College and Research Libraries


Symbols and Storage 
Symbolic Languages in Data Processing. Pro-

ceedings of the Symposium sponsored and 
edited by the International Computation 
Centre, Rome, March 26-31, 1962. New 
York: Gordon and Breach, 1963. xii + 
849p. $34.50. 

Librarians are becoming increasingly 
aware of the potential of the digital com-
puter in problems of information storage 
and retrieval. One of the reasons that work 
in this important field is not moving more 
rapid.ly is the difficulty presented by pro-
grammg. Few working librarians have the 
time to learn how to program a computer, 

.particularly sine~ program problems vary 
from machine to machine. It is for this rea-
son that computer experts have developed 
symbolic languages, defined as "completely 
formalized language (s) structurally and con-
ceptually rich enough for communication be-
tween computers, between computer and 
man, and even between man and man in 
the computer field." Such symbolic lan-
guages, the best known of which are prob-
ably the FORTRAN, COBOL, and ALGOL 
systems, permit communication with the ma-
chine without tedious study of programing 
techniques. There are two drawbacks to the 
~se of such languages, however: (1) there 
IS no generally accepted single symbolic lan-
guage, so that the investigator must at pres-
ent learn a number of languages if he is un-
certain of the computer he may have to use; 
(2) since symbolic languages are used in so 

many ways by so many fields, they tend to 
grow rather rapidly and to become unman-
ageable. The present book represents an at-
tempt to solve many of the problems inher-
ent in computer languages by an interna-
tional symposium. Though it does not seem 
to have reached definitive conclusions, this 
symposium is to be heralded as a step in the 
right direction. 

The book consists of lectures given by the 
participants in the symposium, grouped 
around six panel discussions: Metasyntactic 
and Metasemantic Languages, Languages 
for Aiding Compiler Writing, Philosophies 
for Efficient Processor Construction, Reflec-

JULY 1963 

Review Articles 
tions from Processor Implementers on the 
Design of Languages, Are Extensions to 
ALGOL 60 Necessary?, and Is a Unividation 
of ALGOL-COBOL, ALGOL-FORTRAN 
Possible? The Question of One or Several 

. Languages. The panel discussions are given 
in the now familiar but tedious form of 
transcription from tapes, and one is occa-
sionally unsure of what went on. 

The last question, namely the possibility 
of uniting the FOR (mula) TRAN (slation) 
system developed for the IBM 700 series, the 
CO(mmon) B(usiness) O(riented) L(anguage) 
sponsored by the U.S. Department of De-
fense, and the European-American ALGO 
(rithmic) L (anguage) , which is intended as 
an international standard, is in my opinion 
the most important of ~II for the user of the 
computer who is not himself a computer 
man. It is unfortunate that the conclusions 
of the panelists seems to have been that no 
unification is possible. In fact, it seems from 
the tenor of the discussion that we must look 
for a continued proliferation of computer 
languages. 

It is for this reason that the first section 
of the book, "Theory of Languages-Syntac-
tical Structure and Metalanguages," is so im-
portant. If we are to have a multitude of 
computer languages, the normal noncom-
puter-trained user of the machine needs a 
further language which will permit him 
access to the computer language. Of particu-
lar importance is an article by J. Riguet on 
programing and the theory of categories; in 
this article, M. Riguet outlines the theory of 
categories, which he thinks will replace set 
theory as the basic language of mathematics, 
and shows how programs may be cast in its 
terms. The theory is so simple and easy to 
understand that it seems at least as powerful 
as set theory for purposes of programing. If 
computer languages could be cast in the form 
of some other more powerful and generally 
used language such as the theory of sets or 
categories, the goal of international coopera-
tion in this field could be realized, and the 
artificial restrictions at present placed on the 
growth and proliferation of computer lan-
guages could be lifted. 

Also of great interest is the section on the 
use and design of languages to aid compiler 

339 



wntmg. A compiler is a program which 
equips the particular computer to accept 
programs in a problem-oriented language 
and to transform these into machine lan-
guage instructions. Again, the panel and 
the lecturer represent no unified opinion on 
this matter, and no synthesis is presented. 
With the widespread . international use of 
the computer for language processing, such 
programs should be made available widely. 
The book under review here represents an 
advance in that it is based on international 
cooperation; it is disappointing in that it 
yields so little evidence of a realization of 
this fact on the part of the participants. The 
nonprofessional user of the computer will 
find little comfort in the professional's atti-
tude towards him as revealed in the pages 
of the symposium report.-]ames W. March-
and) Vanderbilt University. 

General Bibliography on Information Stor-
age and Retrieval. By Marshal Spangler. 
[Phoenix, Ariz.]: General Electric Com-
puter Department, 1962. 390p. Apply. 

Literature on Information Retrieval and 
Machine Translation. By Charles F. Balz 
and Richard H. Stanwood. White Plains, 
N.Y.: IBM Processing Division, 1962. 117p. 
Apply. 

Permuted indexes are· falling upon us in 
profusion , but the two under review here 
are the first extensive ones to appear in the 
field of librarianship. Although IBM calls 
its process "Key Word in Context" (which 
shortens into the happy designation KWIC) 
whereas General Electric says "significant 
word in context," the two efforts use essen-
tially the same devices for controlling the 
same body of literature. Large libraries will 
want both of these bibliographies even 
though they are very similar in subject and 
scope. 

The General Electric bibliography lists 
alphabetically by main entry 1550 books 
and journal articles in the broad area of 
information storage and retrieval, including 
such tangential but related subjects as me-
chanical translation, character and pattern 
recognition, speech analysis and synthesis, 
self-organizing systems, and artificial intelli-
gence. Almost all references are to writings 
since 1949, although a few significant earlier 
works are included. 

The General Electric bibliography has an 
index to secondary authors and an index to 
source journals. (This latter index, for ex-
ample, shows that nine items are listed that 
had been published in College and Research 
Libraries. As would be expected, American 
Documentation furnished more items than 
any other journal with 329) . This is fol-
lowed by a chronological index, and the 
book closes with the permuted word index. 
Approximately 7,250 entries in this index 
indicate a ratio of not quite five per item. 

The IBM bibliography contains about 
twice as many main entries, again with each 
entry indexed under an average of five terms 
permuted from its title. There is an author 
index but no source nor chronological in-

. dexes. An interesting sophistication is a list 
of words not used in developing the per-
muted index. Among this homogeneous but 
useless lot of discarded terms are Don, Force, 
Poughkeepsie, September, and Versus. Al-
though more comprehensive than the Gen-
eral Electric list, the IBM compilation is 
printed at a reduction in size giving it a 
somewhat handier format. 

These are good practical examples of what 
the permuted index can do. It will be inter-
esting and indicative to observe how much 
we use them.-D.K. 

Statistics 
Library Statistics · of Colleges and Universi-

ties) 1959-60) Part 2: Analytic Report. 
Washington: U.S. Office of Education, Li-
brary Services Branch, Bureau of Educa-
tional Research and Development, 1962. 
45p. 40c. 

One of the problems which was never ade-
quately solved by the hardworking, deadline-
pushed ACRL Statistics Committee (now 
LAD LOMS Statistics Committee for College 
and University Libraries) was that of pro-
viding for academic librarians a useful or 
significant analysis of the raw data supplied 
in the tables formerly published in this · 
journal each January. When the collection 
and publication of statistics was turned over 
to the USOE Library Services Branch in 
1959, it was with the understanding that the 
data processing machinery of that agency 
would provide a more detailed statistical 
analysis as Part 2 of its yearly publication of 

340 COLLEGE AND RESEARCH LIBRARIES