In previous papers I have suggested that we might profitably consider the history of English lexicography from the perspective of bibliography; that is, that each discrete book bearing a unique title page should be seen as a single edition of but a single text, The English Dictionary. My argument proceeded from an analysis of John Wilkins' and William Lloyd's Alphabetical Dictionary, published in 1668 as part of a larger work, An Essay towards a real character and philosophical language. The Essay proposes a scheme for a universal language; it also contains a short history of English, a grammar in the tradition of Speculative and Philosophical Grammar, and a conceptually ordered list of words which could be called a lexicon, or more mundanely, a thesaurus. I would point out that just saying this much already indicates the range of questions that need to be addressed to adequately understand and appreciate this work. For instance, can the book be defined within a print tradition? Do the stated intentions of the author restrict our understanding of the text? Does the author propose a language theory by way of his method? Is this more a text of the state of the English language of the mid-seventeenth century, or more a text of the state of scientific knowledge of the same period?
The questions that are pertinent to deciding the dictionariness of the text complement the issues that could be raised and addressed in a more comprehensive analysis of the Essay. In this case, we must ask questions concerning the structure and function of the artifact. How closely does this text follow the conventions of the text type called The Dictionary? Or perhaps, does the book present multiple text types, thereby producing an indefinable and unique artifact? Since the multitude of texts that are conventionally categorized under the heading DICTIONARIES have only just begun to be studied as texts, these questions can not be answered with precision for any given text; however, most people know a dictionary when they see one. For the purpose of discussing a dictionary as database, we can agree that dictionary makers and dictionary users assume that the text has, as primary functions, the representation of knowledge (even if it is merely the knowledge of orthography or pronunciation) and the transmission of ideas (even if it is merely the idea of synonymy).
For the most part, the discipline of linguistics (and by extension, lexico-graphy) does not address the problems associated with determining or understanding the transmission of ideas within the field, much less does it consider its own bibliographical history. Of course, there are people working within the history of linguistics, but that field is notable for its lack of focus and for its emphasis on what is called historiography. There is important work being done in the history of linguistics, but my attention is drawn to the history of books and the representation of knowledge. The discipline of literary studies would seemingly offer methods to describe and explain this facet of representing and transmitting ideas; indeed, perspectives from textual and bibliographical analysis prove useful in framing questions about texts within the domain of language theory. When we consider how dependent representing knowledge as lexicographical formulae is upon so-called literary texts, especially as evidence for definitions, it is natural to conclude that lexicography as a discipline (sometimes referred to as metalexicography) must combine methods developed in both linguistic and literary theories. To add the concept of database to the discussion only accentuates the need for a methodology for interpreting lexicographical texts within a larger notion of knowledge representation. A theory of bibliography, considered in a comprehensive manner, could provide all the seemingly divergent methods and perspectives with a mutually supporting place for discourse.
In this paper, I will (1) briefly discuss the dictionariness of the Wilkins/Lloyd text; (2) consider the questions of how we represent meaning and knowledge; and (3) outline some of the textual problems associated with the transformation of dictionaries into databases. These separate points all are founded upon the notion mentioned above, that the analysis of English lexicography, especially as history, starts with bibliography. The definition of bibliography presents its own tradition of argument and counter-argument, but for this occasion I cite the following words by D. F. McKenzie (1986) as representative of the spirit of my inquiry: "[...] bibliography is the discipline that studies texts as recorded forms, and the processes of their transmission, including their production and reception [...] bibliographers should be concerned to show that forms effect meaning. Beyond that, this definition allows us to describe not only the technical but the social processes of their transmission" (p. 4). While McKenzie's argument is directed at the place of bibliography in textual criticism, I see a relevance to the study of lexicography. The phrase "texts as recorded forms" extends to all recorded forms, not merely the traditionally defined "book". The study of texts that record meanings in a variety of somewhat similar formats, texts we call dictionaries, calls us to account for the assertion that "forms effect meaning". I would add that forms also affect our perception of what a text means to us, in the sense that the cultural and historical context of the publication of a text becomes part of the form. One might suppose that form would establish function, or at least follow function: yet for those who would establish the canon of English dictionary, the mere existence of a title page announcing the intention of presenting a dictionary does not appear to be enough evidence of dictionariness for some people; we shall notice this later during the discussion of the Wilkins/Lloyd lexicographical text. The form of a text, and that includes any semiotic detail from italics to taxonomic structure, from microstructure to macrostructure, bears upon the second and third topics of this paper, representing meaning and knowledge and the dictionary as database.
Presently, the history of English lexicography, which after all is a history of a succession of texts, seems to exist outside of literary history; thus, one finds no tradition of textual criticism that is adequately pertinent to the study of lexicography. We can import theories and methods of whatever school of literary theory most interests us, but I find, when attempting this, that the application of a literary method redounds upon its own principles. Indeed, dictionaries are treated as extra-literary; dictionaries support interpretation: I am never quite sure what my colleagues in literature departments mean to accomplish when they send students to a dictionary, since they would probably never send a student to look up poetry (the oddity of the phrase look up poetry registers the difference between text types). Outside of the contentious stew of literary theory, we find that the loosely defined discipline of English lexicography has no standard of measuring the universe of dictionaries; no doubt the very history of publication and marketing has created this situation. It is within this context that I have argued for the inclusion of the Wilkins/Lloyd text in the domain of English lexicography. That the argument had to be made, and to some extent still needs to be made, reflects the sociology of the discipline more than the material evidence that can be brought to bear on the scholarship. Those who look no further than the 1946 publication of The English Dictionary from Cawdrey to Johnson 1604-1755 by Starnes and Noyes to establish the canon of early English lexicography will find a single mention of Wilkins, and that mention has nothing explicit to say about his relationship to the construction of dictionaries; rather, the citation refers to a faction in the Royal Society who would reform the English language. A recent facsimile reprint of Starnes and Noyes that includes a new bibliography of the history of early English lexicography omits all publications printed since 1946 and before the facsimile edition that discuss the place of Wilkins and Lloyd in this history. We must conclude that there must be a need, however unstated and tacit, to exclude the Wilkins/Lloyd text; there is no need for me to rehearse my argument for inclusion until such time as those who would exclude Wilkins reply to my previous claims (Dolezal 1985).
Turning for a moment to the concept of electronic database, we can further confuse the issue of what constitutes a proper dictionary when we conflate all word-lists, concordances, glossaries, thesauruses, and dictionaries (not to mention all non-alphabetically-ordered wordlists: narrative or not) into machine-readable form. From this perspective, it would be absurd to exclude the Wilkins/Lloyd text; on the other hand, it would be wise to consider the bibliographical shape of the database. Thus, we cannot escape the underlying questions implied by the creation of an electronic database. This leads me to the next topic of discussion: the representation of meaning and knowledge. Representing meaning, by which in this discussion I mean recording definitions, depends on textual form. For the purpose of our discussion, definition also includes any classification schemes or semantic diagrams, no matter the larger context. For example, definitions appear not only in dictionaries but in linguistic treatises, glosses of texts, and narrative texts. Moreover, the form, or structure, of a vocabulary creates its own narrative, even if that narrative is limited to revealing the human agency that constructed the form. In other words, a lexicon, a database, a concordance, or a classification system either presents itself (inasmuch as it is the creation of human agency) as an unstated theory of knowledge, or as knowledge and artifact. If form determines meaning, or at least is partially constitutive of meaning, then the way we capture the data for construction of databases becomes crucial; I will discuss this point later in this essay.
In order to illuminate the general point concerning the structure of presentation and representation of meaning I shall use the specific case of the Wilkins text. Werner Hüllen pointedly argues for a differentiation between thesaurus (the Wilkins text) and dictionary. I do not intend to criticize Hüllen's work, but it does merit comment. We do not have the time to fully explore the claims concerning the difference between dictionaries and thesauruses made by him and others, but this statement is relevant to the point I am attempting to make: "[Wilkins'] book is the first collection of words that in our understanding deserves the name 'thesaurus'. However, Wilkins' thesaurus does not give definitions at all but simply lists words in a certain arrangement" (p. 117). Hüllen refers only to the hierarchically ordered compendium of concepts and does not mention the part of the book that has its own title page with the title Alphabetical Dictionary; to be sure the dictionary is also an index to the lexicon, but to a large extent it is self-contained. However, that point aside, if we look only at the so-called thesaurus portion of the book, we find lengthy discursive passages integrated within the classification tables.
Those kind of Moral Habits which serve for the regulating of our Wills and Affections more General, are commonly styled by the name of VERTUE, Honesty Probity, Righteousness, brave; denoting such Habits whereby we are inclined and inabled to observe a due Mediocrity in our Actions. To this is properly opposed the notion of VICE, Sin, Crime, Dishonesty, Trespass, transgression, Faulty, Failing, Infirmity, Oversight, wicked, Improbity, Turpitude, unrighteous, unjust, bad, naught, vile, base, loose, evil, ill, corrupt, venial, heinous, debauched, lewd, lawless, licentious, foul, flagititious, enormous, profligate, Miscreant, Ruffian, Caitiff, Villain, Rakehell, Libertine, defile, pollute...
The thesaurus-like nature of this passage is obvious; however, just as evident is the dictionary-like text. The word-list section follows a general pattern that Wilkins explains in a chapter on interpreting the text; usually, each separate item in lists as exemplified above is included within the Alphabetical Dictionary. The entries provide one-word paraphrases with appropriate grammatical and semantic markers that distinguish among words within and across semantic domains. The preceding illustration from what are called 'Philosophical Tables' appears as an introduction to the concepts that appear within the category of 'VERTUE' (within the system, 'VERTUE' is formally considered a Radical or Integral, that is, an atomic or elementary semantic unit). Within the subcategories, we also can see dictionary-like text; for example, before the lexical items associated with the subcategory 'FORTITUDE,' Wilkins writes: "whereby we are made duly resolute against all such difficulties either of Fear or Discouragement as may hinder us in our duty". Unlike Roget's thesaurus, Wilkins gives discursive text introducing and connecting lists of semantically related items; these commentaries tell the reader how to understand the formal organization on the page (a guide to knowledge) and have the textual elements recognizable as a definition text-type.
These illustrations are representative of the whole; it probably does not matter if we call this text a dictionary, a universal language project, or a thesaurus; from a bibliographical perspective I cannot see how the text can be excluded from a history of lexicography. 'Dictionary' may imply a specific sort of book; 'Lexicography' implies a process that may culminate in a book, but it just as easily could be a process that finds completion in an electronic database. In the case of the Wilkins project, we certainly can understand the text as representing a social, and thus human, construction of a knowledge system.
The specific example tells us something useful about how meaning can be presented and represented. The print medium allows for a limited repertoire of forms. How these forms signify has not been adequately investigated; however, the forms are recognizable as a semiotic system, even if that semiotic system has to be reconstructed from our intuitive understanding of what a dictionary or lexicon or semantic field looks like. Lexicographical texts do not just passively mean; we understand that their function requires that they intend to show meaning. A dictionary that is also an historical artifact should have a place in the textual criticism of its contemporary artifacts: a definition written in the seventeenth century may be a better source for determining the meaning of a word in Milton than a definition constructed on historical principles in later centuries. On the other hand, the epic poem as a method for representing knowledge may have as much validity in some cultures as a grammatical diagram. Surely, we understand more about intellectual life in England if we realize that one twelve-month period comprehends both Paradise Lost and Wilkins' Essay. They both are informed by the same semiotics of printing at a basic level of analysis. They differ, however, in the larger forms: clearly, one looks more like poetry and the other looks more like lexicography. I find that the efficacy of Wilkins' definitions and lexicon can be tested against the meanings written in Paradise Lost; the definitions and the conceptual/semantic system in its turn have the potential to open up Paradise Lost. The dictionary, or thesaurus, or lexicographical text, however one prefers to label the work, exists within an intertextual frame that encompasses all previous texts and the development of texts that become a history of lexicography. Any lexicographical text implies a bibliography, a print history of each recorded word; as a printed record itself, a dictionary becomes part of a bibliographic lineage of inscribed meaning.
The third topic, the transformation of the historical record and the method and structure of meaning into what is called a database, points to the potential of enhancing our understanding and use of historical dictionaries. A master list of all terms and phrases collected in early English dictionaries will allow us to begin the process of measuring the scope of dictionariness; moreover, matching definitions and vocabulary is a task well suited for a computer-driven analysis. However, the consequences of losing the historical form and structure promotes a transformation of the meaning of the texts. This does not mean the transfer from one medium to another is necessarily good or bad; rather, we are creating our own text from historical material. In this way the canon of the English dictionary is realized not as a book or set of books but as an ever-changing database. The limitations of form imposed by the printed page no longer guide the methods of representing meanings and knowledge. The negative consequence of losing the historical form could be balanced by the opportunity to construct richer forms; but we should decide beforehand just what purpose and use there is in a richer form (a task not avidly pursued by print-dictionary makers in their own right). On the other hand, I fear that the textual complexity of a work like the Essay by Wilkins will once more put it in the margins of English lexicography. It was constructed to be a database for a universal language; it was consulted as a lexical database by later lexicographers, and it inspired the work of Roget among others; while it makes sense to make it available in a master vocabulary, it also demands to be preserved as a systematically ordered database. In order to accomplish this requires a method to translate a highly integrated lexical print database into an equivalently powerful electronic database. In this case, a seventeenth-century language system (or scheme, as Wilkins would say) challenges the scope and method of twentieth-century theories of meaning and representation of meaning.
 One cannot even be sure of the primary use and purpose of dictionaries today, much less divine answers from the past (see Greenbaum et al. 1984); does the dictionary have an ideal form? an identifiable discourse strategy? a recognized social practice or function?
 No doubt the general structure conforms to the organization of the best known English thesaurus by Peter Mark Roget; conforms is a rather odd word for a work published 150 years or so before the 'model' thesaurus. More important, the index to the semantically ordered "Philosophical Tables" reads very much like the alphabetical dictionary that it is claimed by its authors to be. To comprehend the dictionary fully, one does need to understand the lexicographical metalanguage used throughout; the dictionary does have the quality of an index in that it relies heavily on cross-referencing and one-word paraphrases (or synonyms). The total impression of the work is more reminiscent of Mel'cuk's Explanatory Combinatory Dictionary (Steele 1990) than Roget's thesaurus, given that a work published over 300 years before the ECD can be reminiscent.
 McKenzie (1986) makes this point: "Whereas libraries have held books and documents as physical objects, computer systems have been mainly concerned to retrieve content" (p. 60). Of course, dictionaries are unlike most books and documents: they already are systems designed for content retrieval. This does not invalidate McKenzie's idea here, but once again reminds us of the complexity of a text-type that might best be named heterotextual.