0. Introduction

The dictionaries of earlier periods are gold mines of information for a variety of types of research. Dictionaries are repositories of the 'word-hoard' of the languages included, providing unique 'holistic' insights into the languages themselves and the cultures those languages are a part of. Dictionaries are also most difficult sources to utilize for these broad spectrum studies. In the past, studies of lexicographic technique have depended on small portions of extant dictionaries, studying all words beginning with ba for example,[1] or ferreting out all references to dialect (but omitting or ignoring all other types of reference to register, style, etc.) Faced with that alternative, computers are not only a means to avoid some of the mind-numbing aspects of the study of dictionaries; they are more importantly the tool to open new ways of viewing the text. They permit us to try textual explanations and interpretations that surpass the best efforts of human memory. This line of research can make these dictionaries, viewed as complete and semi-discrete compositions, key pieces in ever more abstract intellectual discussions of the nature of lexicography and linguistics, the nature and uses of history, the nature of knowledge and/or understanding. The databases comprising these dictionaries, therefore, need to be constructed so that the large questions, as well as the small, can be answered.

We are used to considering dictionaries as tools, yet they are something more as well: these texts bear, as a sort of summa of the cultures they represent, many similarities to a canonical literary text; the use of the analytical tools and theoretical viewpoints of literary and cultural history should bring these documents new attention from a variety of scholarly approaches. Beyond the information contained we need to consider the textual traditions. We need to ask ourselves what choices the authors made and why they made them. In short, to borrow a concept from illocutionary semantics, what was the author doing in saying what he said (and in omitting what he omitted)?[2] Each of the scholarly disciplines that has something to learn from these texts shares some needs in terms of the computerized analysis of the text, and each has certain peculiar needs; yet all interested disciplines stand to benefit from the holistic approach to their contents. To answer important theoretical questions through computerized access to these peculiar texts demands a variety of search and retrieval strategies. Performing those searches, in turn, requires a more complicated structure of the database than we have heretofore accomplished, including cross-referencing to external materials and the ability to retrieve 'negative' tokens (instances in which a feature might have been included or marked, but was not).

The creation of such a structure is beyond the requirements of the preparation of an edition by a single scholar, and therefore has implications for the distribution of these databases as well. Those features which describe formatting and basic structure are part of the electronic edition which can be distributed in fixed-state media: on diskettes or CD-ROMs. Those features which are essentially for interpretation presuppose ideally an interactive network where scholars can insert their interpretative feature markers for other scholars to use, discard, or discuss as they will.

What I propose to do in this article is to present an inventory of the lexicographic features of dictionaries that include French and English in the 16th century,[3] and then to outline the needs of different scholarly disciplines in approaching this genre of text. The basic structure is necessary for those wishing to do research using the dictionary as a tool; the interpretative structure is necessary for studying the dictionary as a cultural and scientific production.

In these texts, we can start with a study based on five criteria,

1. What languages are included and what is the order of presentation?
2. What is the source and nature of the word list? (Relationship between this dictionary and others in its textual tradition; changes in the concept of 'hard word'.)
3. What is the organizational principle of the word list and the entries within the word-list?
4. How has the author/printer availed himself of the new technology of that day, printing?
5. What information within and beyond translational equivalent is given, either for the word in the source or in the target language?

each considered from explicit and implicit evidence:

1. What are the author's explicitly avowed purposes and techniques in creating the dictionary?
2. What are the purposes implicit in the contents of the dictionary?

The sum of the answers to these questions responds to the primary problem we have set for ourselves: What was the author doing (lexicographically, pedagogically, practically and culturally) in composing his dictionary? Lexicographically, the author must justify why a new dictionary is necessary, and why he has made the choices he has in terms of the lexicographic tradition. Pedagogically, most of these authors conceived of their works as part of an instructional programme in one of the languages, although the degree of integration varies widely, from inclusion within a grammar (Palsgrave, Du Wes, Hollyband 1573 and 1576), to independent part of an instructional series (Hollyband 1580a and 1593), and ultimately to those more independent yet (Veron, Baret, Huloet-Higgins, Junius-Higgins). The relationship between the dictionary and language instruction influences many of the decisions on lexicographic practice.

Practically, many of the authors saw their works as aids in translation. In some instances the translation activity was professional (Palsgrave -- for work as clerk in the Inns of Court or Chancery), in some educational (as part of the instructional program -- Baret and Hollyband), in others yet part of a higher intellectual calling (as in the case of Stephanus himself). Baret demonstrates his interest in providing a tool for translation, from Latin or French into English, right on the title page:

An Alvearie or Triple Dictionarie, in Englishe, Latin and French: Very profitable for all such as be desirous of any of those three languages. Also by the two Tables in the ende of this booke, they may contrariwise finde the most necessary Latin or French wordes, placed after the order of the Alphabet, what so ever are to be founde in any other Dictionarie: And so to turne them backwardes againe into Englishe when they reade any Latin or French Authors & doubt of any harde worde therin.

In the introduction to the reader Baret goes on to explain that the dictionary grew out of the translation exercises he used to give his Latin students (the 'busy bees' remembered in the title of his work).

Hollyband's dictionaries are also organized to facilitate translation from French into English, so that students might take advantage of the pearls of French literature that Hollyband recommends: Commines, Amadis de Gaule (already translated into French from Spanish), Marot, Rabelais, Boaistuau. In addition to the basic disposition of the languages (French headwords -> English equivalents), Hollyband explains in his introduction "to the Students of the French tongue" that he has included the principal parts of the French verbs, because the infinitive alone does not help one much when faced with a conjugated form:

In other dictionaries thou hast onely the Infinitive moode of any Verbe, as Aimer, to love: Vouloir, to will or be willing [...]: and yet for all that thou art not much the wiser, because thou knowest not how to begin in the present tense, so that thou canst say in French, to love, Aimer: but thou doest travaile to say, I love, thou lovest, I have loved [...]

In another motivation which Palsgrave regarded as practical but which we might regard as cultural, he explicitly recognizes the social force of the grammar/dictionary he prepared, as a definer of acceptable usage. He is the first, on either side of the Channel, to bring French under the control of "rules certaine", in accordance with the wishes of Geofroy Tory (Champ Fleury).

Finally, culturally these works define the relationship between the languages involved, define linguistic authority, and express a broad range of views about history, literature, religion and a host of other topics. The lexicographers are both reflections of and commentators upon the society represented by the languages of their texts. This richness is a source both of our fascination with their works, and of the complexity of devising an adequate computerized information-retrieval system.

1. Lexicographic Practices

In this first part, I shall present an inventory of features, both common and distinctive, found in these dictionaries. Some authors address the lexicographic issues directly, either in the preface or elsewhere in the body of the text, but the answers provided within the dictionary are usually more telling than those furnished in the prefaces. As we present such features, we must constantly ask ourselves why the authors have chosen to do what they have done. Without such inquiry, we will find only isolated nuggets (however shiny!) and miss the rich seam of words.

1.1. Which languages will be included?

None of the lexicographers studied here justifies the choice of languages, although Baret laments the exclusion of many Greek equivalents in the 1573 edition, a failing which he blames on his printer. The order of languages is one way of showing the direction of translation (others are the nature of the equivalencies, the language of examples, and supplemental means of access).

• French-English
• Hollyband 1580a
• Hollyband 1593

• English-French
• Palsgrave 1530 (Pronouns both French-English and English-French)

• English-Latin + French
• Huloet-Higgins 1572
• Baret 1573/4
• Baret 1580

• Latin-English + French
• Veron 1552

• Latin-Greek-French-English
• Junius-Higgins 1585

By this evidence, Palsgrave, Baret, and Huloet-Higgins are the ones primarily interested in helping English speakers write in French. Why? Palsgrave specifically mentions the clerks, and foreign trade. Baret seems to have had close connections with the legal profession, for he recruited help for his dictionary among his former students who had moved on to the Inns of Court. 500 years after the Conquest French was the language of the law, and the ability to translate into French therefore remained an important skill. The Hollyband dictionaries are the only ones to go in the opposite direction (French -> English). Why? Hollyband was not a product of English legal training, but rather a religious refugee. For him the study of French was a matter of cultural edification, to encourage young Englishmen, like the son of his patron, to read, in their spare time at university, the great works he proposed to them. Should we study in the same framework those texts that start with Latin, or include Latin? Certainly, in the case of Baret, for he notes that the purpose of the French index is to help translation in the direction French to English, and we can assume that the addition of French to his original source (Thomas Elyot) was meant to facilitate translation in the other direction. The same point can be made about Higgins' additions to the dictionaries of Richard Huloet and Hadrianus Junius: French equivalents were added because they were deemed useful to the intended audience of English speakers. The only dictionary in which the relative importance of the French and English equivalencies is in doubt is Veron's, in which both of the modern languages are related to the Latin. Still, the English was created by translating from French to English, rather than by drawing the Latin-English equivalency independently. Even if the work was not intended to facilitate translation between the modern languages, it was the product of such translation. Furthermore, Veron could have simply omitted the French to shorten and simplify the dictionary if he were not interested in relating French to English.

The languages and the direction of the dictionary is one element, like the date and name of author, that should be included in the 'status' window of any computer database. If we switch from an entry taken from Palsgrave to an entry taken from Hollyband, we must be reminded of these crucial facts. Suppose we were comparing lists of vocabulary relating to religion in those two dictionaries. To compare the lists we need to look at all the words of one language. This involves flip-flopping the entries from one of the two dictionaries (already a difficult enough task, see below, 1.3). One dictionary's word-list will not be presented in the original order, and this needs to be clearly drawn to the attention of researchers.

To summarize this section, there was a point to creating dictionaries relating French and English, a point to adding French to previously existing English and Latin dictionaries, and a point to adding English to the one example of a Latin-French dictionary. The creation of these linguistic relations reflects aspects of English society. There was a need, legal, intellectual, and sociocultural, for translation into and out of French. This need can be further specified by a look at the nature of the word lists themselves, and this aspect of the texts must not be lost by the organization of the database.

1.2. Which words will be included?

In their prises de position few of the lexicographers discuss the nature of the words they will include. Some are open in identifying their base sources, literary or lexicographic, others are more secretive. Some identify the nature and source of their additions; only Higgins in his revision of Huloet mentions deletions. In the second half of the century, the question of 'hard words' is discussed openly for the first time, but the term has different meanings for different lexicographers.

1.2.1. Base vocabulary

The ultimate source for most of the vocabulary included in these dictionaries is Latin literature, history, philosophy, and natural history. This bald fact has many repercussions, social as well as lexicographic, but let us focus here on the lexicographic trail. One of the primary goals of Latin lexicography in the Renaissance was the purification of the word stock, that is, in good Humanist tradition, the purging of medieval Latin words, examples and usage. This was Stephanus' motivation for his dictionaries, from which Veron's and Hollyband's dictionaries are derived, and Baret's reason for reworking Thomas Elyot's Bibliotheca (which itself is largely drawn from Calepino). Higgins corrects Huloet (who himself drew on both Elyot and Calepino), and insists on the fact that he went directly to authority, although these authorities were sometimes contemporaries (including Stephanus).

For Palsgrave and the other grammarian-lexicographers (Du Wes, Hollyband 1573 and 1576), the Latin sources are more removed. In these cases the ultimate sources for the 'practical' vocabulary are probably the medieval nominalia. To this small base Palsgrave adds a large corpus of literary examples from English and French literature, although some of these too are based on Latin sources, as we see from this example from the Aeneid via Jean Lemaire des Belges provided in the table of pronouns to illustrate the use of lequel:

Comme fist Anchises, lequel incontinent apres le fait s'alla vanter Lyke as Anchyses dyd, the whiche incontynent after the dede went and vaunted him

However, we must not confuse the source of examples with the source of the word-list. Palsgrave was, after all, trying to teach correct usage of French, something demonstrated by French, not English examples. Even if the examples are most often drawn from French authors, the ultimate base must have been English, as proven by the English words for which no French equivalent is presented. That English source, or those English sources, are still uncertain. Baret mentions that if anyone wants to correct or add to his book, that person should take examples from 'good Authors' (all classical), such as Cicero, Terence, Cæsar and Livy.

1.2.2. Additions

Higgins is particularly forthcoming about the nature of his additions to Huloet's dictionary:

for the better attaining to the knowledge of words, I went not to the common Dictionaries only, but also to the Authors themselves, and used therein conference with them which wrote particularly of suche things, as the place requyred. As in herbes I followed the iudgement of Master Turner, Dodoneus, Pineus, Fuchius and sometime Plinius: In beastes, fishes, and foules, Conradus Gesnerus: In instrumentes of warre Vegetius: In building Vitruvius: generallye Hadrianus Iunius: In names of places Ptolomeus: In the nature of wordes Valla & others: In Phrases Thierre out of which I confesse I have taken a great parte of this Booke, notinge the Phrases wyth S. after the Frenche, for that Stephanus was the first Author thereof: and finallye I wrote not in the whole Booke one quyre, without perusinge and conference of many Authors.

This is important not only for the list of sources, but also for the list of subject areas considered essential knowledge, incompletely covered in earlier dictionaries, and/or sufficiently changed in the course of the 16th century to merit special attention: botany, zoology, war, architecture, geography, philology. To create a computer database that can provide the data necessary to interpret these changes, we need to tag words within these categories from the sources noted, and words that seem semantically related but which have not been so identified (e.g., new Italian architectural terms which entered French over the course of the 16th century). This would clarify how Higgins used his sources (in this particular instance) and help us to understand Higgins' lexicographic goals and the society that shaped these goals. The dictionary provides word meaning and equivalence, and an index to the sum of knowledge expected of an educated man (for the development of the index and its importance see Rouse & Rouse, 1982a).

1.2.3. Hard Words

A further element in the determination of which words the dictionary must include is the 'hard word' principle. In these dictionaries it is not so much a case of including hard words as it is of excluding 'easy' words or features of 'easy' words. The 'hard word' principle is first mentioned in English-French lexicography by John Baret (1573). Baret talks of omitting the 'easy' words, by which he means the most common pronouns, conjunctions and the forms that can be derived automatically (e.g., participles from verbs):

In the tables ye shall finde few harde wordes lacking that are in any other Dictionarie. And as for common easie words, as Ego, Tu, Et, Si &c and Participles also, as Ambulans, Currens, &c I thought it not meete to stuffe this booke with them [...]

Of those words which Baret considered too easy for inclusion, Palsgrave includes all except the participles. ('Participles' and 'articles' are the only parts of speech for which Palsgrave does not provide a 'table'.) He does however, omit examples for and / et because of the regularity of the equivalence:

And: Et. Sans. Ensemble. Que. Et, wherof nedeth no exemple/ for it is in maner euer generall that et, countreuayleth and/ and these other be but seldome used.

For the rarer equivalents Palsgrave does provide copious examples. Hollyband also explicitly recognizes the difference between 'hard' and 'easy' words, but the distinction is made on other, morphological, grounds. He announces in the introduction that he will provide the principal parts of the difficult verbs.

The source of the word-list is a major link between the dictionaries of the period, but we must not let the fact that most of the words have been taken from a specific source blind us to the analysis of the words that the lexicographer has added, subtracted, or altered. To understand the choices made by any one lexicographer, we must be able to retrieve these lists of added and subtracted headwords, and just as crucially, added and subtracted materials within entries. We cannot assume that it is by chance or incompetence that the authors made these changes. To judge the author's purposes, the database must be constructed so that we can recover, for example, all French words from Estienne 1539 that Hollyband omitted in the text of his 1580 dictionary, as well as all French words he included that Estienne omitted; all words drawn from a specific subject authority, as well as words semantically related but not so designated, etc. The analysis of those patterns will enable us better to determine what the lexicographer was doing by making his dictionary.

1.3. The organization of the text: means of access

The organizational principle depends on the textual tradition and the intended use. Variation occurs in the degree of alphabetization, the disposition of words in a single word family, and the separation of lemmata into distinct entries based on differences in meaning. The choices for the textual tradition are (1) the medieval nominalia, in which case the order is by subject matter; (2) a growing medieval tradition for reference dictionaries, organized, at least partially, in alphabetical order;[4] (3) a grammatical tradition in which the words are listed alphabetically according to parts of speech (Palsgrave, and, in part, Du Wes).

1.3.1. Alphabetization

By the end of the century the principle of simple alphabetization was dominant. The word lists were too long, the indexing task too complicated, and the semantic classes too arbitrary to accommodate classified vocabulary. As for the combination of grammatical and alphabetic ordering, the connections between related words in different parts of speech were and are too involved to access easily the words one might search. Alphabetization is generally preferred, although the skill at alphabetizing varies greatly. Higgins mentions the many corrections he had to make in the order of words in Huloet; generally alphabetization is getting better by the end of the century, and this is important for the use of dictionaries as reference tools.

One special problem in alphabetizing is determining which is the head word in an expression. In the table of substantives, Palsgrave must have confused some by listing the following nouns under O for the adjective modifying them:

• Opyn audyence
• Opyn seme
• Opyn courte
• Opyn house
• Opyn warre

Particularly puzzling is the fact that courte should follow audyence alphabetically, and both have the same equivalent (court planiere), yet they are separated by seme. Equally disconcerting for the modern reader is the long list of expressions "I am + adjective/participle", among which are interspersed verbs starting with am- (amass, amaze, amble, amend, etc.). Whatever our modern reaction, we cannot assume incompetence on the part of the Renaissance lexicographer. Palsgrave's organizational pattern for verbs indicates a different perception of the combination "to be + adj." This may be the result of equivalence to a single verb, either in the French word proposed as equivalent, or in a Latin word that served as the equivalent in some source dictionary. Another explanation could be a different grammatical analysis of the copula. To recover his intention, we need to tag both the instances in which he treated these forms as single units, and instances in which he didn't, but might have (the 'negative tokens'). Then we can relate these lists to the table of adjectives he provides, to combinations of "to be + adj." translated by a single Latin or French word in other dictionaries of the period, etc.

The alphabetical index of the target language words marks an important step towards reversible bilingual lexicography, one which Baret explicitly recognizes on his title page (see above). The limitation of Baret's indexes is that he has drawn those word lists from Estienne's French-Latin, Latin-French works, but the basic word-list (in English) comes from Elyot. Therefore some words appear in the index which do not appear in the dictionary, and many words appear in the dictionary that do not appear in the index. Junius-Higgins offers an index only of the Latin words, making it useless for finding French-English equivalents.

1.3.2. Word families

Another organizational principle receives no explicit attention from the lexicographers studied here, and that is the listing of related words under a single headword. This is very important in the Estienne tradition, less so here in the English tradition. Related forms, such as deadjectival verbs, or deverbal adjectives and nouns, generally get their own individual listing, as here in Hollyband 1593:

• Feste, holyday: f.
• Fester, to keepe holyday, to feast
• Iours Festez, dayes being made holy, holydayes
• Festin, a feast, a banquet: m.
• Festoyer, or fester aucun, to feast one, to cleare one: also to celebrate a holyday.

Baret recognizes these related words by indenting them and preceding them with an asterisk, a practice that sometimes reveals the Latin source for his dictionary. The words listed as 'related' are sometimes related morphologically in Latin, but only semantically in English, as we see in this example:

A BAYNE or stewe: a washing place [...] Balneum
*The place in the house where the bayne or stew is. Balnearium
*Perteyning to baynes or stewes. Balnearius
*The mayster of baynes or stewes. Balneator

However, in other cases, these words are related only in English:

The BACKE of a man or beaste. Dorsum
[...] to go backe. Retro cedere [followed by many other examples of return motion using the Latin retro]
*One that runneth backe, and yet will not give over. Tergiversator
To be put back in sute for an office [...] Pati repulsam
Resty moyles going backe. Mulae cessim euntes
To write in rough paper: that stoppeth the penne: also to write in the backe side: to indorce. In aversa charta scribere
*A backe dore: a posterne gate. Posticum
*A returning a comming backe againe. Reversio
*To keepe backe [...] Demotor & Remotor
[etc.]

1.3.3. Meanings and entries

A similar problem occurs when a single headword has a number of different equivalents in different contexts. The choice is to present all the equivalents as if the object referred to were the same, or to list each meaning separately. Veron chooses the former, listing all the various equivalents of clavus in a single entry:

Clavus, clavi, m.g. a nayle, the sterne of a shyp, a byle in the bodye, a default in trees procedyng or commyng of the burning of the same, a knappe of purple in fourme of a nayles head, wherof great number were embroydered in the robes of the senatours of Rome.

Given a similar choice, Palsgrave lists each word separately, even if the English word is morphologically and derivationally one:

flete where water cometh breche
flete of shippes flotte de navires
flete a prisone for gentylman consergerie

The construction of databases that handle multiple dictionaries from this period presents a special problem: because of the variation in the language of the headword, we need to be able to match alternate sides of the a = b equation. What is a headword in one dictionary will be an equivalent in another, but we must be able to pair them up while excluding (for this purpose) words found in the examples. That is, the simple index of words in the dictionary is not enough; the words have to be tagged as headwords or equivalents. This becomes especially complicated when synonyms are provided both for the headword and for the equivalents in the target language. How does one match up the headwords and the equivalents in examples like the following?

Hollyband 1593: Fiance & asseurance, trust, affiance, troth.

Similar match-up problems will occur when the equivalent is a phrase or explanation, not a translational equivalent:

Hollyband 1593: Fleurdeliser, to burne one with an hot floure-de luce betweene the shoulders, as theeves or roges are burned in the hand here in England.

Un Cepier, he which keepeth, or hath charge over the stockes.

For the purposes of comparing entries in which the order of languages is reversed, is it sufficient to mark the English word here as to burne (for fleurdeliser) or, clearly unacceptable, he (for cepier)? The variety of equivalents poses a serious problem for the computerized matching of the contents of different dictionaries.

The lexicographers had a number of choices in organization and presentation: alphabetization or classification; supplementary means of access to words in the target language; disposition of words within a single word family (source or target language); disposition of different meanings/equivalents for a single word (of the source language). The selection, in each case, and variation within individual works, depends on the goals of the lexicographer, and not just narrowly lexicographical goals. Unless we mark the absence as well as the presence of important features we cannot recover that level of intention. Furthermore, modern access is hampered by the different types of equivalency proposed, making it difficult to perform the flip-flop necessary to compare the contents of dictionaries with different language orders. Given the complexity of the indexing task, one understands better Baret's solution, taking the word-lists from Estienne's dictionaries and using them as indexes.

1.4. The use of print technology

As the century wore on, as print technology developed and as print standards emerged, the lexicographers took more advantage of the options available to them.[5] From the simple alternation of size or boldface, they progress to more complicated combinations of these, plus the addition of Italic fonts. At the same time, the use of abbreviations in normal use declined (tilde over vowel to represent the combination of vowel + nasal consonant; y to represent the or that (Palsgrave, Du Wes)). Abbreviations to mark grammatical categories such as gender and conjugation remained. In the second half of the century Baret and Huloet-Higgins used special symbols, such as the asterisk, the paragraph bar, a pointing finger, and a barred C. Baret used the paragraph marker to denote the prime entry within a single word family, and the asterisk to mark secondary meanings or uses. He then used the pointing finger to highlight proverbial examples. Huloet-Higgins used the asterisk to mark obsolete words, and the barred C to indicate examples. As discussed earlier relating to other features, we need to tag those instances in which these symbols might have been used, but were not.

Another convention that simplified access was the addition of alphabet markers at the head of each column -- B ante A, B ante E, etc. Finally, some hidden ways in which print technology influenced lexicography were the ease with which one could steal the word-list (and thus expand upon it), and the expansion of the reading public, which gave the dictionary far more power as an instrument of authority and regularisation.

1.5. Information within the articles themselves

The information within the article is a cornucopia, or a Pandora's box, depending on the mood of the day. I have divided the information into seven categories, but it could easily be more:

1. Pronunciation and Orthography
2. Morphology
3. Syntax
4. Semantics
5. Language Variation -- Sociolinguistic and Diachronic
6. Examples
7. Encyclopedic Information

1.5.1. Pronunciation and Orthography

Unlike their modern counterparts, 16th-century dictionaries rarely indicate pronunciation or orthographic variation. Pronunciation is mentioned only when there is some potential for confusion. For instance, if the word includes vowels in hiatus, the number of syllables might be marked:

Regratier, of foure syllables, to thank one.

On very rare occasions Hollyband also notes letters which are not pronounced, particularly when they would normally be pronounced according to the rules he presents in his De pronuntiatione (1580):

Saoul, but pronounce Sou, full, drunke, strout: m.

In the short word list of the French Littleton (1576) Hollyband regularly marks silent letters by placing a small cross underneath consonants and vowels that are to be left silent. At the same time, he devotes a significant portion of his introduction defending traditional orthography, however badly it corresponds to pronunciation. In the dictionaries of 1580 and 1593 Hollyband also notes special aspects of vowels, either length or quality:

Louche & bigle, he that looketh a squint, goggle eyed: m Pronounce it long Femme, a woman, a wife: but pronounce fame: f.

As for orthography, Palsgrave notes examples in which French orthography varies, usually to condemn one spelling in favour of another:

I Flye from myne enemye or I flye for feare I have of any daunger: Ie men fuys, nous nous fuions, ie men fuys, ie men suis fuy, ie men fuyray, que ie men fuye fuyr, or fouyr, verbum medium. tercie coniu. whiche thoughe I have written with U/ only in the seconde boke to put a difference in orthographye bytwene the tenses of ie fouys, I dygge. yet Alayne Chartier and Johan le Mayre confounde the orthographye in these verbes [...]

Veron repeats Estienne's observations concerning variation in Latin orthography:

Bipartio, bipartis, bipartiui, bipartitum, bipartire, to parte in two, Partir en deux. Some doo wryte, Bipertio & bipertior.

The rarity of comments on pronunciation and orthography reflects the separation of functions within the teaching curriculum. Those questions were normally dealt with in other volumes, such as Palsgrave's first book, or Hollyband's treatise on pronunciation. The use of special indications for silent letters in Hollyband's instructional books was a way of defending non-phonetic orthography in the face of orthographic reform movements both in France (Meigret, Ramus, Rambaud) and in England (Hart). For Palsgrave, the regular references back to the 'rules' of his first book are part of his campaign to bring French under the control of rules, something his chief rival at the court, Gilles Du Wes, proclaimed impossible. To understand why attention was drawn to these particular phonetic and orthographic problems, and not to others which might have been so noted, we need to be able to separate out all instances in which such comments are made, and to check these sources against reference works such as Fouché's Phonétique historique du français.

1.5.2. Morphology

Morphological information is specifically represented by the lexicographers in a number of areas, either by abbreviation or by listing of variable forms. I am not including here the morphological variation found in example sentences or phrases, although that might be worth marking in a database.

Palsgrave and Hollyband mark the gender of French nouns by the use of standard abreviations, and Palsgrave regularly provides the mark of the plural (s, z, or occasionally x) after the singular form. Veron and Baret use abbreviations to indicate the gender of Latin nouns; sometimes Baret uses an article before the French noun to indicate its gender. For words that have gender-specific forms, or concepts that have gender-specific nouns, the tendency is to provide both, either as separate entries (example from Palsgrave) or within a single entry (example from Hollyband):

Palsgrave:

Horner a maker of hornes. cornettier s ma.
Horneresser a woman cornettiere s fe.

Hollyband:

Beuf: m: ou Vache: f: an Oxe or Cow
Adolescent, m: ou adolescente, a yong man or maide

Veron, Baret and Huloet-Higgins indicate the declensional patterns of nouns in Latin by providing the genitive singular form. Roughly the same pattern holds for adjectives as well. Palsgrave lists variant forms of French adjectives, while all the dictionaries that include Latin do the same for that language.

As for verbs, Palsgrave is the only one to identify verbs by conjugational number, which serves as a reference to the grammatical portion of his work. Palsgrave also provides lists of principal parts ranging from one (the headword entry is the first-person singular present indicative), usually for compound verbs when the simple form is listed elsewhere, to three (first-person singular present and passé composé, + infinitive) for regular verbs, to as many as eight for irregulars. Hollyband announces in his introduction that he will provide principal parts for the difficult verbs, but he is inconsistent. Generally, he furnishes the first-person singular of the present, passé simple, and passé composé and future indicative, in addition to the headword entry of the infinitive. Veron and Junius-Higgins list five principal parts of the Latin verbs: first- and second-person singular present and first-person singular perfect, the past participle and the infinitive. Huloet-Higgins lists only the first two forms of the present, and Baret often limits himself to the infinitive. Again we must ask ourselves why the authors made the choices they did, concerning which languages would have morphological detail listed as well as the type and form of detail listed.

For Palsgrave and Hollyband, the primary purpose of the book is to help English speakers use French correctly, therefore morphological information is provided only for that language and not for English. For both of them as well, the variety of information provided in each individual case depends on the morphological detail provided elsewhere, either in the same book (as in Palsgrave) or in separate volumes (as in Hollyband). It is interesting to note here that for Hollyband the variation of the noun and adjective was generally a topic for the treatise De pronuntiatione (which is just as concerned with orthography as with pronunciation), while the verb conjugations are described in a different volume devoted to them. For the others, in which only the Latin forms include morphological information, we can point to the textual traditions on which they are based to explain the presence of declensional and conjugational forms. The lack of conjugational numbering as an abbreviatory convention points to confusion in the numbering of the declensions and conjugations, and to the lack of a direct connection between the dictionaries and the grammars of Latin being composed at the time. To test our hypotheses, we need to be able to sort entries by the type and quantity of information provided. This requires adding a tag to verb entries (for example) to represent the explicit morphological content of the entries (as opposed to the implicit information carried in example sentences and phrases): what forms, how many forms, in what language.

1.5.3. Syntax

Within the dictionaries syntax is often explained implicitly by the use of examples, more rarely explicitly by structural description. Because Palsgrave combined the dictionary with the grammar, there is far greater insertion of syntactic information into the lexical portion. The entire pronoun section is cross-referenced with the grammatical rules presented earlier in book III. In the table of verbs Palsgrave includes a number of long explanations, particularly about compound or periphrastic constructions:

I Do/ is a uerbe moche comenly used in our tonge to be put byfore other uerbes: as it is all one to say/ I do speke [...] and I speke: but in the frenche tonge they use never to put any uerbe that countreuayleth/ I do/ in this sence/ but use the uerbe selfe onely lyke as for I dyd speke [...] they use euer the preter imperfyte tence or the indiffynite tence of the uerbes selfe in their tonge and neuer use to put any uerbe that countreuayleth I dyd in this sence [...]

I Go/ Ie men uas, conjugate in the seconde boke/ & note that like as it is comenly used in our tonge to put this uerbe/ I go/ byfore our uerbes/ where we use no mouynge to a place/ so use they to put the tenses of ie men uas, byfore their partyciples of the present tence/ as the Romant/ Et uont chantans, for chantant, et lors ua deprisant les dames, for lors deprise les dames &c. M. il alla mourir.

Hollyband, repeating a commonplace from the Latin lexicographic tradition, observes that the letter a "serveth many times to express the dative case".

Implicit grammatical relationships are most often represented by generalizing examples. By this term I mean the use of pronominal object to show types of complementation. The choice of pronominal object may also represent the semantic choices available (animate or inanimate):

Hollyband 1593:

Assaillir aulcun, to assaile a man.
Adroit, as c'est un home bien adroit à cecy, ou à cela: an handsome and fit man to this or that thing

Literary examples with nominal complements can serve the same function but in those instances it is less certain that that is the primary function.

Nowhere in the dictionaries do we find the clear description of verbal syntax that most modern dictionaries include: transitivity, intransitivity; ability to take a prepositional or subordinate-clause complement. The inclusion of generalizing examples is a recognition of the importance of this type of information, but the lack of systematic indications is symptomatic of the general weakness of the syntactic portions of the grammars. The notions of complement and governance were slowly developing (see Chevalier, 1968), but still far from widely received. This 'weakness' itself is related to the prevailing attitude toward the word and its part in 'construction'. The primacy of the former might itself be viewed as an artifact of the glossing tradition of Biblical exegesis. These explanations of the syntactic content need to be proven by the kind of evidence that only these databases can provide. In order to extract the data necessary for that task, entries should be tagged for type and manner of presentation of syntactic information.

1.5.4. Semantic Information

Here we need to look at the types of equivalency provided, the use of synonymy in the source as well as the target language, and the definition of terms in the source or target language.

In the source language the presence of a definition may represent the fact that the initial word-list was one based on another language. Thus, when Hollyband inserts the definition of the French headword, it is usually because Estienne gave this French word a French definition, after a Latin headword:

Hollyband 1593: Barbare & qui n'est point de nostre langage [...] (< Veron 1552: Barbarus [...] Barbare. Qui n'est poinct de nostre langue.)

Similar suspicions must be raised if the 'headword' is a relative clause or present participle used as a noun.

Another possibility, sometimes of similar origin, is the use of synonyms in the source language as the headword, or in the target language, where it may represent sociolinguistic variation.

Hollyband 1593: Gratin dei, or falouze, an hearbe which Harts use against venemous beasts.

Junius-Higgins 1585: Faliscus venter [...] A haggise: some call it a chitterling: some a hogs harslet.

In the target language, a simple equivalent is the ideal goal of the bilingual lexicographer, but it is an elusive goal. If it does not exist, or if the lexicographer fears that the simple equivalent might be unknown to the reader, then some explanation is in order:

Translational Equivalent:

Veron: Batrachus, batrachi, m.g. a frogge Une raine, Une grenouille.

Explanation:

Veron: Balnearius, ria, rium, pertaining or seruing to banes or stues, Appartenant ou seruant aux bains

Definition:

Hollyband 1580a: Un Adultere, an adulterer, one that committeth whoredome in mariage.

Sometimes the synonymy simply means that the single word in the source language represents several different concepts in the target language, as in the case of clavus mentioned earlier, or graisse here:

Hollyband 1593: Graisse, fat, grease: f.

Occasionally in the tri- or quadri-lingual texts both the source and one of the target languages include lengthy expositions, most often of the single Latin word.

Such phenomena help us to draw the links between successive generations of the same dictionary, but the selection of one or two examples can be misleading and this is where the computerized dictionary can be extremely useful. Only by comparing complete lists of entries, with sorting by a variety of possible components of the entry, can we understand these changes, these lexicographical choices. Then we can identify all the times Estienne's list provides a definition or explanation instead of a translational equivalent, and we can determine what choices subsequent lexicographers made: to omit the concept because it did not have a simple term in the language of the headword; to admit a latinate neologism to serve as headword; to place the explanatory phrase in the position of the headword, etc. These choices, viewed over an entire dictionary, not in the kind of small samples we are currently limited to, permit us to relate the dictionary to broader intellectual and cultural movements within the society. This is where the historiography of lexicography becomes an integral part of the history of ideas.

1.5.5. Language Variation

With the first commentaries on many of the modern languages came the first efforts to restrain the wild variation those languages exhibited. That variation might be dialectal, either geographic or sociolinguistic, or it might be diachronic. In either case the comments reflect efforts to purify the ancient languages and to 'defend and illustrate' the modern languages. Higgins openly states that he is out to clean up the medieval Latin vocabulary of Huloet, and Palsgrave prides himself on the efforts to bring French under the control of rules. The interest in linguistic uniformity goes beyond the limits of these avowed purposes, and beyond these authors.

1.5.5.1. Geographical variation

Palsgrave was just as likely to criticize English dialectal usage as French, condemning many words as "northen", particularly in the works of Lydgate.

I Carpe (Lydgat) Ie cacquette. prime. this is a farre northen verbe.

Palsgrave and many of the other authors remark on dialectal usage in French, most often commenting on Picard or Gascon forms, since these two areas were the most active trade partners with England. Picard also had a more active literary output than other dialectal literatures after the 13th century.

Baret (in etymology): A Gaill, or pryson. Carcer, eris, pen. corr. m.g. Ergastulum, li, n.g. Gaoyle. It seemeth to come of this picard woorde Gaole, which is derived of this latin woorde Caveola, being a diminutive of Cavea which in Frenche is called Cage d'oyseau [...]

Palsgrave: I Rent I teare a thyng a sonder/ Ie dessire, prime. and ie arrable, Romant. prime coniuga. and ie deschire, prime coniu. and ie despece. pri. coniu. He hath rent my gowne: Il a dessiré ma robbe, il a deschiré, is Pycarte/ and il a despecé ma robbe, as for arrabler, is nowe out of use in comen spetche.

Palsgrave (word choice): [interjections of lamentyng:] henny is used rather in the doutche lande, and where they speake Rommant and Wallon than in France.

Palsgrave (morphology): Ie clos, il clost, nous closons. but the Romant/ clouons, ie closys, iay closy, ie clorray, que ie close, que ie closisse, clorre [...]

Surprisingly, perhaps, given the importance of Normandy in English history, there are few references to the Norman dialect, and even fewer to the Anglo-French dialect that remained the language of the law at this time, and this in spite of the fact that many of the lexicographers had contacts or training in the legal community.

1.5.5.2. Sociolinguistic variation

Less frequent than dialectal information is commentary relating to class distinctions in the use of language. Palsgrave distinguishes between two different counting systems, one more prestigious (septante, octante, etc.), the other the language of "voulgar people" and "marchante men" (soixante-dix, quatre-vingts). Hollyband notes a few words "used mockingly" and Palsgrave includes many words and examples that shock us by their crudity, but neither specifies in those instances that these are terms limited to any class or segment of society. In the following example Palsgrave distinguishes not only between classes, but also between written and oral registers.

Palsgrave: But where as in comun speche they use to saye: je allons bien, je ferons bien, j'avons faict ung grant exploit, and suche lyke, joynyng the first person plurall of the verbe in to je, whiche is the first person singular, suche kinde of spekyng is used of none auctour approved.

Huloet-Higgins: Borowe of Peter to paye Paule, whiche is a vulgare speach, properly where as a man doth borowe of one to paye an other [...] Discouurir saint Pierre pour couurir saint Pol, emprunter argent a gros interest pour payer une debte qu'on doibt a moindre interest, prendre a perte de finance

Huloet-Higgins' proverbial example includes an explanation of the proverb for all the higher-class folk whose delicate ears had never heard such blasphemy, and an interesting feature is that the French explanation is more explicit than the English.

1.5.5.3. Diachronic variation

Chronological variation is all important for a language that according to Tory changed so quickly that pieces written fifty years before were incomprehensible. Palsgrave decided that Alain Chartier, whose work was almost a century old at the time Palsgrave was writing, was the limit of acceptable usage:

suche authours as write in ryme use to varie the termination of substantives very often, bicause of the more just hapyng of this ryme, especially the Romant of the Rose, in whose days the Romant tong was not come to suche certaynte as sith the frenche tong is, so that it were requisite to loke upon other authours that write in prose, or upon suche as have written sith Alain Chartiers time to knowe the right frenche wordes

Sometimes, as he notes in the case of 'turning' the disappearance of the word can be ascribed to the disappearance of the practice it was used to denote.

I turne as a man dothe in a daunce/ Ie me renvoyse. prime coniu and ie me vire. prime coniu. and ie me revire pri. coniu. This terme waxeth out of commen spetche bycause the maner of daunsynge is chaunged/ howe be it/ it is somtyme used.

The only problem was not archaisms, however; neologisms were equally unacceptable, and especially neologisms drawn from languages viewed as threatening the integrity of the author's native language. Thus Palsgrave attacks the borrowings Lydgate introduced, generally calqued on French usage, and Hollyband attacked Italianisms creeping into French.

Palsgrave: I Betraysshe (Lydgate) I go aboute the stretes of a towne or cytie/ Ie tracasse. prime coniu. This verbe is nat yet taken in comen use.

Hollyband 1593: Poltroniser, to play the slouthfull person: it is an Italian worde. [Others are bardache, bocon, goffe]

Palsgrave is the only one I have found to criticize French expressions for being too latinate, as when he rejects ie vous propine for 'I drink to you'. Hollyband's selection of Italian words shows that he subscribes to Henri Estienne's judgment that French must borrow Italian words for certain forms of vice and scoundrels because those vices and those scoundrels simply do not exist in France.

A final recognition of linguistic variation is the insertion of etymological commentary into the articles. This is exceptional, not regular, and can serve several purposes. For Palsgrave, the historical formation of the word is often used as a means of determining what correct usage should be, but his deference to age is sometimes tempered by practical considerations, as when he criticizes the Roman de la Rose for using the etymologically correct but practically confusing ung voylle for une voylle ('a sail'). Baret includes long etymologies that direct the attention beyond the French to the underlying Latin:

Attaynted seemeth to come of the french woorde Teinct, which is also derived of the latin Tinctus. Infectus. Colore vel humore imbutus. And of the old latin woorde Attinctus, came of the feyned french Atteinct, which we yet vse in englishe. There may also be some proovable reason that it commeth of this french woorde Esteincte, which is in latin Extinctus of Extinguo, to put out and so leaving for the s (as frenche men do pronounce) it is almost our woorde Atteynt.

Stein (1985: 283-4) notes (following Sledd, 1947 and Starnes, 1954) that in Baret derivations are not offered for words of Germanic origin. Their explanation is that most of the derivations included are taken from the Stephanus tradition. To this Stein adds that the primary interest was in synchronic contrast of two contemporary languages, and not the historical contrast within one language. These explanations have a certain validity, but do not explain all the cases, and present a logical inconsistency. If Baret is "a lexicographer who contrasted different languages and not different stages of one and the same language" (Stein, 1985: 283) why would he contrast different historical stages of French? Still, she is on the right track, but we need to look beyond the lexicographic to the broader historical context. The French words cited are almost always traced further back to the Latin. In this way the use of French words is not subject to the complaints of the anti-Normanists who were gaining sway in English intellectual circles (cf. John Cheke's attempt to eliminate non-Saxon words from his rendition of the Bible). Germanic words need no justification for inclusion; French words need to be traced back to Latin to gain acceptability. "Feyned French", which may be a reference to Law French or to earlier periods of Anglo-French, definitely lacks prestige.

Veron often notes historical change within Latin, consistent with the Humanist desire to purify that language. A further use is to show derivational morphological developments, as the creation of the noun bacchatio from the verb bacchor. Another use of such familial relationship is to explain non-obvious connections, as between a bird and bees:

Veron:

Balineum, balinei, n.g. & balineae, balinearum, foe. g. à veteribus dicebantur quas nunc balneas dicimus, a bayne, a hote house or stue, Bains, Estuues.
Bacchatio, tionis, foe. g. Verbale, à verbo bacchor, drunkenshyp, the deede of a drunken man, Ivrognerie, Faict d'ivrogne.
Apiastra, apiastrae, foem. gener. a byrde called a mod wale, she destroyeth and devourethe bees, Oiseau nommé Mezenge.

The interest in word families may have been part of the conservative movement on the orthographic questions. One of the consistent arguments for maintaining antiquated etymologizing spelling is that it helps the learner recognize the relationships between words of the same origin. Thus once again, lexicographic decisions are often based on broader issues, ones that can be ascertained with more certainty only through computerized analysis of full texts rather than impressionistic analysis of partial texts.

1.5.6. Examples

Another lexicographical choice for these dictionary writers involves the type and purpose of examples provided. Again, the first aspect to be addressed is the language of examples. Palsgrave mentions Chaucer twice and Lydgate several hundred times, but in none of these instances does he supply more than the base word. Lydgate is most often used to illustrate incorrect usage, either for dialectal words, archaisms or neologisms. For French words, on the other hand, Palsgrave supplies full sentence examples for most of the verbs, adverbs, prepositions, conjunctions and interjections. Literary pedigree is not enough for acceptance. Most examples from the Roman de la Rose are inserted to show outdated usage. Examples from Jean Lemaire des Belges often denote orthographic inconsistencies. Hollyband furnishes French examples from literary, everyday, and proverbial sources. The literary examples, from Marot, Rabelais, and Amadis, are often to help explicate a difficult passage, such as a pun:

Il pisse pour les Tres-passez, Rabelais, doubtfully spoken, for it may be taken as it is pronounced, he pisseth for the dead, alluding to the custome of the Papists, sprinkling the graves of the dead with holy water to clense their soules: but indeed it should be written, Il pisse pour les traicts passez, that is, he pisseth for, or bycause of the draughtes of Wyne or drinke which he hath swallowed downe.

The practical examples appear to be drawn from everyday life, and may tie in to his dialogues for learning French.

Hollyband: Si vous me voulez defrayer, j'yray avec vous: if you will beare my charges, I will goe with you.

Palsgrave: (I go darkelyng): Go fette me a candell I can nat go darkelyng in this house: Allez moy querir une chandelle car ie ne puis pas aller a taston ycy en ceste mayson

It could be that some of Baret's examples are taken from the dialogues that were used to teach Latin in the 16th century (see Massebieau, 1878):

Borowe. [...] A. Will you go with me into the towne?
B. I have no gowne. A. Go borowe one somwhere.

Palsgrave's non-literary examples are equally colloquial, sometimes even crude.

Proverbial examples hold a special place in the hearts of 16th-century lexicographers, for proverbs were thought to embody the 'genius' of the language, the naïve puissance as Meigret put it. Palsgrave, who notes that French is especially rich in such expressions, includes proverbs going in both directions, English-French and French-English:

English-French: A day for the fayre, whiche we use for an adage, meanyng that one cometh to late, whiche they express les secours d'Espaigne

French-English: I Nye as a horse dothe/ Ie hannys, hannyr, secunde coniu. Thou nyest for an others otes/ whiche we express by these wordes. Thou lokest after deed mens shoes: Tu te hannys pour lauoyne d'aultruy, it is an adage in the frenche tonge.

Hollyband also includes a number of French adages. In the Frenche Schoolemaister (1573) he had compiled a list of proverbial expressions in which the French and English matched, something he repeats in the Frenche Littelton (1576) to which he adds "Golden Sayings". All the grammarians including Latin provide proverbial examples from that language. These expressions not only show the 'genius' of the language, but also served as the backbone of the instructional program. Higgins and Nicholas Udall collected and published the "Flowers or Eloquent Phrases of the Latine Speache" from the comedies of Terence. All of the lexicographers were teachers of Latin, or French, or both, and the use of proverbial wisdom was a way of integrating moral with linguistic education.

Examples can illustrate syntax, morphology or orthography (either correct or incorrect), provide practical expressions for everyday life, reveal similarities and differences in the 'genius' of the languages discussed, and introduce the students to the pearls of literature. The choice of function and the choice of language are one more piece of the puzzle in determining what the lexicographers were doing in writing what they wrote, one more element that must be clearly tagged in a computerized database of such texts.

1.5.7. Encyclopedic Information

The final element I shall discuss is the nature of the encyclopedic information included in these dictionaries. Encyclopedic information can be proper names or references to specific events, or it can be explanations of the use of the object denoted by a word.

Event: Palsgrave: I Ryse as commens or subiectes do agaynst their prince whan they rebell: Ie mutine, iay mutiné, mutiner. prime coniu. I remember well ynough whan the comens of Cornewall dyd ryse: Il me souvient asses bien quant les communs de Cornouaille se mutinerent [Reference to uprising of 1497].
Hollyband: Le reveille-matin des Françoys, the booke shewing the falshood of the authors of the massaker or slaughter traitrously committed on the persons of the most noble & faithfull christians of Fraunce, Anno 1572

Place: Hollyband 1580a: La Bourse de Londres est couverte d'ardoises, the Royal Exchaunge is covered with slates.

People: Palsgrave: I Pronostycate I shewe thynges to come/ Ie pronostique. prime. I have sene the boke that dyd pronostycate the comyng of Luther twenty yere or he was borne: Iay ueu le liure qui pronostiqua laduenement de Luther uingt ans auant quil fut né.

This use of proper names can be culled largely from the index of words. More problematic is the thematic linking of entries. I provide here a small sample of the words that might be linked by common reference to the law:

Palsgrave: I Regyster I put a thyng in writynge in a booke of recorde/ Ie registre. prime coniu. My fathers wyll is regystred in the bysshops courte: Le testament de mon pere est registré or enregistre en la court de lesglise, or de leuesque.
I Reioyne as men do that answere to the lawe and make answere to the byll that is put up agaynst them/ Ie liticonteste. prime coniuga. He can nat forsake his iudge nowe for he hath reioyned: Il ne peult poynt appeller de son iudge maytenant car il a liticontesté.

Hollyband: Resusciteur de proces vuïdez, a raiser up of matters already decided and iudged, to renew an olde sute: m.

The range of topics here is limitless, but just as an example, such webs of linked words, on topics such as religion (we see the Protestant attitude towards the Catholic Church in Hollyband's explication of the pun in Rabelais), sexuality, etc. are sure to add to our understanding both of the intentions of our lexicographers, and of the major themes of 16th-century intellectual life. As a starting point for the semantic classification of the vocabulary included we might use something like Matoré (1988) as well as the list of subject areas covered that the lexicographers themselves have provided either on the title page or in the preface. The ability to analyze these dictionaries as full texts provides a test for Matoré's approach and conclusions.

2. Database Search Desiderata

In this section I shall point out some of the common and special needs scholars from different disciplines encounter as they use the computer to analyze these texts. The specific subject areas that I shall treat are: History of Linguistics/Lexicography; Diachronic Linguistics within a specific language; Literary History and Criticism; History of Culture/Ideas. This list is not exhaustive, either of the disciplines that might find things of interest in these databases, or of the approaches possible. It is rather a starting point, and additions, subtractions, and corrections are welcome and necessary.

We can assume certain common needs of all researchers, and these needs are met to varying degrees by most of the commercial or scholarly full-text database organizers currently available. These include the preparation of an index of all words in the text, the ability to search by word or phrase (with 'wild-card' characters or words), by proximity (how many times does word X occur within 5 words of word Y), by matching strings (what is the longest pair of matching text starting with word Z), and finally the ability to combine searches (e.g., matching the set of entries in which Italian is mentioned with the set of entries in which architecture is mentioned). Beyond this base, scholars in different fields will focus on different aspects, even different parts of the text, and will seek to relate the text to different external documents.

The primary shared need of all scholars is to have an accurate rendering of the text, as it appears in the original. This means marking changes in font, page, column and line breaks, marginalia, ornamentation, and, in the case of manuscript documents, variants and changes in scribal hand. An alternative to this type of marking is hypertext linking to graphic images of the documents, by which these features could be recovered without having to insert a large number of tags, but linking to a graphic image does not in itself allow us to isolate all instances of a particular format.

2.1. Historians of Linguistics/Lexicography

The historian of linguistics will analyze these texts to see how changes in the dictionaries reveal changes in linguistic theory and practice. These 16th-century dictionaries are particularly important for they are often the first lengthy commentaries on the vernacular languages in question, and they occur at a time when the study of the classical languages was undergoing a significant change. The first key to the lexicographers' attitudes towards these languages and the inherited linguistic practice comes in the introductions. The special significance of this for the purposes of computerized databases is that those factors which seem to have been given prominence on the title page and in the introduction (e.g., treatment of dialect, source of authority, specific subject areas treated, potential uses for the dictionary) must be tagged as such when they are followed up on in the text. Take, for example, the title of Huloet-Higgins 1572:

Huloet's Dictionarie, corrected and amendet and set in order and enlarged with many names of men, townes, beastes, foules, fishes, trees, shrubbes, herbes, fruites, places, instrumentes, etc. In eche place fit phrases gathered out of the best Latin authors. Also the French thereunto annexed, by which you may finde the Latin or Frenche of anye Englishe woorde you will.

Higgins specifies, first of all, that he has corrected Huloet's dictionary (1552); therefore we need to be able to distinguish those entries which appear in both dictionaries but do not match (and then subdivide them according to how they do not match). Secondly the work has been "set in order": in what ways has the order been changed (and ultimately why)? Next Higgins lists subject areas that have been added. Words that fall within these categories need to be marked, and the sets of entries so marked compared to the set of such entries in the 1552 dictionary. Higgins claims also to have expanded the citations using the "best Latin authors". Who is cited in Higgins that is not cited in Huloet? Do the citations serve the same purpose(s)? Higgins states that the French has been annexed to the English and Latin, inviting tagging of the full extent of French words included (are any entries lacking a French equivalent?) and the nature and source of the French equivalents proposed. Thus the introductory materials, including the title page, are guides to at least part of what needs to be tagged in each dictionary. For the purposes of comparison of dictionaries, all dictionaries need to be tagged for all the criteria included in any one. For instance, even though Palsgrave doesn't mention "beastes, foules, fishes", etc., in particular, to compare his dictionary to Huloet-Higgins we need to be able to compare his list of words in those categories to Higgins'. This may sound like an infinite list, and it may turn out to be so; however, the minimum tagging of this sort, drawn from a master composite list of such features specifically mentioned by the lexicographers is not so long (perhaps a dozen semantic fields). Even if one adds the categories specified by Matoré, there are only 27 main subject headings (107 subheadings).

The author's stated intentions are only one of the aspects that must be noted. In addition the historian of linguistics will need to recover the organizational scheme both of the whole and of the individual entries and sub-entries. The organization of the entries can be quite diverse; for this type of study we need to be able to separate, for example, dictionary entries in which related words are sub-entries under the (perceived) root word; dictionary entries in which alphabetic ordering has been violated; dictionary entries in which morphological, syntactic, or semantic information is provided only for the target language, only for the source language, or for both; dictionary entries in which different meanings of the same head-word are included in a single entry, and those in which they are given separate entries. These are but a few examples of the types of distinctions that the tagging of the database should be able to capture.

In some instances other tags or other search methods could be used to find the positive tokens, particularly through the comparison of intersecting and non-intersecting sets. Suppose a historian of linguistics were interested in the history of historical linguistics. Subcategories within the set of all metalinguistic vocabulary could be used to determine the distribution of secondary information -- which French words are cited as being derived from Greek, for example (a question with political overtones; see Kibbee, 1991b). Here one could search for the set of entries in which key phrases for the introduction of etymological information is used (e.g., "come of" in Baret's derivation of attaynted cited above), and then for the entries in which those phrases occur in proximity with either the word Greek or a Greek etymon. Finding this list, and comparing it with other dictionaries lists of this sort, gives us a sense of what the author might have tagged as of Greek origin, according to the prevailing ideas of the day (exemplified by Henri Estienne's Conformité du langage françois avec le grec (1565), Perion's Dialogorum de linguæ Gallicæ origine eiusque cum Græca cognatione libri IV (1555), and Trippault's Celt-Hellenisme (1580)). The tagging of the negative tokens is more complex, and requires greater expertise. The first group of this sort, those words that have been attributed Greek origin by other grammarians and lexicographers of the day is fairly straightforward (assuming their works are available in machine-readable form). The second group, words for which modern etymological dictionaries have furnished a Greek etymon, would be greatly simplified by the creation of electronic versions of the great etymological dictionaries (such as the FEW). The negative tokens are just as crucial to determining the changes in linguistic theory and practice as the positive. Until they have been identified we cannot recover the intentions of the author.

As these examples illustrate, ultimately we need to link the primary documents we are interested in to a variety of external documents that will help us determine the meaning of the internal information. The external documents important for the historian of linguistics include (but are not limited to) other dictionaries and grammars, source materials (literary and otherwise), biographical information about the authors (e.g., Dictionary of National Biography for England, Biographie universelle for France), historical documents from the period (government records, etc.), much like the Renaissance knowledge database being developed by Ian Lancashire (1992).

2.2. Historians of Language

The historians of linguistics are primarily interested in the types of information provided, the historians of an individual language need to know and be able to sort out the detailed information. An historian of linguistics wants to gather from the database the type and distribution of morphological information included in verb entries, the historian of the language wants a list of all variant verb forms for each verb. The structure of presentation is less important than the raw data. One studying the history of French morphology might want to isolate all instances of variation in conjugational pattern (e.g., switches between -i- forms of the passé simple, and -u- forms). For the history of syntax, one might wish to identify examples which show certain types of complementation. For the history of vocabulary one might tag first occurrences of certain definitions or equivalencies, as well as the author's own indications of usage ('vulgar', 'archaic', etc.)

These semantic features can be isolated in a largely mechanical way, using the search for metalinguistic vocabulary described above. In some cases simple string searches can simplify the identification of morphological characteristics. The variation in conjugational pattern can be extracted from searches for strings -is, -ist, -ismes, -istes, -irent and equivalent forms in -u-. However, the analysis of syntactic constructions cannot be so simplified, and requires tagging by experts not just for major constructions (transitivity, types of complementation), but also for features within the constructions (e.g., animate/inanimate). This is time-consuming, and although useful, perhaps best done for basic research through the analysis of other types of texts (continuous prose). However, the classification of the types of information provided in the dictionaries might provide a counter-weight to the evidence of those other texts. The selection of which information the lexicographers chose to present, compared to the variety of structures extant in the language at the time could be extremely useful to the historian of linguistics.

As in the case of historians of linguistics, historians of individual languages have specific external sources which, linked to these databases, would be useful for their research. In particular, the historian of a language needs to be able to compare the data extracted from these lexicographic sources to standard analyses of such information found in such reference works as historical dictionaries and modern dictionaries constructed on historical principles (FEW, OED), standard histories of the languages involved (Brunot, Fouché, Pope, etc. for French), period dictionaries (Godefroy, Huguet, Middle English Dictionary, etc.), and histories of orthography (Beaulieux, Catach). Of these, only the OED, so far as I know, is available in machine-readable form.

2.3. Literary Scholars

Literary scholars will be interested in the source and use of literary examples, the dedicatory or laudatory verses which often precede the works, the definitions of literary genres and practices. The literary sources, and the examples taken from them, are noteworthy not only for which ones are included, but for which ones might have been included but were not. We can verify, with the Short-Title Catalogue, what translations of French works would have been available to Palsgrave by the time he was composing Book III (the dictionary) of Lesclarcissement (1523-30). Which ones did he use, and how did he use them? What manuscript sources were used? Palsgrave tells us that Gilles Du Wes introduced him to the Roman de la Rose, showing him an old manuscript. Palsgrave's examples taken from that source might help us determine which of the manuscripts of that work was available to him. Knowing that would allow us to judge Palsgrave's use of the potential examples. Once we have identified Palsgrave's sources for the examples he provides for alternative translations for downe,

Downe: Bas. Embas. Ius. Aual.
[...] Embas, as/ Mercure iectant les yeulx embas, & Priam iecta son sceptre embas
Ius, as Pallas ne voulut mettre ius sa chemyse

we can judge (1) which works he was aware of among the list of authors he mentions; (2) which examples of these words he selected and which he chose not to use; (3) (possibly) what translation of the works he found the equivalent phrases in. In the case of examples drawn from manuscript sources, the literary examples cited in these dictionaries might provide interesting variants, or point to manuscript versions now lost.

The dictionaries also provide definitions of and commentary on literary practices and genres of the day. While Hollyband's definition for théatre seems unexceptional:

Théatre, édifice public ou le peuple s'assembloit pour voir les jeux, A Theatre or common place of shewe

it might be worth noting that the French definition is in the past tense. This gains further importance when one considers the example he provides for apprentissage:

Le theatre des jouëurs est un apprentissage de toute impudicité, lubricié, paillardise, ruse, finesse, meschanceté: the theatre or stage of players is a lesson or learning of all lecherie, hooredome, guyle, craft, wickednes.

The preparation of Hollyband's dictionary coincided with Marlowe's most productive dramaturgical period, and this description of the theatre may have been inspired by Marlowe's numerous brushes with the law (culminating in arrest on a charge of atheism one month after the publication of Hollyband's dictionary). Protestant attitudes towards theatrical productions are manifest both by this positive evidence, and by the negative evidence that the words jeu, scène, pièce (de théâtre), etc. are all absent.

To use the dictionaries in this way, literary scholars need not only for the dictionaries to be tagged so that examples are clearly marked as such, but they need to relate the examples to outside information, starting with the full texts of works that can be identified as sources. Another useful connection is to period dictionaries (Huguet, Middle English Dictionary), and to collections of usage from the period (such as the database of the Trésor de la langue française). In this way literary scholars can use the dictionaries to determine how the dictionaries determined meaning and equivalence, how the dictionaries came to be authorities, limiting variation in usage -- in short, how the dictionaries used the authors and how the authors used the dictionaries.

2.4. Historians of Culture

More general historians may be interested in these dictionaries for particular or general reasons. The particular information is the type of encyclopedic information documented through the vocabulary itself or through the examples. This has been exploited in a light-hearted and impressionistic way in articles like Perrin (1972), more systematically and seriously in Matoré (1988). Following Matoré, the historian can choose to look at the vocabulary of the 16th century in a global fashion, uniting all the dictionaries, and adding to them the studies of usage in authors of the period. An alternative approach is to consider each dictionary in its own right, as a snapshot of the author's view of the 16th century. Matoré complains of the "absence de véritables dictionnaires" (1988: 13) either in the 16th century or of 16th-century language. I think this emphasis on the gaps and mistakes (lacunes, erreurs) of the dictionaries in question misses the point; modern dictionaries which try to include all vocabulary in the collective word-stock of native speakers provide just as false a view of modern vocabulary, masking the gaps in each individual speaker's personal lexicon. The OED and the TLF describe a vocabulary far beyond the range of any single native speaker; how is one to know or guess what portions of these collections of vocabulary are in active or passive command of any one speaker? Perhaps one reason Matoré missed that point is that the type of analysis of the dictionary as a text that I have emphasized throughout this article is simply impossible without a computerized database of the material. The single-author dictionary, however indebted it may be to previous lexicographers, captures a singular yet global vision of the society whose words it describes. Displaying what these lexicographers captured to the modern scholarly public requires tagging semantic fields and other encyclopedic information as completely as possible. (There is no doubt that this will demand ongoing additions and corrections by later scholars.) The database that will include a number of these dictionaries must be so constructed that compilation of lists of words within a given semantic field can easily be accomplished, and also comparison of such lists across several dictionaries. The compilation within a single author could supply a particular view of society at a given time. That Palsgrave, an ordained priest, would provide the following equivalent for prestresse, and the following example to contrast the alternate orthography of avec might help explain the climate which favoured the spread of Reform in England:

Preestes concubyne, prestresse. [cf. Hollyband: Prestresse, a she priest]

Avec, or avecques. bytwene whiche I fynde no difference nother in signifycacion nor in use. As/ Voulez vous coucher avec moy, or avecques moy.

The comparison of all the terms related to religion found in Palsgrave and in Hollyband would demonstrate the evolution of religious thought in 16th century England.

As in the case of the previous disciplines discussed, these studies that relate to the history of society itself would profit from links to a broad range of external documents. Those that come readily to mind are modern encyclopedias (some already available on CD-ROMs), government documents and non-literary publications from the period, the biographical sources mentioned earlier, etc. Whether one looks at the dictionary as a complete portrait of one man's vision of 16th-century society, or as an incomplete picture of the full range of the society, the database comprising a single dictionary or a series of dictionaries will be a powerful tool towards understanding the period.

3. Conclusions

I have outlined a few basic ways in which scholars from different disciplines might want to approach these texts. These lists are not exhaustive; they are a point of departure. The types of information we need to be able to retrieve from the dictionaries require certain types of access software: word index with search and retrieval; search and retrieval of tagged elements; hypertext linking. The first might be taken care of more or less automatically. The potential for automatic sorting of the second type depends on the complexity of the tags, although we have suggested some shortcuts. The third is necessarily a collaborative project, and the technical and intellectual problems of that collaboration are not trivial. The results however should remove dictionary research from the category of secondary drudgery to primary intellectual investigation.