The frequency of English phonemes as ascertained by counting their occurrence in connected matter is different from that stated in our investigation. In the limits of our material we tried to give an approximate idea of the absolute quantitative charge of phonemes in the present-day English as an important factor in the linguistic system, whereas by counting phonemes in the given texts we could state only their actual frequency in the texts examined. The results of the former computation tend to be of absolute value for the linguistic system, whereas those attained by the other method are necessarily always incomplete and vary according to the choice and extent of the extracts. Whereas the relative frequency of phonemes depends on the actual occurrence of words, their absolute quantitative charge (which may be perhaps better called their productivity) is dependent on the whole number of words and their forms existing in the language. In short, the former method is concerned with the language as the "parole", whereas the latter involves the linguistic analysis from the point of view of the "langue". Both methods are supplementary, but we have adopted the latter (which is, by the way, more difficult than the former), because it seemed evident that reliable results must be first obtained on this side of the language and a safe basis created for comparison with those of the former method, by which the various styles of the same linguistic system might be characterized. Such studies do not exist as yet, and we must be satisfied with confronting our results with some of those of Mr. G. Dewey's computation.

According to Mr. G. Dewey's table 15, the relative frequency of the "simple sounds" of American English as ascertained in 100,000 words of connected matter (not differentiated from the stylistic point of view), would follow in this order of frequency:

n 7.24%, t 7.13%, r 6.88%, s 4.55%, d 4.31%, 13.74%, ₫ 3.43%, z 2.97%, m 2.78%, v 2.28%, w 2.08%, p 2.04%, f1.84%, b1.81%, h 1.81%, 0.82%, g 0.74%, j 0.60%, tfl 0.52%, dz 0.44%, b 0.37%,

0.96%, 3 0.05%.

In the Standard English pronunciation <r> would probably follow between <k> and <w) (2.44), which would precede <v>; <j> (0.88) would follow after <n>, and <h> (1.45%) would have its place between <tf> and (dz >.

As to the vowels, Mr. Dewey's table has many deficiencies from the phonological point of view, so that it may give only an approximately idea of their relative frequency in the written and spoken language:

(a) In stressed syllables:

i 4.11%, æ 3.50%, e 3.44%, o 2.90%, i: 2.12%, ei 1.84%, a 1.70%,

?

u: 1.60%, ai 1.59%, ou 1.30%, ɔ: 1.26%, u 0.69%, a: 0.69%, ǝ: 0.63%, au 0.59%, eǝ 0.23%. Other diphthongs are not counted. (b) In unstressed syllables:

e 4.63%, i 4.42%, ɔ 0.33%, a 0.2%.

In the prevocalic position the order of consonantal phonemes as to their frequency in 1370 syllables occurring oftener than ten times would be as follows:

one:

In the postvocalic position the order of the phonemes is a different

As to the monomorphemic consonantal combinations found in the syllables occurring oftener than ten times, the order of their frequency is as follows:

40 combinations... 6,954 occurrences (out of 80,905).

25 combinations.... 9,490 occurrences (in addition to 3,647 of di-
morphemic combinations).

FOREIGN WORDS.

In analysing the phonological structure of English mono-
morphemic words, an interesting problem presented itself to
the writer: What are the linguistic tests of foreign words
within the system of the present-day language? If the historical
standpoint (from which it has been hitherto treated) is re-
jected, the question may be solved only on the basis of the
phonological analysis, the importance of which for our problem
was duly emphasized by V. Mathesius in one of his Czech
papers.23 He shows that the tests of the foreign character of
words are constituted by differences in their phonological
make-up from that of the native words. Thus two main groups
of words regarded as foreign from the present point of view
may be distinguished, the one containing those which are
entirely assimilated and do not differ in their phonological
structure from the others, and the other characterized by a
number of phonological features placing them on different
degrees of foreign character. It must be emphasized that these
peculiarities have nothing to do, on principle, with a wider, or
narrower, currency of words, because even the words which
are often used may be felt as foreign owing to peculiarities in
their phonological structure, and vice versa, the extremely rare
words may have the usual make-up of the native ones.

23 Cizí slova se stanoviska synchronického. ČMF. XVIII/1932,
231-239; cf. the summaries in French and English on the pp. 361 and 363
of the same volume.

The tests of the foreign character of words in English are as follows:

1. The use of phonemes, which are entirely foreign to English, e. g. the nasal vowels, <ü, l'> and <y> in Modern French loan-words. As a rule, these sounds are substituted by the English acoustically similar phonemes, or their combinations respectively. Cf. 'ku:pon, or 'ku:pɔŋ coupon, 'buljɔn bouillon, bu:lji: bouilli, 'konjæk cognac, læng linge etc.

2. The number of unstressed vocalic phonemes. Almost all monomorphemic words in which the stressed syllable is followed by more than one unstressed are felt to be foreign.

3. The position of the stress. Simple words not stressed on the first stem syllable are of foreign origin. A notable exception is i'levn eleven. The shifting of the stress in words of the same morphological family is also a mark of their foreign descent, e. g. 'foutǝgræf: fou'togrəfi photograph photography. 4. The double stress in monomorphemic words: 'bə:'lin

Berlin.

5. "Full" vowels in unstressed syllables, e. g. indeks index, empaiǝ empire, bu(:)kei bouquet. For windou window, sinju sinew, see p. 17.

6. The sounds, which have not the function of phonemes in genuine words. Middle English [v] and [z], which existed only as combinatory variants of the phonemes <f> and <s> respectively (cf. fären, of:lēven, sunne, rōs: risen), had the phonological function in French loan-words, such as vīne, veil, māson. This mark of differentiation between the words of Teutonic and Old French origin had disappeared in consequence of the phonologization of the voiced open consonants [v, đ, z] due to the loss of the final unstressed -e in Late Middle English. Cf. [g] in Modern Czech.

7. Some special consonantal combinations (cf. pp. 31) and all combinations of vowels in the same morpheme.

8. A different position of phonemes. Middle English intervocalic phonemes <s> and <f> occurred only in words borrowed from French, e. g. offer, mason, basin. In 15th Century English this mark of differentiation was lost owing mainly to the dephonologization of the length of consonant phonemes.

The Modern English <p> between vowels or before a voiced consonant belonging to the same morpheme, is a characteristic mark of the foreign (mostly Greek) origin of the words, e. g. li: pal lethal, o:pe author, (:)'poriti authority, en'þju: z enthuse. (Such words as smiđi, smipi smithy, heldi, -pi healthy, in which <i> belongs to another morpheme, do not contradict our assertion.)