Reviews

291

D. G. Burnett-Hall and P. Stupples, discusses learning Russian on the secondary level by applying computer-compiled vocabularies; the aim of the coursewas to teach the students to develop speech patterns, as well as to enable them to read Russian. Additionally, the authors' solution to the problem of computer transliteration between the Roman and the Cyrillic alphabet is explained in a five-page, three-part appendix. While the details about the planning and input of the Russian program can be suggestive, Burnett-Hall and Stupples might have told us a good deal more about the output and application. The last category of essays, designed for an expert in a given field, is perhaps best exemplified by the two highly specialized papers in the fifth section, "Computer Applications to Oriental Studies." The theoretical and practical realms opened by these papers may very likely perplex as much as enlighten the general literary scholar. Here, R.D. Bathurst explains the Cambridge project in automatic alphabetization of Arabic words, and Alan Jones describes three Oxford projects in Oriental languages, ranging in scale from an analysis of modern Turkish through a concordance of the Persian poet Hafiz to a multifaceted study of the Koran, which Jones hopes will ultimately lead to a variorum edition of the text. The technique being evolved in the course of this research suggests possibilities to the reader for computer investigation in other areas as well. A different approach is taken by R. H. Churchhouse and Susan Hockey in their work on producing a readable output in the Greek alphabet. Their efforts are also designed to lead towards the output of Greek diacritical markings and, ultimately, Arabic script. Of specific interest to librarians and archivists is Eveline Wilson's essay on "The Use of Formally Defined Structure for the Input of Data from the British Museum Catalogue of Printed Books." The difficulties encountered in this project are simultaneously generic (how to get a text into and out of the computer) and unique (the text is a structured but arbitrary coding of factual data). Wilson's project was an attempt to solve the problems rather than to provide a store of data or an analysis. The sophistication and wide range of the topics taken up by the people who gave papers at the Cambridge symposium surely indicate that the computer has come of age, and that scholars are taking advantage of its ability to process a great deal of data. All this is confirmed by Wisbey's collection. Thus, as an introduction to computer literary studies on a somewhat advanced level, it is a valuable though imperfect work, summing up progress and prospects in various humanistic disciplines up to the time of the conference (March 1970). Since it is the first publication of a series, its greatest significance perhaps lies in indicating the directions to be explored in subsequent volumes. In any case, Wisbey's book cannot be ignored, for it is a major contribution to a computer bibliography.

Metric Analysis of Horace WILHELM OTT, Metrische Analysen zur Ars Poetica des Horaz. Goppingen: Verlag Alfred Kummerle, 1970.19 DM. By James J. Helm THE BULK OF THIS SLIM VOLUME is devoted to the presentation of an assortment of lists and statistical tables describing the metrical characteristics of a classical Latin poem, the Ars Poetica of Horace. While this poem is a primary source for our knowledge of James J. Helm is in the department o f Classics at Oberlin College, Oberlin, Ohio.

292

Computers and the Humanities/VoL 7, No. 5/May 19 73

ancient literary criticism, Ott is interested in its metrical form, and has selected those aspects for study which can most easily be analyzed automatically by the use of a computer. Yet his intent is not simply to present this material in as full and complete a form as possible (although this he does), but also to stimulate interest in the use of the computer and in the production of similar automatic analyses of other Latin hexameter poems, so that the unique metrical characteristics thereby discovered may provide material for a comparative study of the metrical style of hexameter poets. Ott's procedure includes the following steps: (1) prepare the text in machinereadable form; (2) determine the metrical scansion of each line automatically on the computer (95 percent correct); (3) proofread and provide the correct scansion for the remaining 5 percent; (4) establish the primary metrical characteristics of each verse: positions of word-juncture, incidence of elision or hiatus, relationship between word-accent and ictus (metrical beat), metrical shape of each word (in terms of number and metrical length of syllables, together with an indication of elision when appropriate), and metrical shape of each complete line; (5) compile desired tables and lists; (6) prepare the text for printing. Significantly, all of this work was carried out automatically by the computer except for steps (1) and (3). No real definition of metrical style is given by Ott, but some notion of what he has in mind may be inferred from a summary of the information provided. Besides statistical summaries of the characteristics mentioned in procedure (4) above, the following indices were included: metrical indices of elided words, an index of metrical line types (grouped according to the sequence of dactyls and spondees), a metrical index of the line endings (with "line ending" defined as that part of the verse that follows the last word ending before the fifth foot), and metrical and alphabetic indices of the individual words. In short, the emphasis is on the metrical shapes of single words and of entire lines, on word juncture and elision, and on the coincidence of ictus and accent. These are for the most part standard concerns of traditional metrical study. What is new here is not so much the data, nor some revolutionary metrical theory underlying it, but rather Ott's reliance on the computer with the advantages it provides in the areas of the completeness and reliability of the data, together with the speed and the effortlessness of its compilation. Automatic computer scansion of Latin poetry has been carried on by others before, but this is the first time, to my knowledge, that the data has been published in so full a form. There are, expectedly, the drawbacks of which the author himself is aware: while the computer may make it possible to ask new and more complex questions, there are still some major questions concerning a number of important metrical characteristics that cannot (ye0 be answered by mechanical means. Where subjective judgment enters (such as in the location of caesura in certain lines), no unambiguous rules can be given to the computer as a basis for making decisions that will gain general assent. Indeed, scholars may differ widely among themselves on such points. Nevertheless, Ott has programs operational by now which will automatically perform most of the scansion and compile all of the statistical tables and indices. All that is needed to extend the process to a wider selection of authors is to have their texts in machine-readable form. In fact, there are a number of such Latin dactylic hexameter texts available from the American Philological Association Repository of Machine Readable Texts, many of which have already been scanned. The question of whether or not to apply Ott's programs to this material need hardly be raised. There is no real reason not to do it. Ott himself has already been analyzing Ver~'sAeneid in the same manner. As he indicates in this volume on Horace, the study will become much more meaningful when the method is applied to a wider selection of authors. The real question, however, is what to do with the resulting data. In my opinion, the greatest value of the present volume is that it provides a concrete example of what

Rev~ws

293

can be done on the computer, and how well it can be done. But as a publication of data, in spite of some very ingenious attempts to make it easy to use, it is far less useful in this form than it would be if still accessible to a machine. This is to say, while it is far easier to use Ott's tables than to work from a bare unscanned printed text, it would be still easier to extend his study of Horace ff the data were available in machine-readable form. Given this fact, it is doubtful that the expense of publishing similar volumes will be warranted. What now needs to be produced is not a series of volumes of this type, but rather a series of publications analyzing the results: a comparative study of the metrical style of the Latin hexameter poets which draws meaningful inferences from the data so compiled. Then (or in the meantime) the data base could be made available on punched cards or magnetic tape so that the tables can be verified or extended at will. 1 hope that this volume will stimulate the response anticipated by Ott so that such studies may proceed through the collaborative efforts of a number of scholars. But final publication ought to be reserved primarily for the analysis of the material, not just for the raw data itself.

Chinese Linguistic Bibliography WILLIAM S-Y. WANG AND ANATOLE LYOVIN, C L I B O C : Chinese Linguistics Bibliography on Computer. Cambridge University Press, 1970. $23.50. By Mary L. Streeter C L m o c : Chinese Linguistics Bibliography on Computer lists some 3,000 books and articles relating to almost every aspect of Chinese linguistics. These topics include "diachronic as well as synchronic studies of Chinese dialects, Sino-Tibetan languages within and without the borders of China, and the non-Sino-Tibetan languages within China, as well as works of Chinese linguists on general linguistic t h e o r y . . , items dealing with orthography, semantics, etymology. . . . " The works listed in CLXBOC are equally divided between Western and Eastern (Chinese/Japanese) sources, all published before 1968. c r 4 B o c is divided into four primary sections. The first is a list of all the works included, arranged alphabetically by author. Next comes an index of more than 500 topics, keyed to the items in the main entry list. Third are abstracts of selected works; these are not numbered, but I estimate that slightly more than 600 (or 1/5) of the total number of entries have been abstracted. Fourth is an excellent set of colored 15"x22" maps in the jacket pocket. Showing the political boundaries of East Asia, divisions of Sino-Tibetan languages, and the boundaries of each of the Chinese dialect groups, they are in themselves extremely useful research tools. The items included in CLIBOC do not comprise a complete list of every work published on Chinese linguistics. This would entail a tremendous set of volumes, since the field has been in existence for over 1,800 years. Rather, CLIBOC contains the best of

Mary L. •treeter is on leave from the linguistics department at the University o f California at Berkeley.