¹ H. Sweet, Primer of phonetics,³ Oxford, 1906; D. Jones, An English pronouncing dictionary, London, 1917 (second edition, 1926, not accessible to me); Outline of English phonetics,³ Leipzig, 1932.

² See the map by H. Kurath, Lang. 5.159 (1929). Statements about CW in Maître phonétique 3.5.40 (1927) and in L. Bloomfield, Language, New York, 1933.

³ Any transcription shocks and offends all but the few readers who have been inured to the free use of graphic symbols ('algebra'). When old-established renderings, such as [det] debt or [kam] calm, are denounced as dangerous innovations, the critics' choice of examples may perhaps give us a clue to the real difficulty: can it be that the disconcerting factor is really the absence of the letters b and l? Unaccustomed use of the symbol [o] is especially annoying, perhaps because this letter, whose shape resembles the shape of the lips in the utterance of its name, plays a dominant rôle in our first learning of the alphabet and retains this rôle in the graphic fetishism of later life.

The shapes of the graphic symbols scarcely deserve discussion. The reader who prefers the symbol [æ] where I use [ε] does not need any factual basis to justify his preference. Even the reader who claims that the symbol [o] for the vowel of son, mother, come, love is repugnant to the habits of English-speaking people, and therefore prefers to use a v upside down, may be guilty of several fallacies, but he is not tampering with the facts. If he replaces our [o] by this symbol everywhere except in digraph [ow], he will have the lay-out that is traditional in the transcription of B. But the distribution of the symbols is another matter. The theorist whose ratiocinations lead him to demand one and the same symbol for the vowels of calm, psalm and of cam, Sam, or to replace the symbol [č], as in catch it, by the symbols [t] plus [š] or the equivalent, will end with a sorry mess on his hands.

⁴ See my Language 132 ff.

⁵ The plural of sixth is in CW usually [siks] sixths; here the suffix [s] is added with modification, and the form belongs to Type 2 twice over.

⁶ It produces such distinctions as a name : an aim; that stuff : that's tough; see D. Jones in Maître phonétique 3.9.60 (1931).

⁷ For the majority of CW speakers, who have not the distinction of [a] and [α], we transcribe pa, hot [pa, hat]; the question as to the digraphs remains as above, except that if the second alternative is chosen, the symbol [α] in high, how occurs only in these digraphs.

⁸ The customary alternative statement is this: The phonemes [i, e, a, ↄ] occur before [j], as in bee, bay, buy, boy, and the phonemes [u, o, a] occur before [w], as in do, go, now, but in these combinations both the syllabic and the following [j, w] deviate acoustically from their usual shape, especially before [r], as in fear, fair, fire, poor, pour, sour. This statement is equivalent to the one in the text above, and preferable in the way of brevity and clearness, but it annoys some students.

The acoustic analysis of [juw] is so obvious that one hesitates to call it a phoneme. However, it appears only within morphologic units, and it alone enables anything like [j] to occur after initial consonants and clusters, as in few, skew. The only exception to the latter statement seems to be the variant [‘pjεnow] beside [pi’ jεnow] piano; the derivative [‘pjεnist] beside [pi'jεnist] pianist seems to offer difficulty, for a less urbane [‘pijnist] is perhaps the prevalent form.

⁹ In view of my complex linguistic background, both personal and professional, I have tried to describe not my own speech, but that of the people about me. So far as phonemic differences are concerned, this seems to offer little difficulty. As to sub-phonemic features, however, the situation is more complex, and seems to be somewhat as follows : (1) One does not ordinarily react to sub-phonemic differences. (2) When a sub-phonemic difference is striking (as, in a speaker from a distant part of the area) or has been marked out by some special circumstance (such as phonetic training of the observer), it receives a regional or social connotation, and after that: (a) A sub-phonemic variation which the observer himself uses will generally escape his notice; if he tries to observe it, he will suspect himself (rightly or wrongly) of attributing it to speakers who do not use it, and will end in uncertainty. This is my experience with the S forms, (b) A sub-phonemic variation foreign to the observer's own habit is noticeable and seems either provincial (e.g., for me, the backing of [ε], as in hat, southern) or over-refined (e.g., for me, excessive diphthongization of [ow]) or else clumsy (e.g., for me, the use of an S vowel in the word fork).

¹⁰ The late C. W. Alvord insisted upon [ε], not [ↄ], in his name.

¹¹ The B form of this word is unknown to me; B has [α] in daft, and NED gives [α] for the dialectal daff ‘simpleton’.

¹² The word sauce ‘gravy’ has [ↄ], corresponding to B's[ↄ:]; saucy has [ↄ] but also a homely by-form with [ε].

¹³ A mock-B form of fancy with [a] and, in grotesque exaggeration, with [ↄ], is used in CW as a witticism. B, of course, has [æ].

¹⁴ A ‘compromise vowel’ in the forms that are here starred would remove words like pass, glass, grass from the [ε] group of gas, bass, lass. Hence the ‘compromise vowel’ in such types of Standard English is not part of the [ε] phoneme. If one may judge by the forms listed in Jones' Outline 74, the ‘compromise vowel’ is part of the [α:] phoneme, with distribution dependent upon the following consonant.

¹⁵ In yacht, where B has [ↄ], the CW speakers who distinguish [α] and [a] seem to use [a].

¹⁶ The rapid cultural advance, during the last fifty years, of the CW area involves universal schooling by semi-educated teachers, and wide reading; hence there occur elegant reading forms which are foreign to the best educated persons. Thus, the normal CW form of valet is ['vεlej], and the educated speaker uses either this or the B form ['vεlet], but there is also an elegant middle-class form [ve'let], due perhaps to a confused theoretical knowledge of the B form. To this type belong such eccentric forms as ['fεčizm] fascism, ['lεveljej] lavallière, ['šejz 'lↄⁿž] chaise longue.

¹⁷ The B variant of oral with [ow], given in NED, occurs also in American English.

¹⁸ The B variant of docile with [ow] would sound striking in CW.

¹⁹ Similarly, unstressed syllabic [r̩] is marked as syllabic by its surroundings, as in better ['betr], or by a slight stress, as in error, pattern, pickerel ['err, 'pεtrn, 'pikr̩l̩].

²⁰ Can forms like these be the factual basis for the superstition, common among B speakers, that CW uses [r] for sequences of vowel plus [r], as in America?

²¹ In either, neither CW has [ij]; the forms with [aj] are studied elegancies.