Principles of Phonetic Notation

The first requisite of a good alphabet is that it should be capable of being written and read with ease and written with moderate quickness.

Simplicity. For ease of reading, it is desirable that the letters should be as simple as is consistent with distinctness. From this point of view, the Roman letters are superior to the black-letter or Gothic forms still used in Germany, as we see especially in the capitals. Dots and other diacritics, which must be made small, tend to indistinctness.

Compactness.-Ease of reading depends also greatly on compactness. Hence syllabic systems of writing like Sanskrit, in which such a syllable as skra is expressed by a single character, are in many respects easier and pleasanter to read than the corresponding Roman transcription. It is often a matter of surprise that the Chinese characters try the eyes so little, in spite of the great complexity and minute distinctions they often involve. The reason is that every word is represented by a compact square character, all the characters being of uniform size, the strain on the eyesight being further reduced by the arrangement of the characters in perpendicular columns. The superiority of the syllabic principle is strikingly shown by the fact that both the Protestant and the Catholic missionaries in Canada use syllabic alphabets in teaching the Crees and other native tribes to read, on account of the length to which the words run when written in Roman letters. These alphabets consist of simple characters expressing consonants, such as V, turned different ways-<>-to indicate what vowel follows.

Joining.-Ease and quickness of writing require that the letters should be easily joined together, as may be seen by comparing a passage written in Greek letters with one in Roman letters.

The most accurate way of estimating the comparative merits of letters as regards ease and quickness of writing is to count the number of strokes of which they are composed on some uniform plan. Thus i without a dot consists of one stroke, scrpit s of two, š of four.

But this method of calculation leaves out of account the ' aërial movements' of the pen from the line of writing to the diacritic and back again. We see now that writing the single letter š takes as much time as writing the five letters seeee!

Printed Forms.-In printing, the complexity of the letters does not influence speed or ease: the main thing is to have as few types as possible. This is an additional reason for abolishing the use of capitals in phonetic writing-except for special distinctions. In printing it is easiest to have the letters detached. This is highly objectionable in writing, but is generally an advantage in reading.

As regards the relations between the written and printed forms of the letters, it is evidently desirable to avoid unnecessary deviation without, on the other hand, attempting to make print into a-necessarily imperfect-imitation of handwriting. The disadvantages of such an attempt are well shown in Arabic, with its superfluous distinction of initial, medial, and final forms of one and the same letter, the maximum of discomfort being reached when the short vowels are indicated by diacritic strokes printed on separate lines, so that the reader is sometimes in doubt whether the diacritic is to be read above the consonant of the line he is reading or below the consonant of the line above.

Some phonetic transcriptions-such as that of the Swedish Dialect Society and of Trautmann in his Sprachlaute-consist entirely of italics, so as to diminish the difference between the written and printed characters as much as possible, and also to make the phonetic writing stand out distinctly in a page of Roman type. But as italics are required for a variety of other purposes, and as it is a waste of existing material not to utilize the distinction of Roman and italic, it seems better to make the more legible Roman the basis, and use italics for various supplementary purposes; it is always easy to mark off phonetic writing by enclosing it in ( ). The transcription of the Danish Dialect Society Dania is so far an advance on the other italic systems that it utilizes Roman letters for special distinctions of sound.

Having thus determined the general principles on which the choice of symbols is founded, we come to the still more difficult question, how to use these symbols-what sounds or what phonetic functions to assign to them.

National and International Basis

The most obvious way of making an unphonetic orthography phonetic is to select some one out of the various traditional representations of each sound, and use that one symbol exclusively, omitting at the same time all silent letters, and adding marks of stress (accent) if necessary, as in the following specimen of Ellis's 'English Glossic:'

'Ingglish Glosik iz veri eezi too reed. A cheild foar yeerz oald kan bee taut too reed Glosik buoks.'

A system which, like Glossic, writes short and long vowels with totally different symbols (i, ee) is only half-phonetic: it is phonetic on an unphonetic basis. Again, this unphonetic English basis breaks down altogether in some cases. It fails, for instance, to supply unambiguous symbols for the vowels in child and book, full and the consonant in the, which Ellis writes (dh).

The following specimens of French and German spellings formed in a similar way on the basis of the respective nomic orthographies of these languages are taken from Soames's Introduction to the Study of Phonetics :

'Deû pti garson d la vil, Richa:r é Gusta:v, s égarè:r eun jou:r danz un épè:s foré.

'As 'ist doch gevis, das 'in der Vält den Mänshen nicts nohtvändię macht 'als dih Lihbe.'

A fully phonetic system, in which long vowels and diphthongs are expressed by consistent modifications or combinations of the simple vowel symbols, and in which simple sounds are, as far as is reasonable and convenient, expressed by single letters instead of diagraphs, must necessarily discard any one national traditional basis. The best basis on the whole is obtained by making the later Latin pronunciation the foundation, with such modifications and additions as may be necessary. We thus get the 'Romic' or international as opposed to the Glossic or national basis. Thus the passage quoted above appears as follows in my 'Broad Romic' notation:

'inglif glosik iz veri iizi tə riid. ǝ tfaild fɔə jiǝəz ould kən bi tot tǝ riid glosik buks.'

Observe that on this basis the vowel in the English book, French jour, and German gut would be expressed uniformly by (u) in writing all three languages (buk, zuur, guut) instead of in three different ways, as on the Glossic basis.

It is evident that as soon as we have to deal with more than one language there can be no doubt of the superiority of the Romic basis.

A Universal Alphabet Unpractical

If a universal alphabet were constructed which provided symbols for every possible sound, then each language would simply have to select from it the symbols required for its own sound-system. On the other hand, it is desirable for ordinary practical purposes that each language should utilize the simplest and most convenient letters. Thus, if in the universal alphabet (e) were restricted to the close sound of French é, the corresponding open sound being represented by (E), this arrangement would suit French very well. But if it were applied to English, which has not any close (e) at all, the result would be that the simplest and easiest to write of all letters would not be used at all.

Significant Sound-distinctions

Again, for practical purposes we have to distinguish between differences of sound on which differences of meaning dependsignificant sound-distinctions-from those which are not significant. Thus the distinction between (e) and (e) is significant in French, as in pécher, pêcher; but in those languages in which the short e is always open and the long e always close there is no necessity to employ (E) at all: the distinction of quantity in (e, ee) is enough. Even if the distinction of close and open is made in the long e, there can be no ambiguity in writing e for the short sound if it is always open, as in German and English, in both of which languages such a spelling as (men) is perfectly unambiguous.

So also the distinction between the first elements of the English diphthongs in high, how is un-significant, and although neither of them is identical with the vowel of ask, we do not hesitate to write all three uniformly with a-(hai, hau, aask). And as the pronunciation of these diphthongs varies considerably, and as it would be impossible to do justice to all these minute distinctions without a much more elaborate system of notation than is required for ordinary practical purposes, we regard (ai, au) simply as general symbols for a variety of

diphthongs, all of which may be classed under one of two distinct types, both beginning with back or mixed non-rounded vowels and ending with approximations to (i) and (u) respectively.

Superfluous Sound-distinctions

This is connected with another common-sense principle, namely, that of omitting superfluous distinctions. Thus, if a language always has the stress on the first syllable, the stress does not require to be marked at all. If the majority of words have the stress on the first syllable, then it is necessary to mark it only when it falls on some other syllable. It is evident that on this principle the 'smooth breathing' in Greek ought to be omitted, as there are only two breathings, and the absence of the rough breathing is enough to show that the other one is meant. In English it is necessary to distinguish the long open o in naught from the short open o in not, which we ought strictly to do by writing (nɔɔt, nɔt). But as there is no short close o in English, there is no reason why we should not write not with the easier o. Hence it becomes superfluous to mark the length in naught, which finally brings us to (not, not) as the shortest and most convenient phonetic spellings.

Modifiable General Basis

We see, then, that the ideal of a general alphabet for practical purposes is one which gives a basis which is, on the whole, generally acceptable, but can be freely modified to suit the requirements of each language. The better the basis, the less inducement there will be to diverge from it.

If we accept certain mechanical principles, such as utilizing c, x, and the other superfluous letters, avoiding diacritics, testing new letters with regard to their distinctness and ease of writing, and return where practicable to the original Roman values, we shall have little difficulty in arriving at a basis of agreement. No one, for instance, who has given any thought to general principles could hesitate long between ü and y, š and f.

In comparing the sounds of a variety of languages-still more in dealing with sounds generally-we require a much more elaborate system of notation than in dealing with a single language; we can no longer content ourselves with marking