¹ I wish to thank Professor Leonard Bloomfield for his criticism of the manuscript, and for his many pertinent suggestions.

² Of course, in many dialects the [r] in these forms is actualized as a shwa-glide, vocalic length, or zero. But for the general pattern it must be transcribed as a development of the [r] phoneme.

³ Here [ε] represents a vowel of the type found today in English hat, [a] a vowel of the type found in father, and [C] any stop, spirant, affricate, or cluster.

⁴ In Houma, Louisiana, I have heard aunt as [ejnt]. Dr. Samuel McNeely states that this is the standard form in Baton Rouge.

⁵ This phenomenon of an [εj] phoneme distinct from [ε] and [ej] has been noted independently by Mr. J. B. MacMillan for Central Alabama, by Mr. J. W. Watson Jr. for Blacksburg, Virginia, and in my own study of the dialect of Greenville, South Carolina. It probably occurs elsewhere.

⁶ The appearance of aunt as [ejnt] in Louisiana confirms, rather than contradicts, the general theory.

⁷ The feeling, at least in A-D and B-D dialects, that [^|aj ^|ant] is an affectation, probably contributed to the current disfavor.