page" with a typed term (or
string of characters) ;
5. Deleting a tern already entered in the
terminological digest list;
6. Transferring to the terminological digest to examine
the terms thus far entered.
Having transferred to the terminological digest
environment, the program functions are symmetrical, i.e.,
the keys have the same meaning for the terminological digest
list except that pressing the sixth key will transfer the
user back to the point in the text where he left off before
transferring to the terminological digest list. The user
may also enter any of the six instructions on the keyboard
rather than hitting the appropriate program function key;
this is mandatory if he uses a keyboard without such keys.
A typed instruction must be preceded by a > sign to signal
the system that this is not a term to be looked up. Figure 9
represents the flow of control of a terminological digest
generation process for a source text accessible through the
system.
From a human factors engineering point of view, the
keyboard of the display terminal, which provides the main
contact of the user with the computer system, basically
operates like a typewriter but, in several ways, offers
greatly improved performance for a translator or editor who
may be a non-typist. The keys react immediately to being
pressed by the user and so quietly that someone directly
adjacent to him may not be audibly aware of the operation.
Visual communication is made easy by the display of
large-sized characters within sufficient context on the
screen. Since typing on the keyboard "prints" characters on
the screen instead of on paper, error correction is greatly
improved. By moving the cursor* to the error (or any
characters to be changed) and keying in the correct
* a_movable underscore marking the position on the screen
that the character entered from the keyboard will occupy
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
rh Vii GN
characters, previous information is overlaid by the new
data. Changing information in the middle of text can also
be quickly performed with the aid of the cursor: for
example, when data are inserted or deleted in the middle of
text, the immediately following data are automatically
shifted forward in the case of insertions, and automatically
contracted in the case of deletions. Moreover, using the
ERASE INPUT key causes the information just typed (but not
yet entered) to be blanked, while using the CLEAR key
immediately clears the entire display without causing a
disturbance within the system.
Since, at the user's option, the translation of a
queried term may be immediately displayed as soon as the
term is entered, the system could very well be used by
interpreters during consecutive as well as simultaneous
interpretation. As described in [ 41 : 154 ],
It is generally believed that an interpreter
cannot consult reference works or colleagues as a
translator can. This is only partially true. I
have always found it useful to have pertinent
dictionaries (whether general or specialized) in
the booth, which can be consulted by one's
boothmate or by the interpreter himself. Since an
unknown or vaguely known word is likely to crop up
more than once, it should be looked up in one of
three ways; (a) by the boothmate immediately after
it occurs; (b) by the interpreter during his rest
period; or, if (a) is not possible and recourse to
(b) would mean risking a second or even third
encounter with the problematical word, then (c) by
the interpreter himself while he is still
interpreting. Boothmates, too, may be queried
with an inquisitive look and a note can often be
slipped to a coworker in a neighboring booth. Such
consultations should of course never interrupt the
flow of the interpretation. An interpreter should
also be alert to difficult words which a boothmate
or other colleague may have to interpret and
should therefore not hesitate to slip appropriate
notes at the proper time. This is often the case
when one of the interpreters has specialized
knowledge and can sometiues even anticipate the
appearance of a difficult word or expression with
amazing accuracy.
Thus, a display terminal, which has great flexibility
with r3gard to siting and no air-conditioning requirements,
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
could be a valuable aid for an interpreter who has to
consult special dictionaries extremely rapidly for difficult
terms encountered during a discourse. Because on-line
lookup is generally faster than flipping pages [42], such
queries need not infringe upon the flow of the interpreter's
speech. In addition, all terns looked up can be
simultaneously saved by the system in a file representing,
in effect, the terminological "minutes" of a meeting.
Therefore, it is conceivable that rapid on-line lookup of
specialized terminology may enhance the process of
interpretation and increase the fidelity of the
interpreter's work.
Whenever possible, interpreters are urged to collect
all documentation required for interpretation before every
conference and scrutinize this material for specialized
terminology to prepare its translation. Moreover,
just as the translator, the interpreter should
build up a glossary of technical terms both for
his own use and that of his colleagues. These can
be circulated among the interpreters at the end of
each session and are of course kept for future
conferences. Delegates can also usually be
queried after the session on difficult terms
(41: 155).
Thus, most interpreters and translators have their
private terminology lists in addition to the official
reference material. The system and equipment described will
permit the user to rapidly input, update and display
terminology lists and dictionary excerpts via the display
unit and/or typewriter-terminal.
The author would like to express his appreciation to
Warren J. Plath and Mark Pivovonsky for their comments and
criticisms, which were invaluable in the preparation of this
paper. A similar note of gratitude is due to William I.
Bertsche, President of the American Translators Association,
as well as to Professor Etilvia Arjona, Monterey Institute
of Foreign Studies, California, whose interest has greatly
contributed to the write-up of this work.
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83MOOl71 R001800120013-1
t k.IZA GL
[1] D. L. Gold, ,on Quality in Translation," babel, vol.
XVIII, No. 1/1972, pp. 10-12.
[2] D. L. fold, "On Quality in Translation II", babel, vol.
XVIII, No. 4/1972, pp. 29-30.
[3] J. Maillot, "Terminologie et traduction," Meta Journal
des traducteurs, vol. 16, go. 1-2, March-June 1971, pp.
75-91.
[4] E. A. Nida,, Toward a Science of Translatioq, E. J.
Brill, Leiden, Netherlands, 1964, pp. 145-155.
[5] H. W. Sinaiko and R. W. Brislin, "Evaluating Language
Translations: Experiments on Three Assessment Methods,"
Journal of Applied Psychology," vol. 57, No. 3, 1973,
pp. 328-334.
[6] H. W. Sinaiko, "Verbal Factors in Human Engineering:
Some Cultural and Psychological Data," Ethnic Variables
in Human Factors Engineering, A. P. E. Chapaais, ed.,
The John Hopkins Press, Baltimore, 1975, pp. 159-177.
[7] K. H. Brinkmann, "Uberlegungea zum Aufbau and Betrieb
von Terminologie-Datenbanken als Voraussetzung der
maschinenuntersttltzten Ubersetzung," Nachrichten far
Dokumentation, vol. 25, No. 3, June 1974, pp. 99-105.
[8] R. N. Basu, "Barriers to Effective Communication in the
Scientific World," IEEE TRANSACTIONS ON PROFESSIONAL
CO'IMUNICATION, vol. PC-15, No. 2, June 1972, pp. 3C-33.
[9] R. C. Moser, Space=Age Acronyms, Abbreviations and
Designations, IFI/PLENUM, New York, 1969.
[ 10 ] E. T. Crowley and R. C. Thomas, Acronyms and
Initialisms Dictionary, Gale Research Company, Book
Tower, Detroit, Michigan, 1973.
[11] E. T. Crowley and R. C. Thomas, New Acronyms and
Initiaiisms, (Supplement to [9 ]) , Gale Research Company,
Book Tower, Detroit, Michigan, 1974.
[12] 5. T. Crowley and R. C. Thomas, New Acronyms and
Initiaiisms, (Supplement to [ 10 ]) , Gale Research
Company, Book Tower, Detroit, Michigan, 1975.
[13] B. Pugh, Second Dictionary of Acronyms D Abbreviations,
Archon Books, The Shoe String Press, Inc., Hamden,
Connecticut, 1974.
Approved For Release 2008/03/03: CIA-RDP83MOOl71 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83MOO171 R001800120013-1
PAGE 23
[14] R. de Sola, Abbreviations Dictionary, American Elsevir
Publishing Company, Inc., New York, 1974.
[15] E. A. Lacy, "Special Dictionaries for the Electronics
Engineer," IEEE TRANSACTIONS ON ENGINEERING WRITING AND
SPEECH, vol. EWS-6, No. 1, September 1969, pp. 31-35.
[16] A. P. Sayers, Ed., Operating Systems Survey, The COMTRE
Corp., Auerbach Publishers, New York, 1971.
[17] J. E. Sammet, "An Overview of programming languages for
specialized application areas," AFIPS Spring Joint
Computer Conference. AFIPS Press, Montvale, New Jersey,
vol. 4D, 1972, pp. 299-311.
[18] J. M. Lufkin, "Generalization and Interpretation of
Science and Technology," IEEE TRANSACTIONS ON
PROFESSIONAL COMMUNICATION, vol. PC-15, No. 4, December
1972, pp. 108-111.
[19] F. B. Thompson and B. H. Dostert, "The future of
specialized languages," AFIPS Spring Joint Computer
Conference, AFIPS Press, Montvale, New Jersey, vol. 40,
1972, pp. 313-319.
[20] J. A. Bachrach, "An Experiment in Automatic Dictionary
Look-up," The Incorporated Linguist, vol. 13, No. 2,
April 1974, pp. 47-49.
[21] U. FSrster, "Der Sprachberatangsdienst," babel, vol.
XVIII, No. 2/1972, pp. 24-38.
[22] R. Herzog, "Die Anwendung computer-linguistischer
Methoden bei der Kompilation von Fachworterbrichern,"
BeitrHge zur Linguistik and Informationsverarbeitung,
No. 18, July 1970, pp. 26-40.
[23] J. A. Bachrach and L. Hirschberg, "Une troisieme
version du DICAUTOM," Actes de la 26me conf 4rence
internationals sur le raitemeat automat gue des
langues, Grenoble, 23-25 aoat 1967.
[24] T. Longyka, "Technical Translation and Industrial
Terminology," The ATA CHRONICLE, vol. II, No. 1, January
1973, pp. 6-9.
[25] C. J. Hyman, German-English Egglish_German Electronics
Dictionary, Consultants Bureau, New York, 1965.
[26] J. Horn, Computer and Data Processing Dictionary and
Guide, Prentice-Hall, Inc., Englewood liffs, New
Jersey, 1966.
Approved For Release 2008/03/03: CIA-RDP83MOO171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83MOO171 R001800120013-1
PAGE 24
[27] R. N. Basu, "A Jargon Explosion," New Scientist, vol.
29, March 24, 1966, p. 788.
[28] National Academy of Sciences and National Academy of
Engineering, SCIENTIFIC AND TECJNICAL COMMUNICATION - A
Pressing National Problem and Recommendations for Its
Solution, National Academy of Sciences, Washington, D.
C., 1969.
[29] R. A. Evans, "But That Word Really Means...," IEEE
TRANSACTIONS ON RELIABILITY, vol. R-23, No. 3, August
1974, p. 129.
[301 M. Para, "Computerized multilingual Word Banks Can
Provide Terminological Assistance to International
Standards Organizations," UNESCO Symposium on
International Cooperation in Terminology, INFOTERM,
Vienna, April 1975.
[31] F. Kertesz, "How to Cope with the Foreign-Language
Problem: Experience Gained at a Multidisciplinary
Laboratory," Journal of the American Society for
------
Information Science, vol. 25, No. 2, March-April 1974,
pp. 86-104.
[32] i. Horn, "Ubersetzungsverfahren mit textbezogener
Abfrage," Der Spraghmi.ttler, vol. 8, No. 1, January
1970, pp. 17-20.
[33] E. D. Lippmann, "An Approach to computer-aided
Translation," IEEE TRANSACTIONS ON ENGINEERING WRITING
AND SPEECH, vol. BWS-14, No. 1, February 1971, pp.
10-33.---
[34] E. 0. Lippmann and W. J. Plath, "Time-sharing and
Computer-aided Translation," THE FINITE STRING, vol. 7,
No. 8, October 1970, pp. 1-4.
[35] J. Schulz, "Le systeme TEAM, une aide h la traduction,"
META Journal des traductears, vol. 16, No. 1-2,
March-June 1971, pp. 95-104.
[36] J. P. Vinay, "Utilisation Lflectronique de la Banque de
mots," META Journal des tr3ducteurs, vol. 15, No. 1-2,
March-June 1971, pp. 95-1C4.
[37] R. Dubuc, "TERMIUM: System Description," META Journal
des traducte_urs, vol. 17, No. 4, December 1972, pp.
201-219.
[331 F. Krollmann, "Linguistic data banks and the technical
translator," META Journal 3es traducteurs, vol. 16, No.
1-2, March-June 197i,-Pp-
Approved For Release 2008/03/03: CIA-RDP83MOO171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
[39] F. Krollmann, H. Schuck and U. Winkler, "Herstellung
textbezogener Fachwortlisten mit einem Digitalrechner -
eia Verfahren der automatischen Ubersetzungshilfe,"
Beitrage zur Sprachkunde and Informationsverarbeitung,
No. 5, January 1965, pp. 7-30.
[4D] K. Gingold, "A User's Guide to Inferior Translations,"
The ATA CHRONICLE, vol. IV, No. D. May 1975, pp. 6-7.
[41] D. L. Gold, "On Quality in Interpretation," babel, vol.
XIX, No. 4/1973, pp. 154-155.
[42] D. Fredericksen and L. Power, "A Query System for
Reviewing On-line Manuals," Computing Center Newsletter,
131 T. J. Watson Research Center, Yorktown Heights, New
York, vol. 8, No. 4, March 10, 1975, pp. 37-39.
[ 43 ] IBM Systems Reference Library, IBM Si-stem[360 Time
Sharing System Assembler Programmers Guide, Form No.
GC28-2032, IBM Corporation, Time Sharing System
Programming Publications, Kingston, New York, 1972, pp.
157-159.
Approved For Release 2008/03/03: CIA-RDP83MOO171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
APPENDIX: SORT MECHANISM AND MEMORY LAYOUT OF TERMS
IN TERMINOLOGICAL DIGEST GENERATION
This section describes the essentials of the memory
layout and access scan of the terms used for terminological
digest generation, whether performed in order of textual
occurrence or in alphabetically-sorted order. When the
terms selected for the terminological digest come into the
computer, they are available in the original sequence, and,
after (optional) application of a sort routine, in
alphabetically-sorted sequence in the same memory area as
well. They may then be retrieved is either sequence, looked
up in a dictionary according to the procedure mentioned in
[33] and outputted in soft or hard copy as a digest for the
user.
Figure 10 represents a section of the memory with terms
after they have entered the system. The terms, which can be
of practically unlimited length, are placed adjacent to each
other in main memory as they come in, according to a
variable-length storage scheme. Each term is preceded
initially by a length code and an empty memory cell. If the
terms are to be sorted alphabetically, the sorting process
inserts in each such, empty cell a pointer (the "chain
pointer") indicating the position of the next term in
alphabetical sequence. If a terminological digest is
requested in textual term order, this pointer is irrelevant:
The terms are scanned sequentially in memory 'from left to
right' and the dictionary search is performed on each term.
If a terminological digest is requested in alphabetical
order, the terms are first "sorted" by chaining them one by
one in alphabetical sequence.; as each new term is processed,
the partially formed chain i.s scanned from the beginning in
order to determine where the term is to be inserted. After
the sort procedure, the terms are accessed through their
chain pointers in alphabetical sequence for the dictionary
search. The flow chart (Figure 11) describes this
sort/comparison mechanism for variable-length terms.
The sort process uses, in addition to the pointer
attached to each term, five "global" pointers to keep track
of the terms being worked on. These pointers are referred to
on the flowchart as FRESPNT, SAENTRY, GRVALAD, TEMPS, and
PREVOLD.
Looking at a cycle of the sort process, assume a
start-address pointer (SAENTRY- points to a partial sorted
chained sequence and a high-value pointer (GRVALAD) points
to the latest, i.e., alphabetically highest, term in that
sequence. The term pointed at by the high-value pointer has
already been chained to the current term (pointed at by
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
FRESPNT), which is yet to be examined. At this moment it is
not yet known whether the current term is alphabetically
higher or lower than the highest-valued term of the chain:
if it is higher it will remain where it is in the chain; if
it is lower it will be rechained. The current term is first
compared against the term pointed at by SAENTRY. If the
current term is lower, it is attached to the front of the
chain, so that SAENTRY will point to it from now on.
Otherwise, the whole chain is scanned beginning with the
start-address pointer and each of its terms is compared
against the current term. A next-term pointer (TEMPS) is
used to scan the chain and a previous-term pointer (PREVOLD)
trails the next-term pointer by one item, making it possible
to "insert" the current term into the chain when the next
term is not lower than the current term. If the "not-lower"
comparison is not encountered until the end of the chain is
reached, it is certain to be encountered when eventually the
current term is compared to itself. The end-of-chain
condition is detected by the current term having an empty
chain pointer field. In such a case, the current term
remains chained where it is, but the high-value pointer is
advanced to point to the current term. Whatever course the
comparison has taken, the program then chains the next
unexamined term in memory to the last term of the chain,
which is pointed at by the possibly updated high-value
pointer.* The program is now ready to repeat the sort cycle.
A tight comparison mechanism for arbitrarily long
comparan ds (terms) is attached to the sort function, making
it possible to compare up to 232 characters (although such
long terms would certainly never be encountered).
The advantage of employing the above memory layout and
sort mechanism for terminological digest generation is
fourfold:
1. Terms to be sorted: need not be limited or have fixed
length.
Since the "sorting" is actually done by pointer-updating
only, the data (terms) to be sorted are not moved in
memory.
* The advantage of chaining a record to the sorted sequence
before that record is examined is that the check for the
end-of-chain condition can be coded outside the inner
comparison loop, thus making the loop faster. (In addition,
great efficiency and compactness of the comparison/sort
routine in machine code is maintained by the use of
assembler-language programming.)
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03 :CIA-RDP83M00171 R001800120013-1
3. The terms remain in their original order in the memory
area, enabling the user to optionally fetch the terms in
original or in sorted sequence, by retrieving the terms
sequentially or by threading through the chain pointers,
respectively.
4. If sorting is done in virtual memory, merging phases are
not required, since merging is replaced by the automatic
system-paging mechanism[ 43 ].
LENGTH
LENGTH
1POINTERL1-Z TERM,. LENGTH
TERM2
CODE
CODE
CODE
~l
LENGTH
CODE
Figure 10: Conceptual memory section containing terms for a
terminological digest.
Approved For Release 2008/03/03: CIA-RDP83M00171 8001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
SORT/COMPARISON TECHNIQUE
I . SAENTRY, GRVALAD, FRESPNT, TEMPS AND PREVOLD
BEING WORKED ON.
2. pointer IS A LOCATION INDICATOR AS
(CHAIN POINTER FIELD)
SAENTRY POINTS TO FIRST TERM IN THE
SORTED SEQUENCE TO BE LOOKED UP
FOR DIGEST GENERATION
H term [SAENTRY] : term [FRESPNT] >
SAENTRY-0
GRVALAD "0"
FRESPNT I
pointer [O]-*-- "I"
pointer [TEMPS]
empty ? ,
pointer [FRESPNT]-TEMPS
pointer [FRESPNT]
SOCIATED WITH term
FRESPNT- FRESPNT+"
pointer [GRVALAD]-- FRESPNT
CONCEPTUAL MEMORY LAYOUT E
AT THE START OF SORT D
LOCATION
0 I 2 3 4 n
term
TO BE
To
T,
T2
T3
T4 Tn
SORTED
pointer
ASSOCIATED
EMPTY
EMPTY
EMPTY
EMPTY
EMPTY EMPTY
WITH term
E
N
D
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83MOOl71 R001800120013-1
NAME BEDRICH CHALOUPKA
INSTITUTION XONICS, INC., MCLEAN, VIRGINIA
Use following space for an abstract or summary of your project
Title of Project MACHINE TRANSLATION
The Xonic s Machine Translation is a fast and efficient system for
the translation of Russian into English and for translation."of other languages
which have similar grammatical features. This system is representative
of the philosophy that effective Machine Translation is one which simulates
the activities of the human translator.
The computer programs which make up the system are written in
the PL/1 language. They will operate on an IBM 360 or 370 computer. The
entire system uses less than 100, 000 bytes of computer memory for opera-
tion. Translation can be done in three different modes:
(a) Batch - for translation of large quantities of text.
(b) Sentence-by-sentence - for translation of abstracts and short
articles.
(c) Interactive - for translation utilizing teleprocessing.
The system contains supporting programs for updating dictionaries.
Both translation and updating of dictionaries can be done through teleprocessing.
Approved For Release 2008/03/03: CIA-RDP83MOOl71 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
NAME
John Chand.oux
INSTITUTION TAUM Project, University of Montreal
Use following space for an abstract or summary of your project
Title of Project
Meteo Weather Forecast Translation
Meteo is an automatic system for the translation of weather forecasts
from English into French. Public forecasts for the whole of Canada
are directly sent to the system via communications network. The
sentences accepted by the system do not need to be edited or revised.
The remaining sentences are extracted by an interactive editor and
displayed on a screen terminal for translation by a human translator.
Meteo has been operating on an experimental basis 24 hours a day since
December 1975 in parallel with the translation bureau. It will be fully
operational by May of 1976 and presently produces 30, 000 words per
day The actual translation time spent is over 1, 000 words per minute
and estimated costs all inclusive are two cents per word.
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00l71 R001800120013-1
NAME CHANDIOUX John
INSTITUTION TAUM, Universite de Montreal
Use following space for an abstract or summary of your project
Title of Project Leibnitz, Multilingual system
Leibnitz is an international cooperation between
computer translation centers interested in a multilingual
system. Several european groups, the TAUM project from the
Universite de Montreal and a Brazilian group are presently wor-
king on this project. Most parts of the system'are being
written in one of the three languages made available by the
CETA in Grenoble. The first one is the ATEF language, a string
tree transducer for dictionary look-up and morphological analy-
sis. The second one is CETA and is a tree manipulating language
for both transfer and generation. The last one is a tree/string
transducer to be completed sometime in summer of 76.
Each group is either working on the design of an
analyzer or generator for a specific language or on the trans-
portability of the available formalisms. Research is presently
under way on French, German, English, Italian, Portugese and
Russian. English analysis is done by the TAUM team which is
presently experimenting with a parser written in REZO its own
version of Wood's Augmented Transition Networks. All particpa-
ting groups have agreed on a normalized tree representation
for the output of analyzers and input of generators in order
to minimize problems in the design of transfer components. The
first part of the system is expected to be operational within
two years.
Approved For Release 2008/03/03: CIA-RDP83M00l71 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
NAME
Major Lynn M. Hansen
INSTITUTION Foreign Technology Division (FTD)
Use following space for an abstract or summary of your project
Title of Project FTD Machine Translation
FTD has been utilizing machine translation since September
of 1963 when the IBM MARK II system was installed at Wright-
Patterson Air Force Base. The current FTD machine translation
system became operational in July 1970. Since that time nearly
constant improvement has been made through a series of external
optimization contracts and in-house update efforts.
Three major machine translation products are produced at
FTD: the unedited machine translation (MTO) with original
graphics merged onto the computer printout; the preliminarily
or partially edited translation (PET) which includes only those
editorial changes necessary to insure the technical accuracy of
the translated text; and the finished translation (MTF) which
has been completely edited with proper syntactical changes and
then retyped in camera-ready format.
The bulk of the machine translations at FTD deals with S&T
subject matter; therefore, FTD glossaries and lexicographic
routines are basically scientifically oriented. However, the
system as it now exists will provide quite adequate indicative
translations of almost any material.
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00l71 R001800120013-1
NAME
Fred C. Hutton
INSTITUTION Union Carbide Corp., Nuclear Div., Oak Ridge, TN
Use following space for an abstract or summary of your project
Title of Project Georgetown University MT System Usage at Oak Ridge, Tennessee
Ten years' experience in running the programs on the IBM 7090 is
described. The present system, reprogrammed for the IBM 360, is
described and capabilities of the system are set forth., An example of
the use of the language invented by A. F. R. Brown (SLC for "Simulated
Linguistic Computer") used in the preparation of the dictionary and
linguistic routines, will be presented.
Approved For Release 2008/03/03: CIA-RDP83M00l71 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
NAME_ Erhard O. Lippmann
INSTITUTION IBM T. J. Watson Research Center, Yorktown Heights, New York
Use following space for an abstract or summary of your project
Title of Project
Experimental On-Line Computer Aids for the Human Translator
An exploratory computer-aided translation system is being
developed which basically consists of storage and retrieval operations
carried out on line with a computer during the time in which a translation
is produced. The system is not programmed to simulate the human
translator by producing automatic translations. Rather, the user can
call upon the computer's resources as needed in the translation process
to shorten the delay between the initiation of a translation and its finished
version. A combination of terminals, computer devices, and software
is used to perform functions which have habitual human counterparts of
a mechanical nature, e. g., dictionary look-up, dictionary updating,
creation of terminological digests (i. e. , test related mini-dictionaries),
semi-automatic editing generation of cross reference files, text
statistics, printing and lay out, and automatic combination insertion,
deletion, or duplication of text.
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
NAME Automated Language Processing Project; Dr. Eldon G. Lytle, Director
INSTITUTION
Brigham Young University, Provo, Utah
Use following space for an abstract or summary of your project
Title of Project Automated Language Processing Project
The Project emphasizes the refinement of computer-assisted translation,
as opposed to fully automatic translation, and has devised for this purpose
techniques of man-machine interaction which utilize the human for those
aspects of the translation task requiring human intelligence and the computer
for those aspects of the translation task which can be managed mechanically.
Junction Grammar, a new theory of language structure which'captures ling-
uistic universals hitherto unknown, serves as the basis for the system.
Phase I of the development (now operational),provides computer editing,
file management, and dictionary lookup. Phase II of the development provides
computerized analysis, transfer, and synthesis of sentence structure (imple-
mentation 1978-79). Proto-type systems are designed for translation from
English to Spanish, French, German, and Portuguese, but the method is
equally adaptable to any combination of source and target languages.
The primary sponsor of BYU ALP is the Church of Jesus Christ of Latter-
day Saints (Mormon), which annually translates approximately 17,000 pages of
material into more than fifty (50) languages. It is planned that dictionary
lookup and linguistic processing will initially be accomplished at a large
central installation. The output of this processing will then be forwarded
on "floppy" disks to regional translation centers around the world where
residual aspects of the translation and printing task will be accomplished
with the aid of mini-computer work stations.
The Project has a staff of 12 full-time and 18 part-time researchers.
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
NAME Roger C. Schank
INSTITUTION vale University
Use following space for an abstract or summary of your project
Title of Project Computer Understanding of Text
Research at Yale centers around the building of computer programs that wi
understand stories. Two program are currently being developed, SAM and PAM.
SAM is composed of the following
1) an analyzer that maps English into a deep conceptual representation.
2) a script applier that uses its knowledge of contexts to supply missing or
or implicit inferences about a situation.
3) a memory that finds references for things that it knows about in a text so
as to bring its knowledge to bear on the text.
4) a generator that reads information provided to it by (1), (2), and (3) and
states that information in English, Chinese, Russian, Dutch or Spanish.
5) a question answerer that interacts with the script applier to answer questI
about an input text.
SAM is capable of mechanical translation, automatic summary and paraphrase
and question-answering about texts in domains that it has knowledge about.
PAM is like SAM except that it does not have a script applier but instead
has a more general mechanism that to infer the goals and intentions
of the actors in the stories it hears.
Both of these programs are beginning approaches to the problem of
computer understanding.
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00l71 R001800120013-1
NAME Robert .1. Shillman, Ph.D.
INSTITUTION M.I.T.
Use following space for an abstract or summary of your project
Title of Project optical Character Recognition Based on Phenomenological Attributes
A theory of character recognition has been proposed and
a methodology has been developed which is expected to yield
a machine algorithm that will equal human performance in
the recognition of isolated, unconstrained, handprinted characters.
The methodology is based on the study of ambiguous characters,
characters that can be assigned two letter labels with equal
probability, rather than on letter ardhetypes. A description
of the underlying representation of each of the 26 upper case
letters of the English alphabet was obtained through analysis
of ambiguous characters which were generated for this purpose.
The descriptions are in terms of an abstract set of invariants,
called functional attributes, and their modifiers. The
relationship between the physical attributes, derived from physical
measurements upon a character, and the functional attributes
is given by a set of rules called Physical to Functional Rules.
Three different techniques for determining these rules through
psychophysical experimentation have been tested, and the particular
rule for the attribute LEG has been determined. The remaining
rules can be obtained in a similar fashion, and the combined
results are expected to provide the basis for a machine algorithm.
We are currently investigating the Physical to Functional Rules
for the remaining attributes and are also interested in the
way in which the rules are to be combined.
Approved For Release 2008/03/03: CIA-RDP83M00l71 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
NAME Robert F. Simmons
INSTITUTION University of Texas, Austin, Texas
Use following space for an abstract or summary of your project
Title of Project TEXT INFORMATION SYSTEMS
A developmental program is proposed to create a socially useful
system that will integrate several existing natural language processing
procedures into a robust, transportable, General Text Understanding
System for eventual use in applied information centers. The proposal
is comprised of seven tasks: 1. Continued development of quantified
case predicate forms of conceptual memory structure. 2. . Integration
of question.. answering and problem solving procedures. 3. Development
of a human-aided, multi-pass, text-to-memory compiler. 4. Generation
of natural language outputs for summaries, abstracts, expansions,
translations, etc. 5. Generation of special purpose text teaching
materials. 6. Implementation of natural language dialogue capabilities.
7. Development of a textword management system for linguistic
analysis, retrieval and lexicon development.
The work will be accomplished on a DEC10 to enhance the trans-
portability and communication of documentation for the resulting system.
E
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
NAME Dr. Peter Toma, President and Chairman of the Board
INSTITUTION LATSEC, Inc. and World Translation Center, Inc.
Use following space for an abstract or summary of your project
Title of Project SYSTRAN
After havinc developed the SERNA, AUTOTRAN, and TECHNOTRAN
machine translations, I felt that the advent of third generation
computers provided the lcng-awaited opportunity to develop a
large-scale, yet fast and economical, systematically planned,
unified, universal system. That system is SYSTRAN, whose name is
an acronym formed in 1964 from "systems translation."
SYSTRAN is a fully operational machine translation system which
can be installed at any IBM 360/370 site within hours. It has bee
used by the Air Force (translating 15 million words a year) since
1970 and by NASA since 1973. SYSTRAN is fully automatic, requires
no human intervention nor pre-editing. It translates between the
following language pairs at a speed of 300,000 words per hour:
Russian-to-English, English-to-Russian, English-to-French, German-
to-English, and Chinese-to-English. We term it a universal
translation system because of this and because of the ease with
which new translation capabilities can be added.
SYSTRAN's success is due to its strong and very flexible soft-
ware frame, which allows the immediate implementation and testing
of linguistic hypotheses, as well as universality in handling
natural languages. Moreover, its special macro language allows
linguists to program their own rules. The system can be modified
or expanded to any limit at any time. It can never become a
"black box."
The complete SYSTRAN package includes all utility programs,
dictionary creation and update subsystems, source language analysi
programs and target language generation programs, as well as pro-
grams for development of frequency listings, concordance materials
etc. There are separate dictionaries for stem entries and idio-
matic expressions (which are also entered in stem form). Because
lexical items are entered in stem form, and because of a complex
cross-referencing system, it is necessary to enter any lexeme only
once, accompanying it with paradigmatic set information.
Source language analysis programs begin at homograph resolution
proceeding through establishment of immediate constituents (IC's),
to establishment of syntactic relationships of IC's and establish-
ment of clause types and clause boundaries. Semantic analysis is
used not only in selecting proper target language meaning equiva-
lents, but also in establishing certain syntactic relationships.
Target language generation includes structural transformations,
synthesis of target language word forms including insertion of
auxiliaries, prepositions, etc., and rearrangement within clauses
to achieve standard target language word order.
funs ~t,t}er_deyelopment,of#Eronoun translation and prepositional
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
NAME
William S- Y Wang
INSTITUTION
University of California, Berkeley
Use following space for an abstract or summary of your project
Title of Project
Project on Linguistic Analysis
Research on machine translation from Chinese to English under the
direction of William. S-Y Wang was carried on at the project on
Linguistic Analysis (University of California, Berkeley) during the
period 1967 to 1975.. During the early part of the effort, System I was
developed which includes: a) CHIDIC: A Chinese to English machine
dictionary of about 80, 000 entries (60 percent physics, 30 percent
biochemistry, and 10 percent general), and b) Monolithic grammar of
about 4, 000 rules (context-3, phrase-structure rules). In 1973, two
factors caused redesign of the approach toward-the development of
System II. One,. the grammar had become so cumbersome and ad hoc
that its effectiveness as well as its potential for improvement were
curtailed. Second, the sponsor requested conversion of the system from
CDC machines to IBM machines. In response to these factors, System
U is designed along the lines of "structured programming" (i. e. , it is
built on self-contained program modules). It is also designed to be
machine-independent, so that it can be implemented at different computer
installations.
Efforts in research and development have been aimed at an operational
system. We have experimented with numerous trial sentences as well
as several "live" texts (from articles of 3, 000 characters in length) and
have accumulated machine texts of over 560, 000 characters. System U
is incomplete, lacking especially the machine-editing of output to
conform to those morphological features absent in Chinese but required
in English.
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
NAME Yorick Wilks
INSTITUTION Dept. of Artificial Intelligence, Univ. of Edinburgh, UK.
Use following space for an abstract or summary of your project
Title of Project An Al Approach to MT
The present system takes in paragraphs of English on line and out-
puts paragraphs of French. It is very small with a vocabulary of about
5-600 word senses, but that is very large for a project of this sort. By
"this sort" I mean projects that aim for some deep semantic representation
of the input language and from which the translation is produced. There is
no separable syntactic stage in this work; the text is fragmented (into
clause and phrase length pieces) by the program, and semantic structures
are attached directly to these. These semantic structures are called
templates and correspond to "mini-assertions. " That is to say, the pro-
gram seeks to display the input as a sequence of mini assertions. These
templates are constructed out of formulas, already available from a diction-
ary, for the word sense of the input. Each word sense has a formula for it,
and much of the work in the program is ascertaining what is the correct
word sense (and so correct formula) for an input word. The formulas are
tree structures built up out of different types of semantic primitive. A
formula has internal rules operating on these semantic primitives that
enable it to express the meaning of the corresponding word sense. Once the
templates have been formed up, various kinds of inference rules operate on
them , to produce deeper semantic representations, so as to resolve
remaining ambiguities of word sense, prepositions or pronoun reference.
When a clear single temple representation has been obtained, a French
representation can be generated from it.
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
NAME MICHAEL ZARECHNAK
INSTITUTION Georgetown Univeristy School of Languages and Linguistics
Use following space for an abstract or summary of your project
Title of Project Georgetown University General Analysis Techniques-(GAT)-
Simulated L
The Georgetown University Russian-English System is running on IBM
360/70 .CPU time for 2000 words C 9 seconds. The texts translated
include scientific,tecinological,and economic materials.
M.Zarechnak in close cooperation with the linguistic research staff.
The linguistic statements are coded in symbolic language designed by
Dr. A.Brown ('SLC'-Programming Language). Input/output is in Assembler
language.
A dictionary entry contains a split or unsplit Russian stem, grammati-
cal coding, lexical number, and English part. The clustered entries
are recognized through special local operations when the calling signal
occur within the sentence under processing.
Syntactic analysis is partly based on morphosyntactic markings and
partly on semantic coding.
Users: Primarily scientists at ORNL. Users' comments essentially favo-
rable*
The unddited translation is used primarily for information purposes,
although in a few instances, the translations were-post-edited when
the user requested it.
The quality of the present translation. is the same as it-was in 1964.
No linguistic improvements were inserted in the system although there
area some linguistic programs ready to be inserted.
The semantic level will be added. Its underlying procedured_re based
on the semantic collocational and collicational distributional patterns
as observed in the real corpora, with such generalization as these cor-
pora would suggest:. It is hoped that after large corpora will be des-
cribed both semantically and analytically, then some theories might be
developed and tested deductively for the improvement of the next MT
cycle. Each sentence is scanned from the left to the right, and from
right to left at least forty times, following a path of certain prio-
rity-based strategies.. All these scannings in both directions are
grouped into four levels: word recognition, syntagmatic,syntactic, and
synthesis of English. Some parts or. the synthesis are independent of
the Russian input.
Size of the dictionary: (950,000 stems.
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 NAME
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
C. Loh
INSTITUTION
Chinese University of Hong Kong
Use following space for an abstract or summary of your project
Title of Project
Chinese University Language Translator (CULT)
The Chinese University Language Translator (CULT) is a
Chinese-English computer-translating system which is unique in that
it utilizes pry:-editing of the source text as opposed to post-editing of
the target text. The system is essentially a "pragmatic" one, in that
the rules for handling complex strings previously requiring pre-editing
are introduced as needed. CULT is made up of four modules: Dictionary
look-ups, Syntactic Analyzer, Semantic Analyzer, andOutput. Among
these, the most limited is the Semantic Analyzer, which seems to rely
more heavily on pre-editing than the other modules.
CULT is currently being used to translate two Mainland scientific
journals, ACTA Mathematics Sinica and ACTA Physica Sinica. The
computer, however, is not a person, which means it cannot experience
the feeling of the original text and is not of much use for translating
f ss~ .S'
literary n al, Nevertheless, scientific and some non-scientific
works are also within the scope of its capabilities.
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
NAME Jim Mathias
INSTITUTION CE TA (U.S. A.)
Use following space for an abstract or summary of your project
Title of Project Chinese-English On-Line Retrieval
The CETA (Chinese-English Translation Assistance) group is building
a machine readable dictionary file for use in on-line retrieval and for
development of dictionaries and indexes for use of human translators.
The experimental on-line retrieval system can store an unlimited number
of entries. The current file of 640, 000 machine readable entries is
divided into approximately 110, 000 general entries , 10, 000 colloquial
entries, and 500, 000 scientific and technical Chinese-English entries.
The experimental system designed for an IBM 360 illustrates the facility
of computer storage, retrieval, and display of Chinese characters and
Roman alphabet as well as other scripts. It also illustrates the facility
of computer techniques for indexing Chinese characters and special
adaptability for synthesizing Chinese queries to search telecode sorted files.
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
 Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1
NAME Friedrich Krollmann
INSTITUTION Federal Bureau of Languages
Use following space for an abstract or summary of your project
Title of Project FRG Translation Aid System
Germany's Federal Bureau Computer Translation Aids System
contains over 700, 000 foreign language (English, French, Russian,
and Portuguese)- German entries of a technical and scientific nature.
These entries can be accessed in a number of different ways depending
on the needs of the user. Thus, the programming of the system allows
for more specialized foreign language-German glossaries and lexical
concordances, as well as linguistic analysis and frequency counts on
the technical vocabulary of a given language.
Approved For Release 2008/03/03: CIA-RDP83M00171 R001800120013-1