Smithsonian Institution, and moreover at universities there are graduate students to work on the records. However, the Atlas should be under the direction of some national organization like the Linguistic Society of America, that scholars in all parts of the country might feel free to use the material. The compiling and the organizing of the mechanical records should be done under the Director. Motion to strike out preservation of was carried.

BRYAN: moved that the organization of the depot be left to the Managing Committee. Carried.

It was decided not to spend over twenty minutes apiece on Questions VI and VII.

VI. What sort of training should the Director and his field assist

ants have?

MÜLLER: mentioned again the evidence that Edmont heard nearly perfectly. He preferred that no mechanical methods be used except to test the hearing of the field workers.

FLETCHER: All the investigators should have a hearing test, as hearing is frequently defective.

METFESSEL: Their ability to hear differences should be tested as well as their ability to hear sounds through the whole tonal range. HANLEY: Hearing pure and simple is a matter of natural endowment, not of training. On the other hand the hearing of speech differences involves more than mere hearing. Mind training is needed. For instance, a glottal stop is not ordinarily recognized unless an observer knows what it is, through phonetic training. Moved that a knowledge of phonetics and skill in the use of phonetic transcription be a requirement for field workers. Seconded.

MOORE: All the workers should be together in one place before they are sent out, and tests should be made to determine how nearly alike their transcriptions are. In this way, also, the Director would get information which would be valuable for his later check-up on their material.

RUSSELL: Until Fletcher's system of testing was developed four or five years ago, it was impossible to tell about a person's hearing ability. Therefore, Müller is certainly wrong: it was impossible to know whether Edmont heard correctly. No two persons hear exactly alike, and we do not realize that we have defects. The differences in the field workers' hearing must be ironed out by the Director with the aid of the mechanical records and the hearing scores of the workers.

KNOTT: Three things are necessary in a worker: phonetic training, knowledge of machines, and ability to write shorthand.

HANLEY: It would be hard to find persons skilled in all these things. But surely good hearing and phonetic training are necessary. Besides, one of the two workers on a team must have skill in handling the machine.

KURATH: proposed as prerequisites (1) that the investigator hear well, (2) that he be trained in phonetics, (3) that he have considerable experience in recording the known variations in American speech by means of a phonetic alphabet, (4) that he know the variations in morphology, syntax, and vocabulary, that have already been observed, and (5) that he be trained in a special technique of collecting the material, the technique to be worked out by the Director and approved by the Managing Committee.

CHAIR: Discussion of VI must stop now.

BRYAN: Nothing has been said of the Director's training.

VII. Are class dialects to be treated along with local dialects?
In particular, should standard colloquial speech be carefully dis-
tinguished from illiterate or vulgar speech?

MOORE: moved that an important purpose of the Atlas shall be to distinguish between vulgar and so-called standard speech. No one knows now just what the difference is.

CURME: preferred the word popular to vulgar.

STURTEVANT: queried whether this was a matter of linguistic geography. How are non-geographic variations to be represented on the maps? KENYON: VII is the most difficult question on the list. How are we to get a definition of standard English? We shall have to proceed with that question entirely in the background.

HANLEY: agreed with Kenyon. The workers should get full case histories of their subjects, to make the material collected more useful. The field workers should collect the facts, but should refrain from

classifying and interpreting them.

KURATH: Working definitions of cultivated colloquial speech and of popular speech could be adopted. Case histories would be useful for the classifications of speakers in one or the other group. The difference between the two groups should be represented on the maps; they would not complicate them unduly. In any case, in choosing the subjects, a working definition is required, and it might as well be

laid down in so many words. Scholars using the maps would be in a position to reclassify the forms with the aid of the case histories. MOORE's motion carried: It is the sense of this Conference that an important aim of the proposed investigation is to collect information about and discriminate between the local popular and the local standard colloquial speech.

VIII. Can the plan for immediate execution be simplified by the
temporary omission of any area, such as Canada or the region
west of the Rocky Mountains?

RUSSELL: There should be no elimination, especially of the area west of the Rocky Mountains. Colonization of the Spanish Southwest was very early, and the Spanish influence has spread and continues to grow stronger, unlike that of other foreign language groups. The Mormon group also is interesting linguistically, because of the many shifts of location and the large foreign language influence which they have undergone.

PROKOSCH: The German Atlas excluded, on political grounds, German Austria and Switzerland, though there are no linguistic boundaries between them and Germany. It would be as much of a mistake to exclude Canada in the present project. He was also against the exclusion of the Far West, though scientifically more could be said for that than for excluding Canada. KRAPP: But limitation is necessary.

Moved that work shall start in areas which have relatively clearly defined speech characteristics. CARRUTHERS: supported Prokosch on the absence of a linguistic boundary between the United States and Canada. The only difference between them is that the settlement of Canada began about 100 years later.

KURATH: All the English-speaking parts of North America should be included. Limitation should be in the number of places studied, not in area. It would be better to study 200 communities scattered over the whole area, than 500 in a restricted area. KRAPP'S motion, incorporating Kurath's ideas, was adopted: It is the sense of this Conference that the investigation should begin in those areas known to exhibit distinct dialect characteristics, but that it should ultimately be extended to embrace all English-speaking America.

The session was adjourned at 12.50 P.M.

THE THIRD SESSION was called to order by the Chairman at 2.30 o'clock on Saturday afternoon, August 3, and discussion was at once begun of the lists of dialectal peculiarities which had been prepared and submitted in advance to the members of the Conference.

A. Phonetic features.

STEINBERG: What is the relative importance of vowels and consonants in a study of American dialects? Some means of mechanical recording are deficient in reproducing sibilants.

KURATH: Vowel variations are much more numerous and, in a sense, more important. Most of the consonant variations are of such a nature that even an imperfect instrument could record them. The substitution of ƒ for th by some Negroes, as in thru, may give trouble in recording.

STEINBERG: Will study be given only to the male voice?
KURATH: It is important to study the speech of both sexes.
STEINBERG: The male voice is more easily reproduced.

The

KURATH: That fact would have to be taken into consideration. phonetic variations are listed in the Circular as 'regular' and 'sporadic'. The classification may not be entirely correct. The specimen text should include only the more regular variations, and should be supplemented by other material to show the irregularities in series like log, frog, cog, clog, etc., and roof, root, soot, room, etc. BUCK: No. 51 should include the New England [ben].

HOPKINS: The pronunciation of the pronoun I should be included. (An unidentified speaker added that the stressed and the unstressed forms of I must be distinguished.)

KURATH: said that there is a full list of such variations in his American Pronunciation. Some of these must be studied.

AYRES: objected to the transcription of the second variant in 41. The New York pronunciation is [31]; i.e., curl and coil have about the same pronunciation in New York.

HANLEY: added the 'clear' or 'front' l in Southern Nellie, Alice, etc. Mention was made of w for l, as in bio for bill; KURATH cited crawed for crawled, where the l sound is lost.

KENT: A third variant [rat] is found for root.

It was agreed that any further suggestions on No. 1 should be sent to Kurath.

A Vote for Nine Members of a Managing Committee, the American Council of Learned Societies to choose the Committee as

far as possible on the basis of this vote.

STURTEVANT: moved that the vote be taken by ballot, the ballots not to be counted, but to be turned over to the American Council as they come in. Carried.

KENYON: Moved that the members of the Conference be given until they leave the meeting to fill out and hand in their ballots. Carried.

B. Vocabulary.

KRAPP: The list in the circular needs addition and revision. Suggested that half a dozen experts make out lists of about 300 words. KENYON: The field workers should become very familiar with lists of vocabulary variants such as that in the index to Dialect Notes, since words incidentally spoken may escape the hearer who is not watching for them.

CHAIR: called for method of procedure. Shall the members of the Conference make random suggestions or consider a general principle for the selection of a vocabulary list?

KURATH: It is possible to set up guiding principles. The following groups of words should be represented (they are represented in Krapp's list):

(1) Topographical words, e.g., knoll, knob, butte;

(2) Family words, e.g., for father and mother;

(3) Literary terms, to show the influence of the literary language on colloquial speech;

(4) Newspaper words, to show the influence of the press; and so forth.

KENYON: added names of farm tools, and of the more familiar birds; also calls to domestic animals (these have already been studied), and household words.

KENT: All these groups could not be used in all sections. Some classes would lack words for many farm tools.

KNOTT: Will rural, small village, and urban communities be examined and classified separately?

BLAKE: The 'symbolic' words should be studied: the verb to be, the chief pronouns, adverbs, and prepositions. The entire essential vocabulary, say 1000-2000 words, might be studied in each community.