There have been, in the lexicographical European tradition, two major types of dictionaries – the one with notional arrangement is older (and was for a long time more important) than the alphabetical one (Hüllen Reference Hüllen1999). In Britain, the collections by Wilkins (Reference Wilkins1668) and Roget (Reference Roget1832) have established some continuity in modern times (Hüllen Reference Hüllen2004). However, none of the European languages has ever had a thesaurus covering the lexis of its entire linguistic history. Such a project was conceived at Glasgow University as early as the 1960s, the most comprehensive historical dictionary, the OED, serving as a database, and Roget's scheme initially planned to serve as a pattern for the Historical thesaurus of the Oxford English dictionary (HTE)'s notional structure. Since the OED includes only words recorded after 1066, the rich wordhoard of Old English was treated separately and published in due course as the Thesaurus of Old English (Roberts, Kay & Grundy Reference Roberts, Kay and Grundy1995).
The original investigators of the HTE did not foresee all the immense problems inherent in the project nor the forty-five years it would take for the paper version under review to appear. It is a consolation that the prime mover, Professor Michael Samuels, lived to see its completion.
Nearly 4,000 pages with three columns in small print per page are a visual indication of the enormous work that has gone into the enterprise. More difficulties become evident from the introductions to the two volumes and many scholarly papers published on the subject (e.g. those edited by Kay & Smith Reference Kay and Smith2004). Problems, some unforeseen, include necessary adaptations caused by moving away from the Roget tradition of classification (which proved too idiosyncratic even for a synchronic categorization of Present-day English) and the technical changeover from boxed paper slips to electronic storage in the 1980s. The result is now a unique and admirable compilation of more than 800,000 entries in 251 categories descending from 69 major classes in ‘The world’, 83 in ‘The mind’ and 99 in ‘Society’, which are broken up on as many as seven classificatory levels. A second volume, somewhat fatter than the actual thesaurus, comprises the word index, i.e. it lists all the lexical items contained in the OED.
All the HTE data, then, come from the OED and its various complements. In the individual categories, lexical items are chronologically arranged without meanings under (groups of) specific headwords. A very welcome addition in the HTE is the supply of Old English words (taken from Roberts & Kay Reference Roberts, Kay and Grundy1995 and from dictionaries) in alphabetical arrangement at the beginning of entries, though without approximate dates provided. The OE items show the enormous number of losses in many lexical fields resupplied (often from French) after 1066. No two thesauruses published for any of the modern European languages agree in their notional structure (cf. Fischer Reference Fischer2004) – partly in consequence of the differences in the individual Weltbild reflected, and partly because of the compilers’ decisions. With the historical dimension of HTE added, differences to other thesauruses are bound to increase: categorization must take into account not only semantic change (largely covered by the OED), but also notional changes. (For a very simple illustration of the interplay of cultural and linguistic history see my analysis of town through the ages, in Görlach Reference Görlach1997: 135.)
Obviously, HTE users cannot retrace the unending discussions which resulted in the classifications under specific headwords ultimately printed – and even where we may doubt the wisdom of individual decisions we should refrain from nitpicking (cf. Kay Reference Kay2004). The editors’ decision to use a modified folk taxonomy for their classification results in a ‘timeless’ scheme; this obviously replaces medieval categories such as ‘water animals’ or ‘flying creatures’ with those of the modern (or at least Linnean) biology of fishes and birds/insects – a decision which could now stand in the way of an analysis of conceptual as against semantic change. Less conspicuously, an analysis of, say, the value system of Chaucer's knight, who believed in trouthe and honour, fredom and curteisye, becomes difficult (cf. Héraucourt Reference Héraucourt1939).
Individual HTE categories start with general items, before more specific and restricted ones are enumerated, or lists of ‘particular types of . . .’ are added at the end. Since notions cut across parts of speech, a fixed order of nouns, adjectives, adverbs, verbs, prepositions, phrases and interjections is carried through, even where the semantic core may be felt to be verbal. Datings are taken over from the OED, and so are stylistic and regional markers – the unevenness of the OED in this respect warns us that these data are incomplete or may be misleading.
There are logical reasons for the top-down categorization using seven digits, but it is sadly apparent that this makes the system extremely unwieldy. The 251 categories, preceded by a headword (often the hyperonym), introduce items with more restricted meanings, i.e. adding distinctive semantic features. However, the notional scheme used means not only that hyponyms are included and polysemic/homonymic items broken up into their individual senses, but that many lists of words are also included which have no semantic, only encyclopedic, affinity. Great expectations have been voiced regarding the enormous insights and research possibilities that the HTE will provide for scholars investigating semantic change in the history of English lexis. However, a few warnings may here be in order:
(1) The HTE is organized on a timeless notional basis and even where evidence for semantic change can be derived from individual chapters, this change is likely to relate to changes in worldview rather than meaning.
(2) Polysemy (and less importantly homonymy) are important motors for semantic change, including the increase and decrease of semes and sememes. Relevant lexical items can be spread over a great number of categories, though not all so impressively as in the 93 categories listing the notoriously diversified noun stock. Here the arrangement in the OED (whose editors tried to group together semantic developments) remains more helpful for the researcher.
(3) Categories in the HTE group together items of different classes:
(a) Enumerations of semantically unrelated items (persons, materials, legal diction, ecclesiastical terms, etc.). Since the classification here applied is notional, a semantic analysis even in the data collection phase would have to be based on a search of widely divergent categories. If I were to complement my study of ‘text types’ (Görlach Reference Görlach2004), which lists some 2,000 items, I would have to go to the HTE chapters ‘Communication’ (pp. 1483ff.), ‘Law’ (pp. 1367–86), ‘Religion’ (pp. 1460–83) and a few others – sections in which, on the other hand, only a few items would qualify for inclusion in my corpus.
(b) Lists of semantically indistinct/bleached items which contrast only in relative frequency or gradual intensity, etc.
(c) Lists of hyponyms whose individual senses are more or less in semantic opposition, contrasting with the hyperonym and with co-hyponyms. Sometimes the semantic unity of the heading is, however, doubtful: consider p. 915 ‘mind, soul, spirit, heart’, p. 926 ‘idea, notion, concept’, p. 927 ‘mental image, idea, concept’, and the like.
To summarize: the HTE is an invaluable tool for finding comprehensive material for studies of semantic change if the above warnings are taken into account. It can then yield exciting data at the interface of changes of worldview and meaning – if changes in speakers’ attitudes and stylistic parameters are also considered as fully as possible. In the past few years, the Glasgow team has been very generous in giving researchers access to their unpublished data for studies in semantic change: this has also brought some synergetic effects such as those from Diller's studies on the ‘emotion lexicon’ and Fischer's on principles of change. Corpora and thesauruses (cf. Diller Reference Diller2008) make available huge amounts of data and this leads us to expect a new boost to studies in historical semantics (and other fields of English historical linguistics) – hopefully, there will be enough scholars in future generations to make full use of the enormous supply of information assembled in works like the HTE.
For a work of this complexity, it comes close to a miracle to find how impeccable its standards of editing and printing are. Both editors and publisher are to be congratulated on producing this opus magnum at such a very reasonable price.
