¹ See Charles H. Grandgent, From Franklin to Lowell: A century of New England pronunciation, PMLA 14.207–39 (1899).

² Edited by Hans Kurath, Bernard Bloch, and others; 6 vols. (Providence, 1939).

³ Walter S. Avis, The mid-back vowels in the English of the eastern United States: A detailed investigation of regional and social differences in phonic characteristics and in phonemic organization (University of Michigan, 1955: unpublished doctoral dissertation, available in microfilm).

⁴ The area referred to is substantially that shown as Area 1 among the speech areas of the Atlantic States in Hans Kurath and Raven I. McDavid Jr., The pronunciation of English in the Atlantic States, Map 2 (Ann Arbor, 1961). See the line of demarcation drawn on Figure 1 below.

⁵ See Hans Kurath, Bernard Bloch, Marcus L. Hansen, and Julia Bloch, Handbook of the linguistic geography of New England 52–3 (Providence, 1939; second printing 1954).

⁶ See Avis 106 ff.

⁷ For a full description of the diaphones recorded, see Avis 81 ff. For a succinct discussion of free and checked vowels, see Kurath and McDavid 3–5.

⁸ See Avis 92 (Table 14) for a detailed listing of such contrasts.

⁹ The unitary interpretation of syllables adopted here has most recently been defended in Kurath and McDavid 3–5. See also Avis 7–8.

¹⁰ See Avis 94 (Table 15) for a statement of the proportional occurrences of these allophones in 32 words, including compounds.

¹¹ Informant 160.2 in Billerica ‘uses [θ] at home in such words as coat, broken, but avoids it in public speaking’ (fieldworker's comment).

¹² It appears that /gθnə/ in the context am I going to ... occurs surprisingly seldom among a wide variety of pronunciations.

¹³ The following maps in Kurath and McDavid illustrate other words discussed in this article: 20 ago; 30 coat, whole, and stone (wall); 121 ewe /jo/; 123 home; 124 loam; 125 won't; 126–8 yolk.

¹⁴ See E. Bagby Atwood, A survey of verb forms in the eastern United States Figure 20 (Ann Arbor, 1953).

¹⁵ Atwood 25.

¹⁶ Informants for the Linguistic atlas are identified by a decimal number, in which the part before the decimal point denotes the community and the part after it the individual informant. They are characterized as follows: Type 1, persons with little or no formal education, little reading, and restricted social contacts; Type 2, persons with better education (usually through high school), wider reading, and more varied contacts; Type 3, persons of cultivated background with superior education (usually through college), wide reading, and extensive contacts; Type A, old, or regarded by the fieldworker as relatively old-fashioned; Type B, young, or regarded as relatively modern. Thus the designation 1A refers to an old man with little or no formal schooling, 3B to a relatively young man with considerable education, and so on. See Kurath, Handbook 41–2.

¹⁷ See Kurath, Handbook, ch. 3, for an outline of New England settlement history; references to settlement history in this part of the discussion are based on that outline and on the detailed accounts of communities and informants in ch. 4.

¹⁸ Kurath, Handbook 34 (Chart 16).

¹⁹ See Avis 133–8 (Tables 20–3) for the data on the dissemination of /θ/ and /o/ by informant types in the northeastern region.

²⁰ For a detailed treatment of the evidence for New England outside the Northeastern region, see Avis 142–66.

²¹ The maps referred to in fn. 13 above illustrate the dissemination of /θ/ in this settlement area. See also Avis, Figures 54–61 and Tables 33–5.

²² Area 4, Map 2 in Kurath and McDavid.

²³ It must be remembered that the Atlas records here under examination were made almost thirty years ago; many of the informants are now dead. In view of the recessive character of /θ/ in the 1930's, it seems improbable that the feature is still common now. If someone were to record comparable data for /θ/ and /o/ today, the extent of drift in New England would almost certainly be even more remarkable than is shown by this study.