Introduction to DARE Volumes in Print

Excerpt from the Dictionary of American Regional English: Volume I (1985)

Cassidy and Hall with DARE Volume I

Frederic G. Cassidy and Joan Houston Hall with a copy of Volume I of DARE. Photo courtesy of Jeffrey Steele, 1985.


A scholarly project that has required many years of work, involving hundreds of people and considerable expense, cannot be fully understood without taking account of its inception, its aims, its progress. The history of the Dictionary of American Regional English (DARE) goes back almost a hundred years, to the founding in 1889 of its sponsoring body, the American Dialect Society (ADS). It is no coincidence that in the same year Joseph Wright took the first step toward editing his English Dialect Dictionary (EDD); that fact was well known to the eminent American philologists who founded ADS, whose purpose evidently was to do, for the English language in America, what the English Dialect Society had done in gathering materials for Wright and sponsoring his work.

The list of names, though a century old, is still a roll of honor, deserving remembrance. Among the most eminent members were Edward S. Sheldon, Charles H. Grandgent, George L. Kittredge, Francis J. Child, James B. Greenough, John M. Manly, Benjamin I. Wheeler, and William D. Whitney. They stated, as the purpose of the Society, “the investigation of the English Dialects of America with regard to pronunciation, grammar, vocabulary, phraseology, and geographical distribution.” The first issue of the Society’s publication, Dialect Notes (DN), appeared in 1890. It continued publication for forty-nine years and produced six volumes, chiefly of word lists but also including other apposite studies.

Though production of a dictionary was evidently foreseen from the first, collecting was left to individual initiative; there was never a real campaign or an adequately planned effort to cover the country. Word lists, besides, often lacked the kind of precise information necessary for compiling a dictionary. When the Society was seventeen years old, William E. Mead made this point in no uncertain terms:

The most ardent admirer of the achievements of the Society must admit that all the investigation that has been done, however thorough for certain districts, is fragmentary in the extreme...To one who has reviewed the whole situation it seems clear that the time has come when we should definitely abandon the drifting policy which we have followed and set out on a systematic investigation. If our chief aim is to publish detached studies of a district here and there as chance may offer them, we shall doubtless accomplish something of value...but if we cherish the hope that by such means we shall, within a reasonable time, succeed in preparing an adequate dialect map of our vast country and in bringing together a sufficient amount of trustworthy material for an American Dialect Dictionary worthy to stand beside the English Dialect Dictionary, we are optimistic indeed. (DN 3, 168–169.)

Mead estimated that “the work of the Dialect Society must be multiplied twenty-fold before a dialect dictionary can be safely undertaken.” Pointing to the rich fields, especially the South, at that time but scantily studied, he outlined a plan of organized collecting, state by state. He deplored the daily disappearance of material as speakers of dialect died. And he insisted that this kind of investigation could not be made without money—more money than the Society was taking in from its membership. Hoping for the support of some wealthy individual, he concluded: “If such assistance can be obtained, we may hope to accomplish a work of lasting importance.” That was in 1906.

The chief moving force in the following years was Percy W. Long, who in 1917 began the formation of a master file. Yet only two years later he warned that, despite some progress made, “the Society’s chief enterprise may be said to be in a critical position” (DN 5, 74–75). It was the period of the First World War; the making of the file had been delayed; Mead’s hoped-for Maecenas had not been found.

A new step was taken in 1929 when the office of Editor of the Dictionary was created and Long was elected to fill it. In his report, however, the new Editor remarked, “there is no immediate prospect of beginning work on the Dictionary itself.” The lists of words and expressions already printed in Dialect Notes, he said,

now total over 30,000 entries...No doubt it seems to many that the Society’s dictionary should without much further delay begin to assume some form practicable of publication. An examination of the data now available shows that a dictionary of a kind could be prepared from them. It would probably include the greater number of existent dialect words and locutions. It would probably indicate with some approach to accuracy the sections in which they primarily appear. On the other hand, it could not compare at all favorably with the English Dialect Dictionary.

Even to equal the EDD it would still be necessary to gather dialect materials from literature, to have many county glossaries, and to have local correspondents to whom questions could be submitted during the process of editing. That was in 1930. The time was not yet ripe for the kind of effort necessary to produce the kind of dictionary that had been envisioned.

Meanwhile, the Society took part in two other projects of a related sort: the Linguistic Atlas of the United States and Canada, whose first publication was the Linguistic Atlas of New England;1 and the Dictionary of American English.2 Dialect Notes ceased publication in 1939 but was succeeded in 1944 by the Publication of the American Dialect Society (PADS), which has continued to print word lists and special studies. For a time, the hope of producing the dictionary was greatly weakened. During her presidency, Louise Pound revived it to the extent of personally encouraging Harold Wentworth to produce his American Dialect Dictionary, a small compilation based chiefly on Dialect Notes and his own extensive reading.3 But no field collecting was undertaken, large parts of the subject area remained unexamined, and the resulting volume was never accepted as the Society’s intended dictionary. The dictionary was not forgotten, however: in the late 1940s two meetings were held especially to plan it. At last, the present writer, asked to say how he would go about it, replied in “The ADS Dictionary—How Soon?” (PADS 39) and in 1962 found himself appointed Editor and encouraged to get on with the work.

<< back to top

Preparatory Steps

Studies already made were ready to be used. First, the new Editor had done fieldwork for the Linguistic Atlas, completing fifty workbooks in Wisconsin. This work had proved clearly that numerous local differences in Wisconsin speech could be correlated with settlement history and other social factors. Second was a statewide study, the Wisconsin English Language Survey (WELS), in which, by means of a mailed questionnaire, fifty Wisconsin natives in twenty-five chosen communities had returned a large number of variant responses. This questionnaire ultimately became the basis of that used by DARE. The purpose of WELS was, within the limits of practicability, to gather as much of the local lexicon as possible, in all its variations. This could not be done efficiently by random questioning. For WELS, therefore, the items printed up to that time in the DN and PADS word lists—approximately forty thousand of them—were analyzed and sorted by semantic categories. The questionnaire was then constructed of questions covering those categories which had proved to be the most fruitful in regional variation.4 Collecting could thus be undertaken with some assurance of maximal results. As noted, WELS became the pilot study on which the DARE Questionnaire was based.

The Editor had laid down two essential conditions for production of the Dictionary: full-time direction and full-scale financing. Fortunately for the latter, the U.S. Office of Education, just at this time, was supporting a few projects of research in the humanities. The Dictionary received support for five years, during which field collecting over the entire United States could be accomplished. The hope was that in another five years the collections could be edited and the Dictionary produced. But the success of the first step was so resounding—such a huge and valuable corpus was collected—that the years of editing stretched from five to ten and beyond.

The DARE Questionnaire

In the summer of 1964 the Chief Editor and Audrey R. Duckert worked out the essentials of the project and gave it its name.5 The acronym was taken to be a hopeful augury. For the essential nationwide campaign of data collecting, the primary tool required was a series of questions to be asked uniformly in all areas so that results would be comparable. The WELS questions were therefore completely revised, made over from their postal form for use in direct oral interviews, and put into a fixed phrasing which the field-workers were instructed not to alter.

Intended for use in personal interviews, the DARE Questionnaire (QR) begins with the neutral subject of time in order to allay possible suspicions of some hidden purpose on the part of the investigator. Next come weather and topography, equally neutral and safely concrete. Houses, furniture, and household utensils follow, with dishes, foods, vegetables, and fruits. And so the questions continue to more abstract topics: honesty and dishonesty, beliefs, emotions, relationships among people, manner of action or being, and so on—41 categories in all with a total of 1847 questions. In all, 1002 QRs were completed in as many communities.

In an attempt to avoid prompting specific replies, questions were phrased without using words that might possibly be given as answers. For example, the word rise is not used in asking about the sun, though sun itself is unavoidable: “What do you call the time early in the day before the sun comes into sight?” would not prompt sunrise rather than sunup, dawn, or some other word. (By far the most frequent response was dawn. The language of the questions was kept as simple and common as possible, in order to elicit an immediate, spontaneous response. Secondary and other responses, though not sought, were carefully recorded when given. The immediate, spontaneous response, however, is as close to normal local usage as an investigator can hope to come. It avoids inhibitions and second thoughts on the part of the informant; he or she does not have time to change the answer under the influence of educational or other social pressures. That the informants are aware of such pressures is revealed by such secondary responses as “I don’t know if that’s correct, but it’s what we say here” or “I shouldn’t have said snuck; sneaked is the right word.” Fieldworkers were instructed to record all such self-critical remarks; they are of great sociolinguistic value, especially when used alongside the informants’ biographies.

Most of the questions seek to establish the regional or local name for a single object or idea. For example, qu. R2 briefly describes a dragonfly and asks for its name. To this question 79 different replies were given, among them snake feeder (chiefly N Midl, also S Midl), snake doctor (chiefly Midl, Sth), mosquito hawk (chiefly Sth), spindle (coastal NJ), ear-cutter (NH, WI). Open questions were necessary when a whole class of objects had to be referred to: “What different kinds of oak trees grow around here?” (qu. T10). Informants might name as many as 20, or none at all, but at least the best-known kinds of oak in each region would usually be mentioned, including a number of local folk names, such as pin, post, Spanish, chinquapin, overcup, shim, chair bark oak. More than 130 names for oaks were proffered.6

Because the phrasing was so important, the questions were printed in a preliminary form and tried out in the first 75 QRs. In the final revision, some low-yield questions were dropped, a small number of new ones added, some troublesome ones rephrased. The QR was then printed and used without further change in the remaining 927 communities. As noted, the yield was enormous; the data gathered on living American speech are unequaled elsewhere. Nor does DARE exhaust them as a linguistic resource: they contain a great deal of unexploited data on syntax, for example.

The QR is not perfect, however. As Jules Gilliéron noted after making the Atlas Linguistique de la France, a questionnaire, to be appreciably better, would have to be made after the fieldwork had been done.7 In the nature of questionnaires, the question, if properly phrased and understood, and the answer, if responsibly given, should ideally produce a reversed definition. For example: “What do you call a container for coal to use in a stove?” Responses: coal bucket, hod, pail, scuttle, and so on. Reversed, this becomes a definition: scuttle, a container for coal to use in a stove. The method works relatively well for simple material objects like coal scuttles, less well for abstract things or emotional matters.

Questions intended to elicit one part of speech, or one form of a verb, were often answered in another: qu. GG34a, “To feel depressed or in a gloomy mood: He has the _____ today,” aiming at a noun, got the reply “feels blue” or “in low cotton.” To qu. X6, “If a person’s lower jaw sticks out prominently, you say he’s _____,” one informant replied “box chin”; the question, intended to elicit an adjective, produced a noun. This kind of switched response served as a check in making the questions: they had to be simple and idiomatic to keep from misleading the respondent, whose reply, unless hesitant or obviously uncertain, was a sound guide to his or her normal syntax or idiom.

The section of the questionnaire covering wildflowers proved unsatisfactory: it is difficult to describe a wildflower briefly in nontechnical terms with assurance that the informant will visualize the intended one. To compensate, a special questionnaire was devised, keyed to color photographs.8 Each picture was shown to the informant and the response recorded in a special list. Since this greatly increased the fieldworker’s burden, it could not be carried out in more than a few communities, but it did furnish a substantial and more accurate body of regional wildflower names.

<< back to top


It was decided that one thousand communities could be investigated in the five years for which support was assured. (This is a somewhat looser mesh than that of the Linguistic Atlases, but in areas where they can be compared, findings have corresponded well.)9 Communities were chosen in each state, the number proportional to the population and taking settlement history into account. (The planners of DARE were fully aware that linguistic variation does not go by states. Population statistics and settlement history are usually published by state, however.) The aim was to choose relatively stable communities, distributed according to the states’ composition, and communities of various types, so that the aggregate would reflect the makeup of each state’s population. Five community types were established:

  1. Urban: a specific section within a metropolitan area, with a population usually in the thousands within a total of a million or more.
  2. Large city: a section within a large nonmetropolitan city with a total population in the hundreds of thousands.
  3. Small city: a city independent of any metropolitan attachment, with a population in the tens of thousands.
  4. Village: an independent town closely attached to its surrounding rural area, with a population in the thousands.
  5. Rural: an area with unconcentrated population—farm dwellings, small crossroads settlements—and a population in the hundreds.

DARE recognized as a “community” any group of people living fairly close to each other and sharing the same commercial facilities, social organizations, and the like. Even within metropolitan areas such communities, or subcommunities, exist with a sense of focus based on ethnic, religious, and other characteristics. Contrariwise, quite small independent rural communities, though close together, may keep themselves apart on similar grounds. (For details about the DARE communities see the List of Informants.)

<< back to top


After a short period of preparation, data gatherers were sent to the chosen communities to find informants and get them to answer the QRs. These fieldworkers were chiefly graduate students trained in English language and linguistics and able to handle phonetic transcription; a few undergraduates and faculty members were also among them. The Chief Editor used the QR and interview method himself, comparing them with Atlas practices and seeing what details needed revision.

It was the task of the fieldworker, sent to the designated community, to find an informant who would answer the questions in the QR. It usually required more than one informant to complete the QR, but since each was given a personal identifying number and every response is coded, responses cannot be misattributed. For each informant a page of biographical information was made, including name, address, social factors (sex, race, age, education), amount of travel, chief occupations, associations, family background on both sides, a brief description of the community, and a description of the informant’s speech and attitudes toward language. These facts have greatly helped the editors to evaluate the data. It was also the responsibility of the fieldworker, in choosing informants, to attempt to balance social factors overall for his or her geographic area. Fieldworkers were closely monitored so that such matters could be controlled, and each fieldworker’s first QR was checked critically before the worker was permitted to continue. The total number of fieldworkers was 80: 51 men and 29 women.

One question—a minor one as it proved—had to do with Black informants and fieldworkers. Would a Black fieldworker establish a closer relationship with a Black informant than a White fieldworker could? To test the possibility, Black informants were interviewed by both White and Black fieldworkers, but no significant difference in the results was detected.

The DARE interview method was based on that of the Linguistic Atlas but with some innovations. It served remarkably well. Fieldworkers were instructed to avoid suggesting responses. When, rarely, they had to prompt an answer, they were to mark it sugg for suggested. If they heard something of linguistic interest in an informant’s free speech, they were to mark it con for conversational. If the informant said “We don’t have that here,” the fieldworker wrote NH for not here; if there was no response, he wrote NR. If the fieldworker had doubts about the response, he marked it with a question mark. If he was quite certain of something that others might doubt, he double-underlined it or wrote [sic]. Informants almost always took the procedure seriously, but a few were occasionally tempted to invent or to give humorous replies. Sometimes an informant misunderstood and answered at cross purposes. Fieldworkers were encouraged to make notes of any kind beyond the limits of the questions; their marginalia, quoted as FW Addit (fieldworker’s additional information) have often proved quite valuable.

The fieldworker was also required to make a tape recording of each of his or her chief informants from each community speaking freely for twenty minutes or more—preferably on a familiar topic so that the speech would be relaxed and normal. To check on more formal pronunciation, the informant was also asked to read the children’s tale “Arthur the Rat” (famous among phoneticians), in DARE’s revised version.10 These two types of recordings permitted comparisons, often highly instructive, between an informant’s speaking and reading styles. Tapes were made not only of the regular informants who answered the QR but also, as opportunity offered, by a number of auxiliary informants. In all, 1843 audiotapes were made; they constitute a unique record of American pronunciation, drawn from many levels of life and all fifty states. The tapes were studied over a two-year period not only for regional pronunciation but for other features, especially lexical ones that might go into the DARE files.

<< back to top


As with communities, the attempt was made to choose informants so that the aggregate would reflect the native population of the United States in its diversity. To qualify as a DARE informant, a man or woman had to have been born in the community represented or very close by, and could not have traveled or stayed away long enough for his or her speech to be affected. If the family had been in the community for generations, so much the better. Informants with foreign-language backgrounds had to use English as their home language, whatever else they might be able to speak in the community. Biographical information was compiled for every informant who completed at least one section of a QR, but not always for auxiliary informants who happened to be present during an interview and furnished an occasional word. The choice of informants was generally balanced with an eye to the five social factors already mentioned—community type, sex, race, age, and education—but with a deliberate weighting toward older people. Folk language is traditional, and older people remember many things that young ones have never heard of. DARE’s 2777 informants included 1368 men and 1409 women, ranging in age from about 18 to over 90. Old informants (60 years and over) made up 66 percent of the total, middle-aged (40–59 years) 24 percent, young informants 10 percent. The levels of education were as follows: less than fifth grade, 3 percent; at least fifth grade, 24 percent; at least two years of high school, 41 percent; at least two years of college, 31 percent; unknown, 1 percent. The racial distribution was Whites 92.7 percent, Blacks 6.7 percent, American Indians .3 percent, Orientals .3 percent. In addition, 244 informants made tapes but did not answer the QR, and 21 answered the wildflower QR only. These people are listed separately at the end of the List of Informants.

It must be understood that no attempt was made to prorate every one of these factors state by state; to have attempted to use such precise criteria would have made the task virtually impossible. Further, neither the choice of communities nor that of informants was randomized; on the contrary, the intention was to maximize the collection of materials by going to the places and people most likely to furnish the largest amount of appropriate data.

<< back to top

Processing the Data

By 1970 the 1002 QRs had been completed, but the sheer quantity of material gathered had become a problem. If every question had elicited one response each time it was asked, there would have been 1,850,694 responses—but multiple answers brought the total closer to 2,500,000. The planners of DARE had realized from the first that the mass of data would be large and that computer processing must be considered. In 1965, however, though numerical computing was well advanced, the handling of alphabetic material was only beginning. Nevertheless, the collections were coded for sorting, filing, and other forms of processing, and fortunately, as the collection grew, computing methods also advanced. DARE was ultimately able to save much time because of this, and to develop its unique map-making capacity.

As the completed QRs began to come back from fieldworkers, a group of graduate students—the modern counterparts of Dr. Johnson’s amanuenses—were employed to prepare them for computer input. Every page of every QR had to be numbered and stamped with the informant’s code, which consists of the two-letter state abbreviation plus the individual’s designated number. Thus, for example, OH47 is informant 47 from Ohio; VA3 is informant 3 from Virginia. Also marked were the sections of the QR that each informant had answered. Assuming no undiscovered typographical errors, every response can thus be traced back to the QR in which the fieldworker first recorded it. Data from the QRs were then fed into the computer, and ultimately the entire collection was printed out, question by question, with the responses listed alphabetically and the codes given by state and numerical sequence. The editors then condensed this mass of printout, removing anomalies,11 and it became the Data Summary, which will appear in the final volume. It still lists the QR questions in code sequence but now gives the responses in descending order of frequency.12

While the QRs were being processed, the other part of the DARE corpus was being gathered and put in order to form the Main File. This contains both oral and written, published and unpublished materials. Chief among published sources were DN; PADS; American Speech; the Journal of American Folklore and other folklore journals; the Federal Writers’ Project state guides series; more than five hundred books of regional literature; and selected newspapers from every state, especially county and small-town newspapers. Under a grant from the American Council of Learned Societies, a special program was carried out in which 114 diaries were excerpted. These sources are generally accessible in good libraries.

In addition, DARE has used a large body of materials that are not generally accessible. Notable are the personal collections of Gordon Wilson and Joseph Hall, made over periods of about 35 years each and generously given outright to the project. They contain records of local speech respectively from the areas of Mammoth Cave in Kentucky and the Great Smoky Mountains in Tennessee and North Carolina, gathered directly from lifelong residents before the communities were dispersed and the areas made into national parks—communities therefore now forever gone. Other personal collections given to DARE include those of Atcheson L. Hench, Kelsie B. Harder, and Russell Tabbert.13

Special mention should be made of materials gathered for but not used in the Linguistic Atlas of New England (LANE). Thanks to Raven I. McDavid, Jr., the original LANE workbooks were furnished to DARE. Similarly, the aluminum disks of New England (and other) speech recorded by Miles L. Hanley and Guy S. Lowman, Jr., in 1932–1934, among the earliest of such recordings, were transferred to audiotapes and the lexical items added to the DARE Main File.14 (In 1984 these disks were archived in the Library of Congress.)

DARE also has several special files that are not otherwise accessible:

  1. The WELS file, covering 50 Wisconsin speakers, with audiotape recordings of their speech.
  2. The “Northup File” of natural history, with the folk names of plants and animals cross-referenced to their scientific names, and with bibliographical references to a large number of published sources.
  3. The “Latest File”—an ongoing file of everything that continued to come in from a variety of sources after the compilation and computer indexing of the Main File. This began in a very small way but over the years grew to more than 70,000 items.
  4. Letters from the public, often in response to news articles or radio or television appearances. Replies of this kind, though of varying quality, are often useful sources of information.

DARE fostered two studies of special vocabularies in the hope of gaining understanding of certain questions concerning regionality. The lexicon unquestionably varies regionally in widespread activities such as farming, housekeeping, fishing, and lumbering. It was decided to study two extensively practiced activities to check on possible connections with geography. The activities chosen were coal mining and tobacco growing and marketing. Mary Ritchie Key collected for DARE the vocabulary of tobacco growing in eight states, and Dennis R. Preston carried out the study of coal mining in nine states.15 Both studies give evidence that certain features of occupational vocabulary are affected by general regional usage.

<< back to top

The DARE Maps

The method of coding every question in the QR, every response, every community investigated, every informant, and the most relevant social facts about each made it possible to display the data in ways helpful to the editors and to the user of the Dictionary. The most innovative of these computer-aided features are the maps that appear in the Dictionary as part of the treatment of strikingly regional words or phrases. Where a feature was found, and where it was not found, can be very quickly understood from a map. Manual mapmaking is laborious; mapmaking by the DARE computer method is so rapid and easy that it became an investigative tool of great value in the editing process.

The DARE maps are populational, not areal—that is, they say nothing about square miles of territory but show the number and distribution of the informants’ (speakers’) communities, this number having been prorated to density of population, state by state. The states appear on the DARE map as nearly as possible in their proper spatial relation to one another, and in something like their proper shape, but each is enlarged or diminished according to the prorated number of communities in it. Within each state a rectangular space is assigned to each community; these too are placed generally in their proper relative positions. The symbol indicating a specific response is then printed in the rectangle assigned to the community where that response was given. This means that empty spaces are also significant: in those communities some other response was given (unless, as occasionally happened, the question was not answered). For further detail, see The DARE Map and Regional Labels, pp. xxiii–xxxv.

Maps are used only when the data warrant them; if responses are very few or very many, a map is not a useful means of dealing with them. Each map is placed as close as possible to the treatment of the word or phrase being illustrated. The map is an aid to visualizing the region of use; details of use are in the accompanying text.

Maps are based on information derived from field collecting using the DARE Questionnaire. They are always geographical, but some are also used to show social features. Some social maps, and some contrastive maps showing more than one word or phrase, appear in the Map Section of the final volume.

<< back to top

Inclusions and Exclusions

The title of DARE had it been made in the seventeenth century, might have been “A Dictionary of That Part of the English Language that is commonly spoken in the Colonies of North America in Sundry Regions, Provinces, Tracts, Districts, and smaller Settlements, but not in that People as a Whole, Detailing its Varieties as Projected by the Native Folk.” Condensed rather drastically for twentieth-century digestion, it came out as Dictionary of American Regional English. This leaves the key word “regional” in need of sharper definition. What does DARE include, what exclude, and how are the lines drawn?

DARE does not treat technical, scientific, or other learned words or phrases—or anything else that could be considered standard.16 Beyond that general exclusion, two criteria have guided the editors: (1) Any word or phrase whose form or meaning is not used generally throughout the United States but only in part (or parts) of it, or by a particular social group, is to be included. (2) Any word or phrase whose form or meaning is distinctively a folk usage (regardless of region) is to be included. Some terms are widespread without prevailing nationally (compare the maps for baby carriage and baby buggy),17 while others may be current in a single community (for example, arab, street vendor, is used only in and around Baltimore). Regionality, then, as defined in DARE, bears no relation to the size of the area of use, so long as it is less than total.

For DARE folk usage is that which is learned in the home and in the community, from relatives and friends, not from schooling, books, or other outside forms of communication. It is traditional and largely oral; it includes anything that can be called “dialect” in the United States. But here the definitions of traditional scholarship break down. For the English Dialect Dictionary, and for the European dialect atlases, especially in the nineteenth century when the folk speech was being collected, a large part of the population was still firmly localized away from cities, not greatly affected by schooling, living in traditional ways, and keeping traditional speech. Not so in the United States, and less and less so in the twentieth century, when easy communication, mechanized farming, public education, the growth of cities, widespread migration, and most recently the tremendous success of radio and television, have disrupted even the more traditional communities, blurring former lines of dialectal division and creating a relative uniformity in speech. Even if most people still stay in one place and keep up their former ways of life, change, or pressure for change, is everywhere. And change in ways of life brings with it accelerated change in language. In the old-world sense there is little “dialect” spoken in the United States. Yet it is possible to see more “uniformity” than actually exists. In both city and countryside, distinctive regional and social differences in pronunciation, vocabulary, and syntax persist, as DARE shows.18

DARE does not cover artificial forms of speech such as Boontling (the private language used by the inhabitants of Boonville, California, and its environs) or any whose purpose or effect is to exclude the general listener. Criminal argot is of this kind, an in-group code intentionally separated from the general idiom. DARE includes such terms only if they have escaped into wider use. The same policy is followed with occupational language, sometimes called trade jargon, which only its users understand. For example, the hobo word gump, a chicken, is not known otherwise, but the jazz musician’s cool has long since come into wider use. DARE does not attempt to cover restricted occupational vocabularies—especially when the occupation is highly specialized or esoteric, such as the sexing of chicks or the wrapping of cigars. On the other hand, the vocabularies of widespread occupations such as farming, housekeeping, mining, lumbering, cattle-raising, which involve entire communities and even entire regions, are necessarily included.

Regionality is affected also by the nature of some occupations. The language of sailors, for example, folk-created and traditional, is geographically limited to use on ships wherever they go. Along seacoasts it is brought ashore and enters landsmen’s usage. But on ships sailing all over the world it becomes to some extent international. In reading sea literature for DARE the attempt was made to collect only such nautical usages as were distinctly American. Even so, regional labels that normally apply ashore do not apply to sea terms. To a lesser degree, perhaps, the language of other kinds of transportation—railmen’s, truckers’, even airline employees’ talk—is not regionally limited and is likely to be much the same everywhere, a mixture influenced by the geographical sources of its practitioners. Again, treatment in DARE depends upon whether such occupational terms are known only within the groups or have escaped into more general usage.

The language of children’s games has been treated fully in DARE because it is almost entirely of folk origin and of oral preservation, and because it shows considerable regional variation. It represents a notable combination of the traditional with many local differences. The same game, judging by the name, may be played by different rules, whereas the same game, judging by the manner of play, may have several different names. An example of the latter is the once familiar stick-and-peg game, variously called cricket, jippy-sticks, kitty-cat, knick-knock, peeny, peewee, tee-toe, whip-stick, and so on. There is no question that many variants are regional—see, for example, the maps for Andy-over, Annie-over, anti-i-over, and Antony-over. The language of children’s games preserves, in disguise, many quite ancient words and ways of speech.

The language to be collected was, then, any form of American English as spoken in the various regions of the United States. This necessarily included American “Black English,” but DARE went further and included Gullah, the only Creole English now surviving in the continental United States, and Hawaiian “pidgin,” actually Creole. As English-based folk dialects spoken by native Americans, these deserve inclusion.

A persistent problem was that natural regionality often comes into counterplay or conflict with linguistic regionality. A topographic feature, a kind of plant, a climatic condition, a bird or other animal form not found everywhere in the country, cannot have variant names where it does not exist. Folk names will be found for it only where it does exist. The most common folk name will usually become the standard one outside its natural area or for those who hear or read about it but do not know it directly. Within a natural area, however, there are often several local folk names (not always clearly distinguished), and DARE includes these. For example, the pasqueflower is also called windflower, gosling, badger, and other names; the most common name for Cathartes aura is buzzard, but it is also called carrion crow, turkey vulture, red-neck buzzard, and so on. The scientific identification of folk names is not always possible, but DARE assigning special editors to this task, has corrected many vague or erroneous attributions of the past (see, for example, the entries for apache plume, congo snake, fly poison plant, and rosemary pine).

The widespread confusion in the use of the term slang has led to its rejection by DARE. Popularly,slang is used to cover any kind of unconventional usage, especially if it has been condemned in schoolrooms. Attempts to define it are as numerous as their attempters, and while there is some core of agreement, the word remains so imprecise that its use as a scientific term has been challenged. Finding slang too indefinite and too often used merely to condemn, DARE dispenses with it and uses more definitive and objective labels.

Some other distinctions need to be noted. Speakers of all kinds have, in the well-known phrase, a vocabulary of use and a vocabulary of recognition. That is, everyone has some acquaintance with a great many more words than he or she habitually uses. DARE seeks, through the spontaneous immediate answer to a QR question, to find out what word the informant (and presumably others in his or her community) does habitually use. For the country as a whole these may be relics or innovations, commonplaces or rarities. DARE attempts to label them appropriately. Obsoleteness is difficult to prove; many present folk terms were once standard or literary, but have gone out of general use. Yet some are revived, or, like the coelacanth, are thought to be extinct and then are suddenly discovered alive and well. Accordingly, DARE uses the label obs sparingly, normally if the most recent example found is a century or more old. With few exceptions, terms have not been entered unless they are in current or recent use. As a part of DARE’s function as a historical dictionary, changes of status are also attended to: the decay of formerly widespread usages, the spread of formerly restricted usages, folk terms that have made their way into written or literary use.

It need hardly be said that borderline cases have been numerous. Most teasing have been questions arising from insufficient evidence. A single word or phrase given by an informant, not followed up at the time by the fieldworker, and nowhere found again, may nevertheless have the ring of genuineness. Here the decision to keep or to reject must rest on the editors’ “feel” for the language, their experience and judgment. At the collecting stage it was appropriate to err on the side of inclusiveness; the collector could not know what other support for a strange form might be waiting in the files. At the editing stage, however, forms had to demonstrate their right to inclusion. Putative loanwords had to be found in actual use in English-language contexts. Apparent individualisms had to have analogs that made them seem likely to be more generally used. In the DARE entries, a double dagger (‡) is prefixed to entries that may well be folk usages but for which conclusive evidence has not been found.

<< back to top

Treatment of Entries

DARE entries are normally presented in three parts: the opening section, the definition, and the supporting quotations. The first part, using many abbreviations, conveys the basic information about each entry: headword or ‑words, part of speech, pronunciation, variant forms, etymology, geographic range, usage, cross-references, and editorial notes. In the second part, meanings (if there are more than one) are numbered, with alphabetic subdivisions if necessary (see, for example, the treatment of be). Because standard senses are not normally treated, the historical sequence of senses is sometimes not obvious. Editorial notes are included when deemed necessary. Boldface type is used for cross-references and to give prominence to dates and regionality.

Entry Form

Most entry forms are single words, but there are also numerous compounds and phrases. All are entered under the standard spellings or under established dialect spellings, with cross-references at variant spellings.19 Entries for which there was no established spelling—those recorded by phonetic transcription or on disks or audiotapes—have been spelled by analogy with the most similar standard spellings and with due attention to etymology. Examples: ooch over, to move oneself along (as on a bench); patalca and kitarber, variants of catalpa.

Entries are strictly alphabetical, except for spaces or punctuation marks, which are taken into account only as they move an otherwise identical form down the alphabet.

As to the vexing question of whether to spell compounds open, hyphenated, or solid (coal hod, coal-hod, coalhod), DARE follows Webster’s Third New International Dictionary except in the rare instances when new evidence indicates that a form should be spelled otherwise. (Quotations from written sources of course reproduce the spelling of the source.)

When variant forms are almost equal in frequency, they are equally usable as the headword. The choice then rests on such considerations as etymology, alphabetic convenience, and clarity of presentation. When one variant is clearly more frequent, it is taken as the headword; the others are listed after part of speech and pronunciation, preceded by “also.” Example: cabbage pea . . . Also cabbage bean.

Such entries as the plant name angel’s-trumpet raise the problem of the apostrophe. Attested written forms are angel’s, angels’, and angel; spoken forms are ambiguous—there is no way of knowing how the speaker would have written them, and the fieldworker’s spelling only reflects personal graphic habits. To present all these small variants economically, DARE uses swung dashes and parentheses to indicate repeated and optional elements. Example: angel’s-trumpet . . . Also ~ trumpets, angel(s’) trumpet(s).

Phrases are entered under the word or words that form the stable core, though other parts of the phrase may vary. For example, he wouldn’t know beans from barley may also have other initial pronouns (she, they), other auxiliaries (can’t, doesn’t), other verbs (tell), and especially other contrasters (beans from butter, B from a bull’s foot, and several more). The phrases are entered with the core words first, as beans, not to know and B from (a) bull’s foot, not to know, with appropriate references to other phrases of the same meaning entered elsewhere in the alphabet, such as A from B, not to know.

Parts of Speech

The traditional nine parts of speech are indicated by abbreviations immediately following the entry forms: n noun, pron pronoun, adj adjective, v verb, adv adverb, art article, prep preposition, conj conjunction, intj interjection. Other labels used occasionally in this position are exclam exclamation, part particle, and phr phrase.

Noun includes single words and also compound nouns and those derived by conversion from other parts of speech. Examples: arrow, arrow chase, Adam-and-Eve root, boiled owl, carry-in. Phrasal nouns are treated simply as nouns. Virtually any noun can be used attributively; DARE notes attributive uses that seem significant.

Verb is a general label; transitivity or intransitivity is not indicated unless there is immediate need to note the distinction. The definition makes it evident whether the use being illustrated is transitive or intransitive. Nonfinite verb forms are labeled infin infinitive, pple participle, vbl n verbal noun, ppl adj participial adjective.

Verb phrase is the label for a verb regularly construed with an adverb and forming a loose unit with it: fall away, hog down. It is also used for set verb-plus-object phrases, such as eat dried apples, to become pregnant.

Adjective phrase, adverb phrase, preposition phrase, and conjunction phrase are so labeled. Examples: dry-headed, adj phr; side by each, adv phr; over against, prep phr; as how, conj phr.

Distinction is made between interjection and exclamation: the interjection stands apart from syntactic context and has no formal syntax of its own, whereas the exclamation has its own syntax, surface or underlying, and may be joined to the adjacent contextual unit. Examples of interjections: ooch! phooey! Crimanetly! Examples of exclamations: Andy-over! King’s ex! Carry me out with the tongs!

A few entries could not be classified with any of these labels. They have morphological status as affixes (prefix, infix, suffix) or inert phonetic units. Examples: a, ker-, -ma-.


DARE has supporting oral data, recorded in phonetically written form or on phonograph disks or audiotapes. No pronunciations inferred from spellings are given. Pronunciations follow the indication of part of speech and are given in broad International Phonetic Association (IPA) characters enclosed in vertical lines. Pronunciation-spellings—those in which writers have sought to suggest actual pronunciation through spelling—are also listed fully. The reader may infer the pronunciation from these, up to a point; but such inferences must not be taken as equivalent to recorded speech.

Pronunciations given phonetically or by various kinds of respelling in the quotations are generally retained. Forms of transcription earlier than IPA are updated to it; otherwise, as with other quotations, they are unchanged. The only variances from IPA characters are č, ⊥, ⊤, ⊢, ⊣, DARE uses ˄, ˅, ˃, ˂, following Linguistic Atlas practice, to indicate positions of articulation respectively higher, lower, farther back, and farther forward from the positions of “cardinal” vowels. Narrow IPA transcriptions, when given, are placed in brackets. Occasionally, to avoid misunderstanding, a hyphen is used to indicate that a disyllabic vowel cluster rather than a diphthong is intended, as in carry |ˈkæ-i|.

Variant Spellings

DARE for reasons of both pronunciation and orthography. Up to a point, as noted, past pronunciations can be inferred from past spellings; sometimes, also, variant spellings make it possible to follow the stages through which new or unfamiliar words acquire settled forms.

DARE differentiates among spelling-pronunciations, pronunciation-spellings, and eye-dialect spellings. A spelling-pronunciation is one in which a speaker follows the written form of a word, pronouncing all the letters even though some of them are properly silent or have other values than the ones he or she gives them. For example, the speaker may pronounce the l in palm, calm, balk, chalk, or give full value to all the letters of boatswain and colonel. Spelling-pronunciations are a byproduct of the shortcomings of English orthography, which has more than one value for many letters or combinations of letters and more than one way of spelling the same sound. Past methods of teaching spelling in the schools by letters and syllables have contributed to the production of many spelling-pronunciations in folk and regional speech. In this instance, writing brings about changes in speech.

Pronunciation-spellings are those in which a writer tries to represent the nonstandard pronunciation of a speaker. The word calm, with the regional variants |kɑm|, |kæm|, |kam|, |kɑrm|, |kɑlm|, appears in writing not only in the standard way (which will do for any variant but does not specify the one used) but also as cam, ca’m, carm, cyaam. Some of these may merely indicate that the writer was a bad speller. Others may be attempts to reflect some nuance of a specific variant. Most often the writer is trying to characterize an individual, fictional or not, through his speech. Some writers do this very badly, with apostrophes sprinkled wherever the pronunciation does not agree with standard spelling; others succeed better by using a simple “phonetic” representation. To indicate an “r-less” pronunciation of through, for example, th’ough signals to the eye, which then translates for the ear—a two-step process; thoo, by contrast, signals directly to the ear. In each case, of course, context confirms the intended meaning.

Eye-dialect spellings are those which an author uses intentionally to suggest that a character’s speech is substandard and that the person is illiterate, even though those spellings correspond to pronunciations that are perfectly standard. Cum, slay, thawt for come, sleigh, thought are phonetically correct but suggest that the character has had little schooling. Eye-dialect was widely used as a device of folksy or popular humor by such nineteenth-century writers as Artemus Ward and Petroleum V. Nasby, and in this century by Ring Lardner.

The gradual settling down of a new word, often a loanword from a foreign language, into an accepted American form may sometimes be followed through variant spellings. For example, Spanish vaquero, borrowed into southwestern usage, includes among its variant spellings: 1827 bakhara; 1847 baccaro; 1862 bukkarer; 1873 buccahro; 1889, 1910 buckayro; 1890 buckhara; 1900 buccaroo; 1907, 1919 buckaroo. These are pronunciation-spellings that attempt to render the foreign word as said by speakers of American English. Its form varies until a standard spelling becomes established—in this case with a peculiar adaptation to the influence of -aroo or -eroo, a suffix that became popular in the early part of the twentieth century.

Though some of the spelling variants recorded may individually seem trivial (and a few no doubt are), they have value in the aggregate. As noted, they may help to show changes in pronunciation, naturalization of loanwords, processes of word formation, and other historical developments in the language. Alternate spellings are usually listed alphabetically, but are sometimes grouped by similarity of form.


DARE does not give etymologies for all entries. Those well treated in general dictionaries need not be repeated. Compounds or phrases whose components are standard in form and sense, such as Abolition War, about to die, acting pole, bullneck, bluejoint, need no formal etymology. When appropriate, DARE indicates how the phrase differs from the sum of its components. Here etymology and definition work together.

Etymologies and explanations of origins appear in square brackets. They are given, when possible, for certain types of entries:

1. Words and phrases that most other dictionaries do not treat. Examples: Abe Lincoln fence, bobbasheely.

2. Foreign loans—words or phrases—usually traced only to the proximate source language. Examples: aber nit, abri, aguardiente, apee.

3. Any entry in which an existing etymology can be improved. Examples: abiselfa, appaloosa.

4. Entries in which nonstandard forms or meanings require explanation. Examples: ace-boon-coon, amarugian, anchor ice.

A distinction is sometimes made between echoic and imitative (abbreviated imit), echoic referring to names derived by similarity of sounds, imitative referring to instances in which some feature other than sound is prominent. Thus zoom would be called echoic; zig-zag would be called imitative.

When an origin is unknown but DARE’s research has turned up possibly relevant information, this is guardedly recorded as an aid to future research.

The Trademark Register of the United States, this fact is stated in the etymology or in a note. DARE’s entries reflect actual observed usage and make no judgments as to the legal status of the word.

Geographical Labels

Areas within the United States are popularly understood in different ways; there is no general agreement on what such terms as North, South, Middle West include or exclude. In the interest of uniformity, the editors of DARE devised a list of geographical divisions for use in the entries. These geographic labels are often qualified with esp (for especially) or the somewhat stronger chiefly. In quoted matter, of course, the authors’ labels have been respected.

Usage Labels

As noted earlier, the editors of DARE have sought to keep usage labels and notes factual and objective, to describe usage as it is (or as it was during the period of primary data collecting, 1965–1970), never to say how it should be or to express the editors’ opinions about it. (Other people’s opinions, frequently expressed in the quotations, often throw light on changes of attitude over the years.) Emotionally loaded labels such as illiterate, vulgar, provincial have not been used.

Labels give information of four kinds: (1) amount of use, as rare, occas, freq, usu; (2) currency, as obs, arch, old-fash, hist; (3) type of user, as rural, Gullah, grade-school educ; (4) manner of use, as joc, derog.20 (All abbreviations are identified in the List of Abbreviations, pp. clii–clvi.)

The basis of many of DARE’s usage labels is furnished by the social factors mentioned earlier: the informants’ age, sex, race, level of education, and type of community. When these correlate in some notable way, varying from the average, a label is called for. For example, in qu. H35, for dropped eggs, 83 percent of the responses were from old informants, whose percentage in the total of informants answering was 70 percent; thus the label somewhat old-fash was clearly called for. In qu. U27, for beau dollar, 79 percent of the responses were from Black informants; thus the label esp freq among Black speakers was justified. These are statistical statements with no further overtones.21

Labels are given to help the reader to a clearer sense of the geographic or social status of a word or phrase: they cannot be expected to be exhaustive or absolute. They are valid as far as DARE evidence goes—no farther. Fuller detail can be found in the database in the DARE archive.


DARE uses dated quotations to support the definitions. In choosing which examples to use, the editors have preferred “defining” quotations to those in which the word or phrase being illustrated is merely mentioned or used in passing. Original sources have been preferred to derivative ones. As to the choice and number of quotations, the editors have sought to give the earliest example found in American use, at least one per century, and a late or recent one. Quotations should also illustrate and justify the geographical label or note. Quotations are placed in brackets (as in OED) when they do not exactly illustrate the entry because of differences in form, sense, or region of use.

An important feature of the documentation is the use of oral sources. Examples taken from Dialect Notes are among the earliest obtainable; these and the PADS word lists have all been taken into the files. American Speech has been fully read. The Linguistic Atlas materials (including unpublished workbook notes taken by the fieldworkers) have been used copiously, as have the Hanley recordings and the DARE audiotapes. But of course the most valuable part of the entire corpus is the Data Summary made from the DARE QRs. These oral sources are examples of unedited, genuine American usage by native speakers throughout the nation whose regional and social attachments are known.

Next in value are such written materials as diaries, personal letters, unedited accounts of travel, adventure, and daily life; these are as close to actual usage as any written materials can be. After these come regional literature written by authors who knew their regions thoroughly and did not romanticize or stereotype their characters. Among older writers of this kind might be mentioned Edward Eggleston and Joel Chandler Harris; among more recent ones Marjorie Kinnan Rawlings, Julia Peterkin, and William Faulkner. But many regional writers have accepted stereotypes and clichés, have exaggerated or overdone the regional speech, or, worse, have got it wrong. The editors have been wary of this class of literature and have quoted from it cautiously if at all.

With travel accounts a difficulty arises. Suppose a native of Massachusetts writes about his journey to Texas. Two regions and two kinds of usage are involved here: the traveler’s home idiom as he writes, and the Texan usages on which he comments. The editor must keep these apart.

Quotations of data from DARE QRs are necessarily formulaic, and therefore may not be immediately easy to read. But one soon becomes accustomed to them, and as they contain the most valuable data, the reader is advised to be patient. They give the question asked in the QR, with its code, and list the responses, with the codes of the informants responding so. The question is sometimes given in full, sometimes abbreviated, sometimes only referred to, according to evident need. (For the full text of the QR, see pp. lxii–lxxxv.) Every question code is an implicit cross-reference to the Data Summary in the final volume, which lists every question and every response. DARE quotations therefore display in condensed form the best oral evidence, to which other sources are additional.

The Bibliography in the final volume lists every source quoted in DARE, giving the reader all the facts necessary to track a quotation back to its origin. To save space, short-titles are used in the DARE entries. Short-titles retain the first or most important noun or verb and enough other words to ensure identification. Examples: Simon Ansley O’Ferrall, A Ramble of Six Thousand Miles through the United States of America, becomes 1832 O’Ferrall Ramble; William Prince, A Short Treatise on Horticulture, becomes 1828 Prince Horticulture.

Dating of Quotations

No problem of dating arises in quoting from a book that was printed only once, but other situations are more difficult. For example, one may want to quote from a record dated 1695 that was first printed in 1705 and is now accessible only in a 1970 facsimile of an 1870 printing. The Bibliography in the final volume will tell all this, but the DARE entry will never give more than two dates, the earliest and latest: here 1695 (1970). This is enough to identify the source and say that the original record was not used, but the facsimile. Any single date tells the reader that the original source was used. When a single date is given, pagination of the quotation is from that edition; otherwise it is from the edition published at the second date.

This is the general rule, but a number of complexities had to be dealt with. DARE uses five standard forms for indicating dates:

1. date

2. date (date)

3. date in date

4. date Author Title [for/of date]

5. [any of the above] Author Title (as of [date])

1. The date of the earliest edition in which the quotation appeared. Also used for microform reproductions. Example: 1938 FWP Guide MN.

2. The date of the first edition of a work plus that of the later edition from which the DARE quote was taken. Situations in which this format is used include: (a) a later printing from the same plates (not from reset type); (b) a new edition with complete resetting of type, therefore usually with different pagination and possible revisions in the text; (c) a photofacsimile edition of an earlier work; (d) a posthumous edition of undated material, using ante with the author’s date of death as the date of composition, as in a1862 (1865) Thoreau Cape Cod; (e) a diary, with the two dates referring to the dated entry and the date of publication.

3. Matter quoted in a later publication: the first date is that of the quoted matter, the second that of the work in which it is quoted. Examples: 1789 in 1889 Washington Writings; 1637 in 1850 CT (Colony) Pub. Rec. This form is used for letters or speeches quoted in a biography, collections of government documents and historical society publications, collections of an author’s works known to have been published earlier, and in similar situations.

4. A work in which the title contains a date other than the date of publication. This occurs in almanacs and in the annual publications of societies and governmental departments. In the DARE reference the first date given is the date of publication. Examples: 1869 TX Almanac for 1870; 1850 MI State Ag. Soc. Trans. for 1849.

5. Material recollected from a time considerably earlier than the date of the publication. Example: 1962 Morison One Boy’s Boston 54 MA (as of 1900).


In a dictionary that pays special attention to variance of forms, there is necessarily a high proportion of cross-references. In DARE, variants are cross-referred to the entry under which they are treated. Many cross-references within entries call the reader’s attention to synonyms, related forms, or analogous forms. Any cross-reference in bold type is to a DARE entry.

Definitions and Sense Divisions

The intention in DARE definitions is to use standard but simple words, to avoid the technical or recondite. When, as occasionally happens, a word of popular speech suits the purpose better than a standard one, it may be used as well. Natural objects (plants, animals) are defined in everyday terms and also, when identifiable, with their Latin genus-species names. When the only evidence is in a single quoted source, or when the source gives an adequate definition, no other is usually given.

Definitions may be preceded by: (1) grammatical notes such as also attrib, constr as pl, in passive usage; (2) delimiters such as In marble play, Of cattle, Among loggers; (3) semantic or stylistic indicators such as fig, transf, in ironic use. Synonyms may be defined simply by cross-reference to the best-known equivalent elsewhere in the Dictionary.

The division of senses is a necessity of presentation. Most sense divisions are easily set up on the basis of the quotations at hand. But there are inevitably a number of borderline cases, which must be put on one side or the other of the line. Where they are put rests ultimately on the editors’ perception, experience, and feel for the language.

The sequence of senses is ideally historical and logical, but many terms, especially those of everyday speech, exist in use long before their first known appearance in writing. Senses may also develop simultaneously in more than one direction. When dated evidence is lacking and more than one sequence of development seems logical, it is often impossible to establish the exact filiation of senses. Clear presentation is then the best resort.

James Hulbert, second editor of the Dictionary of American English, wrote in his almost poetic “Consolation of Lexicography”:

I know of no more enjoyable intellectual activity than working on a dictionary. Unlike most research, lexicography rarely sends one on fruitless quests. One does not devote days, months, or even years to testing an hypothesis only to decide that it is not tenable, or to attempting to collect evidence to prove a theory only to have to conclude that sufficient facts are no longer in existence to clinch it. It does not make one’s life anxious, nor build up hopes only to have them collapse. Every day one is confronted by new problems, usually small but absorbingly interesting; at the end of the day one feels healthily tired, but content in the thought that one has accomplished something and advanced the whole work towards its completion.22

This picture is broadly true, but the tint is too rosy. To attempt a work of the scope of DARE is to give hostages to imperfection. The task of covering all the regional variation in American English is beyond human accomplishment. Compromise is inescapable. On occasion, contrary to Hulbert’s optimistic words, editors have spent days, months, and even years in tracking down individual items, correcting bibliographic references, finding original editions, establishing correct authorship, dates, pagination, the exact wording and spelling of entries—only to have the findings condensed into a few inconspicuous lines no different, in appearance, from the easy and obvious entries. The editors have repeatedly had to settle for inconclusiveness when in their hearts they felt that further evidence could be found “out there somewhere” if only they could abandon all else and go in pursuit of it. But that way madness lies—or at least the guarantee that a tangible, published dictionary will never be made. One of the things a lexicographer must learn—sometimes painfully—is how and when to let go. If all facts are equal in the eyes of Science, the realities of the world are fraught with inequality.

In DARE the editors have sought to produce a work of useful scholarship, one also that will testify to the wondrous variety and creativeness of human language, and specifically of the English language as it is used regionally in the United States. To this task they have given many years, much thought, and much hope. May DARE prove a worthy monument to their labors.

<< back to top

  1. Linguistic Atlas of New England (LANE), by Hans Kurath et al. (Providence: Brown University, 1939–1943).
  2. A Dictionary of American English on Historical Principles (DAE), ed. Sir William A. Craigie and James R. Hulbert (Chicago: University of Chicago Press, 1938–1944).
  3. American Dialect Dictionary (ADD), by Harold Wentworth (New York: Crowell, 1944).
  4. The WELS questionnaire, revised, was published in 1953: Frederic G. Cassidy and Audrey R. Duckert, A Method for Collecting Dialect, PADS 20.
  5. The ADS had referred to the projected dictionary, over the years, as “American Dialect Dictionary,” but in 1944 Harold Wentworth had preempted this title. Actually, the state of dialect in the United States makes the present title more appropriate.
  6. In such lists there is much duplication of widespread names such as white oak, black oak, red oak, live oak; many others are named only once. The coverage is therefore not as full as for questions about a single object. Nevertheless, open questions collect classes of words that could not be covered singly.
  7. “Le questionnaire . . . pour être sensiblement meilleur, aurait dû être fait après l’enquête.” Etude de Géographie Linguistique, Pathologie et Thérapeutique Verbales, 1 (Bern: Librairie Beerstecher, 1915), 45.
  8. The photographs were taken from the best book available at the time: Homer House, Wild Flowers (New York: Macmillan, 1961).
  9. DARE averages 20 communities per state, ranging from 86 (NY) to 2 (NV); 20 states have more than the average, and 31 have fewer than the average. The Linguistic Atlas coverage is more intensive and affords the less populous states more than their proportional share.
  10. The story was first used for phonetic study over a century ago in England by Henry Sweet. In this century it was adopted by W. C. Greet at Columbia University, who made phonograph recordings of it with his students from many parts of the United States (Victor and Victrola, “American Speech,” Under the Direction of H. M. Ayres and Cabell Greet, Columbia University, nos. 65–76). At some point, the young rat, né Grip, was renamed Arthur, because phonetic variation would be greater in the latter name. For the DARE version of the text, see the Guide to Pronunciation, p. xliii.
  11. This was done always on the basis of the QRs. No evidence was rejected. Editors made spellings consistent, removed differences in fieldworkers’ practices, and unified matters of form, such as the proper form for multiword phrases. The condensation was achieved by grouping together all responses with nine or more informants and itemizing only those with eight or fewer.
  12. In general, the most frequent response is considered standard. When two alternate terms are used with nearly equal frequency, neither can be considered the exclusive standard. Examples are twilight and dusk (370 vs. 365 responses), and neigh and whinny (387 vs. 383 responses). All are used in every state; no one clearly dominates. All must be considered standard. With such synonymous pairs, however, there is usually a difference of geographical concentration: the areas of use overlap but are not congruent. Thus whinny is strong in the North, weak in the Southeastern and Southern states; neigh is much weaker in the North, more evenly distributed through the country as a whole. What we have here are regionally standard words.
  13. Gordon Wilson, Sr., had made a card file of about 9000 lexical items, gathered orally from more than 225 local people, with exact biographical information, and with phonetic transcriptions, definitions and usage notes, and annotations showing in what dictionaries they were treated. In the 1940s he made many recordings of the regional speech. All his materials were put into the DARE Main File; the original cards and audiotapes are in Bowling Green at the Kentucky Folklore Society. Joseph Sargent Hall’s card file, also put into the Main File, consists of material from interviews with about 230 informants, with brief biographies and transcripts of audiotapes. Also very useful was his monograph, “The Phonetics of Great Smoky Mountain Speech,” American Speech 1942, no. 2, pt. 2. The Hench Collection may be consulted in the University of Virginia Library, Charlottesville. Kelsie Harder not only answered the WELS questions but also excerpted letters his mother had written to him during his army service—a valuable source for linguistic evidence from central-western Tennessee. Russell Tabbert, collecting data for a Dictionary of Alaskan English, put his collection at DARE’s disposal in 1980.
  14. The tapes are in A. L. Davis’s collection, with a copy in the ADS Archive, University of Massachusetts, Amherst. A full account is in the DARE archive: Margaret Waterman, “The Hanley Tapes,” 1974.
  15. The study of coal mining has been published, with an annotated glossary, in PADS 59.
  16. In a relatively small number of instances a standard word is entered (though not itself illustrated) so that nonstandard terms may be listed together under it. Examples: anhinga, blue jay, fringed polygala.
  17. Baby carriage clearly prevails throughout New England and is found, though less commonly, in the rest of the country; baby buggy, also widespread, though sparse in the Atlantic states, becomes the prevalent form from the longitude of Ohio westward to the Pacific. In other words, the areas of use overlap: people who say baby buggy probably have heard and would understand baby carriage, and vice versa. One term is no more standard than the other. Yet there is a significant regional difference in their areas of concentration, and DARE must include them. Such borderline cases are relatively few, however: the great majority of the words and phrases treated in DARE are more narrowly local, and many are totally unknown outside their own small areas. Not many people outside New York City know that a sliding-pond is a playground slide; not many readers of these lines would be able to gloss skillypot, grass onions, jockey-box, work brickle, genavy.
  18. Dialect differences are most distinctive as to pronunciation, secondarily as to vocabulary; syntactic differences are relatively few and less conspicuous.
  19. Except in rare cases, the headword spellings follow those of Webster’s Third New International Dictionary of the English Language (Springfield, MA: Merriam, 1961).
  20. Jocular implies intentional or self-aware joking by an individual; humorous implies a wider consensus.
  21. The computerized map-making program works out statistics on which quite exact application of usage labels may be based. For example, it sorts on the exact numbers and identities of informants responding with the specific word or phrase being examined, and on each of the five social factors about each informant. For examples of the use of such statistics, see animal 1, all in, all the, and their usage notes.
  22. J. R. Hulbert, Dictionaries British and American (London: Deutsch, 1955), 42–43.

<< back to top