Hostname: page-component-89b8bd64d-b5k59 Total loading time: 0 Render date: 2026-05-07T11:15:43.368Z Has data issue: false hasContentIssue false
  • English
  • Français

Correlation Methods of Comparing Idiolects in a Transition Area

Published online by Cambridge University Press:  22 April 2026

David W. Reed  and
John L. Spicer
David W. Reed
Affiliation:
University of California
John L. Spicer
Affiliation:
University of Minnesota
Get access Rights & Permissions [Opens in a new window]

Extract

The transition area as a crucial but difficult problem in American linguistic geography has been ably presented in a recent article by Alva L. Davis and Raven I. McDavid Jr. The relative lack of pattern in the data from transition areas, compared to the easily mapped patterns found in relic and focal areas, is probably responsible in part for the slight attention that has been paid to transition areas before the appearance of this article. Davis and McDavid properly conclude that transition areas will assume increased importance as research in American linguistic geography moves away from the numerous focal and relic areas of the Atlantic coast toward the increasingly heterogeneous transition areas of the western United States. Certainly the collection and analysis of dialect vocabulary in California has revealed to the present writers a situation even more varied and complex than that of northwestern Ohio.

Davis and McDavid conclude that it will be necessary to collect much more complete information from transition areas if one is to correlate speech patterns with the historical and cultural complex. They suggest no new methods, however, by which the limited data now available from transition areas may be more adequately analyzed and understood. In the process of analyzing our own material from California, we have come to the conclusion that the speech patterns of traneition areas grow much clearer when viewed as quantitative rather than as qualitative phenomena.

Information

Type
Research Article
Information
Language , Volume 28 , Issue 3-Part1: July-September , September 1952 , pp. 348 - 359
Copyright
Copyright © 1952 Linguistic Society of America

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

Article purchase

Temporarily unavailable

References

1 Northwestern Ohio: A transition area, Lg. 26.265–73 (1950).

2 Cf. Hans Kurath, A word geography of the eastern United States 12–3, 39–40, 47–8, 54–6, figs. 5A, 42, 58, 61, 62, 66 (Ann Arbor, 1949).

3 Jan Czekanowski, Na marginesie recenzji P. K. Moszyńskiego o książce: Wstęp do historji Słowian [On P. K. M.'s marginal criticism of the book: Introduction to the history of the Slavs], Series 2, Vol. 7 (Lud, 1928; reprinted Lwow, 1928). Title cited from Kroeber and Chrétien (op.cit. in fn. 4), as corrected by Chrétien.

4 A. L. Kroeber and C. D. Chrétien, Quantitative classification of the Indo-European languages, Lg. 13.83–103 (1937).

5 Kroeber and Chrétien, The statistical technique and Hittite, Lg. 15.69–71 (1939).

6 Id. 70.

7 The quantitative method for determining linguistic relationships, Univ. Cal. pub. in linguistics 1.11–20 (1943); Culture element distributions: 25. Reliability of statistical procedures and results, Anthropological records 8.469–90 (1945).

8 Quant. method 19. Chrétien now agrees that this conclusion is overcautious, and suggests that the nature of the material may occasionally permit rather small samples.

9 Chrétien's preference for the entire statistical population rather than a random sample is justified only if we wish to determine the absolute value of particular relationships in order to compare them ultimately with other relationships outside the immediate field of investigation. For practical purposes, the linguist wishes simply to order whatever is being related in some fashion that will be meaningful in the immediate context. For us the significant conclusion to be drawn from Chrétien's use of the chi-square test is that in any ordering of data there will be a dead zone around the zero coefficient—a zone within which the ordering is not reliable. The larger the sample, the smaller this dead zone.

10 For general though difficult discussions of the theory and methods or correlation, see R. A. Fisher, Statistical methods for research workers6 (Edinburgh, 1936); G. U. Yule and M. G. Kendall, An introduction to the theory of statistics12 (London, 1940).

11 For a more complete description of method than can be given here, see Kroeber and Chrétien, Quant. classif. 83–5, 98–103; Chrétien, Quant. method 12–6.

12 That is, the number of elements from the list under consideration that are absent from both groups of variates. Another common formula (G) assumes that d is always infinite. Values of V (see below in the text) approach corresponding values of G as an upper limit. The Q6 formula has been favored here because it gives twice as wide a spread of values as the G formula. The relative values of Q6 and G do not differ significantly except when d is very small.

13 V is identical with Kroeber and Chrétien's rhk. V is the symbol more recently employed by statisticians for this value.

14 For a readily understandable account of this device, see Chrétien, Quant. method 17–9.

15 E.g. Yule and Kendall 534–5.

16 Fisher 77, 128–33, 230–8.

17 P = Perrysburg, D = Defiance, O = Ottawa, VW = Van Wert, US = Upper Sandusky. In each community the informant designated by a subscript 1 (P1 etc.) is the more old-fashioned, i.e. has had fewer contacts with speakers from outside his community than the other informant. For a thumbnail sketch of all the informants, see Davis and McDavid 266 fn. 9.

18 The difference between O1 and O2 is partially explained by the fact that the field record for O1 was made by Frederick G. Cassidy in 1939, while all the other records were made in June 1949 by McDavid.

19 McDavid informs us that D1 was the most isolated rural informant interviewed in northwestern Ohio, and by far the oldest.

20 Certain terms in the lists are labeled Northern or Midland by Davis and McDavid. We have added labels to the following, on the basis of indications in Kurath's Word geography (pages here cited): beller, New England (19, 62); sugar grove, w. Pa., n. Va. (36, 76); swingletree, Delaware River, s. N. J. (13, 47, 58); sook, general Midlands w. of the Susquehanna (14, 30, 38, 63); sick in his stomach, s. Pa. (78); ground worm, Chesapeake Bay and e. Pa. (46, 74).

21 As A. H. Marckwardt justly observes, however, this indication is perhaps illusory, since Davis and McDavid greatly simplified the phonetic transcriptions of the field records before constructing their pronunciation table. If they had taken into account the full range of symbols and diacritics used in the field records, pronunciation would probably seem less homogeneous than vocabulary. Phonemic analysis of the speech of each informant would provide the only satisfactory approach to a comparison of pronunciation and vocabulary on the score of homogeneity.

22 The quantitative distribution of cultural similarities and their cartographic representation, transl. by Chrétien and Kroeber, American anthropologist 51.237–52 (1949).

23 Milke's term for these lines is isopleth, a word used with a slightly different meaning in meteorology. Because of the difference in meaning and because the phrase graded area is well established in dialect geography, we have preferred to coin the somewhat barbarous isograde. (The only pure Greek alternatives are isotax and isobath. The former carries an irrelevant suggestion of syntax and taxeme, while the latter seems to have been used as the name of a patented floating inkstand.)

24 Cf. Milke 248.