Iowa Materials for the Linguistic Atlas of the Upper Midwest

Almost forty years have elapsed since it was proposed at the 1928 meeting of the Modern Language Association that steps be taken to begin the making of a Linguistic Atlas of the United States.¹ The initial momentum of the project was such that the fieldwork for the monumental three-volume, six-part Linguistic Atlas of New England was completed within the 25-month period from the summer of 1931 to September, 1933, by a team of nine linguists, and the editing of the first two volumes was completed by the spring of 1939. The entire six parts of the New England atlas were published between 1939 and 1943. Since then, fieldwork and collation of data for the various regional atlases comprised in the grand project have proceeded at a less impressive rate, generally operating with insufficient resources — both financial and human.

Among the regional atlases for which the fieldwork has been completed and which are in the collation and editing stage is the Linguistic Atlas of the Upper Midwest, which will present the linguistic geography of Minnesota, Iowa, the Dakotas, and Nebraska. This regional atlas is under the editorship of Professor Harold B. Allen of the University of Minnesota, one of the fieldworkers for the atlas of the North-Central region.² The fieldwork for the Upper Midwest was completed in 1957, Allen himself having collected more than half the field-records. For Iowa, however, the fieldwork was done in the summer of 1950 by two graduate students at The University of Iowa, Frank S. Hanlin and Virgil A. Peterson. The Iowa fieldwork was directed by Rachel Harris Kilpatrick,³ one of the nine field-workers for the Linguistic Atlas of New England. Professors John C. McGalliard of the Department of English and Erich Funke, then chairman of the Department of German, served as an advisory committee for the project. In preparation for the summer of fieldwork, Mrs. Kilpatrick was appointed a lecturer in the Department of German, where she gave a special course during the spring semester of 1949-50 in American linguistic geography and the methods of fieldwork. From the students in this course, Hanlin and Peterson were selected as fieldworkers and given fellowships by the Graduate College. The field project was further supported by a $1500 grant from the Graduate College. Hanlin and Peterson were sent to Minneapolis for a brief check by Allen on their training and for final instructions, and the fieldwork began. (The frequent reports sent from all over the state to Director Kilpatrick by the two fledgling fieldworkers during the summer constitute an interesting firsthand account of the grueling pace, the frustrations, and the satisfying triumphs of dialect fieldwork.) The survey was completed in a period of ten weeks, and duplicates of the field transcriptions were bound and placed in the Special Collections Department of the library, along with some mechanical recordings of the speech of selected informants. Concurrently with the field investigation, some 200 lexical questionnaires were distributed by mail throughout the state, and these completed questionnaires also constitute a part of the Iowa linguistic survey materials now in Special Collections. Since the field transcriptions are the primary materials for the compilation of the atlas, we shall begin the description of the materials with them.

Field transcriptions are the end product of a process which includes a great deal of careful preparatory work before the field-interviewing even begins. In fact, the beginnings of the Iowa survey can be said in one sense to go back many years before 1950 to a long tradition of gathering and publishing word-lists from widely separated parts of the country. These lists of regionally peculiar terms furnished the primary material for the “worksheets” — lists of items to be elicited in field interviews — which guided the survey of New England, and which, with modifications dictated by cultural, economic, and geographical differences, have been the basis for all the subsequent regional surveys. A carefully designed worksheet assures the basic uniformity of data gathered in the field, and the comparability of results from region to region. The worksheets for the Upper Midwest, like those for the other regions west of the Atlantic Seaboard, are a shorter adaptation of the original 800- to 850-item questionnaires used for the eastern areas.

The Upper Midwest worksheets, revised by Allen in 1950 from those compiled by Hans Kurath, editor of the New England atlas, in 1939, comprise 574 items. Following the pattern set by the New England survey, the individual items are arranged topically in an order designed to facilitate eliciting them in a natural conversational transition from topic to topic.

The worksheets are designed to elicit various kinds of information — lexical, phonological, and grammatical. The following section from the Iowa worksheets will serve as an illustration:

The fieldworker attempts to elicit these items in natural conversation, without himself pronouncing the words and phrases he wants the informant to say. In the set of items given here, items 1, 6, and 7 have regionally varying pronunciations, and are thus included here primarily as phonological items. Items 4 and 5 are verbs whose inflectional patterns are known to vary regionally and socially in other parts of the country, and thus constitute two bits of grammatical data for Iowa. Items 2 and 3 are known to have lexical variants elsewhere (three potential variants, for the guidance of the fieldworker, are listed in the worksheet for item 3, as are morphological variants for items 4 and 5, as well as a lexical variant for item 4).

The fieldworker enters in his field-book, with one page for each set of items, a phonetic transcription of the informant’s pronunciation of each item as it is elicited, and adds — also in phonetic notation— any other form of potential or actual linguistic interest which the informant utters in the course of conversation.

Quite obviously a single interview consumes many hours. As a rule, an interview would be distributed over three or four sessions, since neither informant nor fieldworker could be expected to tolerate more than two hours or so of this kind of conversation. Assuming about eight hours per interview for the forty-eight Iowa informants, we would calculate about twenty interviewing hours per fieldworker each week for the ten-week period of the survey — a tight and trying schedule, when we consider that many hours were spent in locating suitable and willing informants, and in traveling from community to community.

In addition to the compilation of worksheets appropriate to the area to be surveyed, there is the preliminary task of selecting the communities to be included in the survey. These must be selected, of course, for even geographical distribution, but a more fundamental consideration is that of the historical settlement pattern. Dialect differentiation cannot be studied altogether apart from either the patterns of linguistic change or the history of population movement. Therefore it is imperative that the communities surveyed include (a) the early settlements, (b) the more important secondary settlements reflecting the spread of the population, (c) representative contemporary flourishing and stable towns, and (d) towns which have, for one reason or another, “gone downhill.” Such a distribution of communities insures that the linguistic data gathered are to some extent revealing of the historical basis of dialectal features.⁴

The selection of informants — apart from the preliminary broad decisions about the total number of informants, the number to be chosen in each community, and the general qualifications and classification of informants — is a part of the fieldworker’s job when he arrives in a community. In keeping with the general principle that the survey of the Upper Midwest seeks to get at the oldest accessible stratum of English usage in the region, one representative each of two main categories was selected in each community: (a) a speaker at least seventy years old, with a minimum of formal education, and (b) a middle-aged informant with somewhat more schooling. Both, of course, were native residents of the community, preferably members of the oldest families.

The second type of materials from the Iowa survey is the lexical questionnaires. Some 200 of these were distributed by mail to informants sixty years old or older in more than eighty counties. In this sort of postal questionnaire, specific objects or activities are described and the informant is asked to circle the one or more terms listed beside the description which corresponds to his own usage, and to write in any term for the described artifact, etc., which he uses but which does not appear in the list given. He is also asked to give certain particulars about himself, such as his age, the length of time he has lived in the community, the birthplaces of his parents and grandparents, the extent of his formal education, etc. The postal questionnaire, which in this instance includes about 165 separate items, provides a means for quickly surveying a large number of informants over a large geographical area, with a minimum of time and personnel. This kind of survey, of course, is impracticable and unreliable for any sort of data except lexical, but within this limitation it provides a useful check on and supplement to the more thorough and informative, but time-consuming and laborious, field investigation.

Finally, there are — or were — mechanical recordings of a number of informants who were interviewed by the fieldworkers. According to a report submitted in January, 1951, by Mrs. Kilpatrick upon the completion of the project, tape- and wire-recorded samples of connected speech had been made for a number of selected informants, and these had been transferred to twenty-one double-faced discs. These were to have been deposited in the “Phonetics Laboratory” (the predecessor of the present Language Laboratory in Schaeffer Hall). These discs, unfortunately, seem to have evaporated. A search into the remotest comers of the Language Laboratory has failed to turn up anything resembling them, and no one connected with the original project has any solution to the mystery of the evanescent discs. All that remains of the recordings is four spools of wire, which have recently been sent to Allen at the University of Minnesota, who has access to a wire-recorder (a rarity among machines in this age of tape), and who should soon determine whether after sixteen years anything audible remains of these recordings. It is most unlikely that anything does, as wire-recordings tend to lose their magnetic imprint rather rapidly.

The loss of these recordings is lamentable for several reasons. First, their loss means the loss of a check on the phonetic accuracy of the fieldworkers’ transcriptions. Second, a recording of running conversation almost always contains linguistic data which the field-worker, in his attention to the worksheet items, has overlooked. Finally, to judge from several pages of handwritten notes included in the miscellaneous papers relating to the project, the contents of the recordings would be of some cultural and historical interest: one elderly informant tells of a visit to his community by the notorious Frank and Jesse James, another describes the fate of a pair of horse-thieves in Shelby County in 1883, another describes a Confederate raid during the Civil War, another gives a detailed account of a “chivaree” (a noisy serenade to a newly wed couple), etc.

What opportunities for research are there in the materials of the Iowa survey? The primary use of these records is to be, of course, their incorporation into the Linguistic Atlas of the Upper Midwest. But as has already been said, the atlas itself is a presentation of data in organized form; the interpretation of the data is another task. Quite distinct from the Linguistic Atlas of New England itself and from the projected regional atlases of the eastern states for example, but based on the atlas data, are such studies as Hans Kurath’s Word Geography of the Eastern United States (1949), E. Bagby Atwood’s A Survey of Verb Forms in the Eastern United States (1953), and Hans Kurath and R. I. McDavid’s The Pronunciation of English in the Atlantic States (1961). A number of shorter studies of particular linguistic features of areas in which atlas fieldwork has been completed have also appeared, considerably in advance of the publication of the atlases themselves.

One study based on the Iowa survey materials was done almost concomitantly with the survey — a Master’s thesis by Laura B. Rush, who worked on the distribution and tabulation of the lexical questionnaires, entitled A Lexical Study of Twelve Items in Iowa. Using the data from the same lexical questionnaires, the present writer and a graduate student in linguistics, Charles Houck, are preparing a study in which the computer is used to interpret statistically the entire corpus of data from these questionnaires, and in which these results are compared with the results of a traditional interpretation in terms of individual regional features in the speech of a “transition area” — or area of regional mixture — such as the Upper Midwest.

A number of interesting extensions of this approach are possible, using the field transcriptions themselves, and including phonological and grammatical data as well as lexical, to say nothing of a wealth of more limited studies of the distribution of particular features or groups of features that may reflect the patterns of settlement in early days. Moreover, studies of changes in the speech patterns of the communities investigated in 1950, by further fieldwork in the same communities but with younger informants, might yield interesting insights into the linguistic effects of increasing industrialization and population mobility.