Phonemes and Phonological Units*

Josef Vachek

The fourth volume of the present Travaux contains the well known “Projet de terminologie phonologique standardisée” (pp. 309—322). Some of its introductory definitions will have our attention. They run as follows:

“Phonème. . . — Unité phonologique non susceptible d’être dissociée en unités phonologiques plus petites et plus simples “Unité phonologique. . . — Terme d’une opposition phonologique quelconque.
“Opposition phonologique. . . — Différence phonique susceptible de servir, dans une langue donnée, à la différenciation des significations intellectuelles.

The definitions just quoted have so far been accepted without any discussion.¹ And yet, what has been dealt with by the phonologists as “phonemes” and “phonological units” in their practice, i.e. in phonological descriptions of various languages and idioms, does not correspond to what is implied by the definitions of the Projet, if they are carefully considered and consistently developed.

In proving this, we shall at first mutually confront the definitions as they have been quoted above.

The “phonological unit” is said to be the term of any phonological opposition. But the concept of the “phonological opposition” appears to be defined in the Projet more broadly than is usually realized by linguists. It covers namely not only oppositions as 1 : r in English low : row but also cases like bl- : gr- (in blow: grow), and even oppositions like hiɘr- : pɘteit (hero : potato), feim- : konšienš- (famous : conscientious): in all of the mentioned pairs of words we are faced with phonic differences responsible for the differences of intellectual meanings in English. A consistent application of the definitions contained in the Projet must, then, necessarily lead to the acceptance of bl-, gr-, hiɘr-, and even pɘteit-, feim-, and konšienš- as “phonological units”. None of the phonologists, as far as I know, has ever used the term in this broad sense. It is not surprising, because the term, as defined in the Projet, is so broad as to render it useless.

As a consequence of this, the phonologists either avoided the term “phonological unit” altogether or used it in a different meaning, ignoring, or failing to notice, the inconsistency of such a use with the definitions of the Projet. As a representative of the latter group of linguists, we may mention V. Skalička,² even if the term he uses is not “phonological units” but “phonological elements”. As far as his clever but rather fragmentary expositions can be followed, phonological units are understood to be the smallest phonic items in language, capable of differentiating intellectual meanings.

I should like to show that the new significance of the term “unité phonologique” requires a slight, but a substantial, change in the definitions of the Projet.

We have seen that the definition of the “phonological unit” in the Projet is too broad; and the same applies to the definition of the “phonological opposition” as it appears there. That is why both definitions cannot, in the wording of the Projet, be of any use in phonological practice. The phonological unit in the revised sense is not a term of any phonological opposition but of what maybe called a simple phonological opposition. This, again, should be defined as a minimum phonic difference responsible for the difference of intellectual meanings.

It has been found of use, then, to distinguish the broad concept of the phonological opposition in general from the narrower, subordinate concept of the simple phonological opposition, as it is only the latter (subordinate) concept which may lead us to an adequate conception (and definition) of the phonological unit.

Moreover, it will not be entirely without value to establish and to define another concept, viz. that of a complex phonological opposition. It is coordinate to the simple phonological opposition, and subordinate (again) to the phonological opposition in general:

It may be defined as a non-minimum phonic difference responsible, in a given language, for differences of intellectual meanings.³ Further consideration will show that the establishing of the new concept is not due to the hair-splitting of a logician but that it is of high importance for the basic phonological problem — the definition of the phoneme.

To return to the definition of the phonological unit, as adapted in the above lines, it is hardly necessary to dwell upon the fact that phonological units, being terms of minimum phonic differences, must take part in their minimum differential character. To put it differently, a phonological unit must be indivisible into smaller phonological units.

This requirement is in entire agreement with the character of Skalička’s “phonological elements”. Skaliika, however, hesitates to draw all the necessary conclusions from the character. The problems we are facing now may be expressed in the following way.

The definition of the phonological unit, as adapted here, covers exactly the same field as the definition of the phoneme as contained in the Projet. Does this mean that one of the two terms appears superfluous? Or does, perhaps, the definition of phoneme, as styled in the Projet, also need rectifying?

Linguistic facts give ample evidence that neither of the two terms discussed is superfluous. Both the phoneme and the phonological unit represent distinctly different concepts, even if in many cases individual phonemes equal individual phonological units. To illustrate this statement by some examples, let us again recall several cases of phonological oppositions. In a pair of words like glow : grow the terms of the difference are undoubtedly l. : r. But in a pair like bad : pad the difference cannot be stated as b : p, if it is to be a minimum difference. The minimum difference here is “sonority: O”, as b, not only phonetically but also phonologically, is composed of an archiphoneme p + the mark of correlation (in this case, sonority). In other words, it is only the presence of sonority, as opposed to its absence, which is responsible for the difference of meanings of the two forms, bad and pad. Obviously, it is only the sonority which, being indivisible into further phonological units, may claim the character of the phoneme, as defined by the Projet. In fact, however, no phonologist has ever used the term phoneme in case of this kind: b and p as wholes have been termed phonemes, and the sonority, as a mark of correlation, has been, so to speak, included within the frame of the b-phoneme. Linguistic practice here goes the right way in opposition — even if it is subconscious — both to the definitions of the Projet, and to the witty and concise but not precise statement of Karl Bűhler, who denotes phonemes as diacritica of morphemes (see the fourth volume of the present Travaux, p. 295). It is evident that in cases like ours only the sonority of b is the diacriticon (as has already been emphasized by Skalička in his above-mentioned paper). A phoneme like b, consequently, contains two phonological units — the p-unit and the sonority unit. Even more phonological units, however, can be united in a phoneme, as, e.g., in the Russian b’-phoneme, the units concerned being p +; sonority + palatal character. Inversely, there exist phonemes containing one phonological unit only (such as, e. g., p, l, r in English).

The conclusion to be drawn from the above considerations is that the definition of the phoneme as given in the Projet is inadequate and needs rectifying. And it is in this point that Skalička hesitates to go further. Upon the whole, he may be said to have clearly recognized the difference between phonemes and phonological units but tries to reconcile the fact with the definition of the phoneme in the Projet. He states (p. 133 of the above- mentioned paper) that “the archiphoneme together with the mark of correlation constitutes the lowest, further indivisible unit. The correlative relation of two (or several pairs of) phonemes is considered as their private affair which has no bearing on actual speech”. To this statement more objections can be raised. Firstly, that the correlative relation very often has a bearing upon actual speech (as is shown by the alternation of correlative phonemes due to particular groupings of phonemes — that is, to actual speech — such as in English iš ši ˂ iš ši ‘is she?’). Secondly, that the definitions of the Projet are not established inductively from actual speech but proceed, in a deductive way, from the fact that there exists such things as significant (or signal-like) oppositions — the basic assumption without which no language (not only the spoken language) is possible — and that, therefore, the criticism of these definitions must keep to the basis of the Projet and its terms and concepts, and not shift the problem to another domain, viz. that of actual speech. (Even there, however, the bearing of the correlative relation may be traced, as has been pointed out above. )

The necessity of an amendment of the phoneme definition as given in the Projet is thus obvious. How can it be effected?

What we certainly know is that the phoneme equals the phonological unit so long as the unit remains single; as soon as there are two or more simultaneous phonological units, they all enter the same, and one single phoneme. Two or more successive phonological units, on the other hand, are equal to as many (two or more) phonemes. To put it differently: there can, and do, exist simultaneous phonological units but there cannot occur simultaneous phonemes.⁴

It appears that the clear delimitation of our two terms, the phoneme and the phonological unit, can be done only in terms of time. The phoneme might be defined as that part of a word which cannot be divided into successive phonological units. This, however, would imply the introduction of a new term into the sphere of concepts employed by the Projet. And we may indeed avoid that introduction by making use of the term of complex phonological opposition established in the above paragraphs of this paper.

The final wording of the definition will then run as follows:

The phoneme is a part of a term of a complex phonological opposition which is sometimes divisible into simultaneous, but never into successive, phonological units.

This definition is by no means contradictory to my previous definition given in American Speech X/l935 (p. 250), where the phoneme is stated to be “a signal-like counter of the language which becomes manifested in actual speech by means of (two or more) sounds which are (1) related in character, and (2) mutually exclusive as to their phonic surroundings; all exceptions to (2) must be accounted for on morphematic grounds only”, ⁵ Both definitions refer to one and the same thing and both have their advantages and disadvantages. The definition quoted in this paragraph has the unquestionable advantage of being clearer and more workable than the definition employing the terms of the Projet. The advantage of the latter is that it can afford the omission of the phonetic term “sound” and the grammatical term “morpheme”. But this very advantage becomes a disadvantage if the definition is applied in phonological practice, i. e. in phonological interpretation of a given language. In this most urgent task of phonological research my definition given in American Speech will be found much more useful.

In conclusion I shall try to give the French wording of the definitions corrected or established in this paper:

Opposition phonologique simple: Différence phonique minimum susceptible de servir, dans une langue donnée, à la différenciation des significations intellectuelles.
Opposition phonologique complexe: Différence phonique non-minimum susceptible. . . (as above).
Unité phonologique: Terme d’une opposition phonologique simple.
Phonème: Partie d’un terme d’une opposition phonologique complexe, découpable parfois en unités phonologiques simultanées, mais jamais en unités phonologiques successives.

The definitions given suggest some interesting consequences both for phonological theory and practice. Their discussion, however, must be left to another occasion.

Notes

1. As the latest spe ciment of an approval of the definitions, see N. S. Troubetzkoy’s paper »Essai d’une théorie des oppositions phonologiques« in Journal de Psychologie normale et pathologique XXXIII/1936, pp. 1 —18 (see esp. pp. 9 and 10), where they are only slightly modified (»cette distinction n’apporte aucun changement essentiel«).

2. See his paper on Problems of phonological oppositions written in Czech (K otázkám fonologickych protikladû. Listy filologické LXIII, 1936, pp. 133—139).

3. This concept, as is easily seen, covers the above oppositions bl- : gr-, hiɘr- : pɘteit-; etc., but not such as l : r, which fall within the scope of the simple phonological opposition.

4. Cp. also Skalička p. 133; he, howewver, does not fully realise the of the fact.

5. In principle, the wording of the definition dates from 1932; it may be found in my paper on »Prof. Daniel Jones and the Phoneme« in Charisteria Gu. Mathesio. . . oblata (Pragae 1932), pp. 25-33.

*From Travaux du Cercle Linguistique de Prague, VI: 235-239 (1936)