Phonology & Orthography:
1. Determining the choices

DP (Design Principles ) 1, 2, 3, 4, 5, 8, 11.

[Choices for a language designer] [The Briefscript project choices] [Design Principle #3] [Design Principle #11] [Conclusion]

This, the first of the pages on 'Phonology & Orthography, discusses the general choices available to a designer of a constructed language and then looks more closely at the available choices for Piashi.

Choices for a language designer

In his essay on phonology (one of eleven "Essays on Language Design") Rick Morneau has listed five possible choices for designers of constructed languages: a designer may choose phonemes:

that the designer is familiar with;
that appeal aeshetically to the designer;
that will maximize the phonemic inventory;
that most people in the world already know or can learn to pronounce easily;
based on morphological, syntactic, semantic or other requirements.

Choice #I was the choice of Reginald Dutton and, indeed, seems generally to have been the choice of other designers of IALs such as Volapük, Esperanto, Ido, Interglossa, Glosa, Intal and others. The designers started with what they knew, eliminated one or two sounds which they deemed 'difficult', and occasionally added one or two other 'easy' sounds. We shall disregard it as it is essentially a personal preference and may well not coincide with DP 11.

Similarly, we shall also ignore Choice #II, which is often the choice of those who design 'artistic' constructed languages, either for their own private use or for use in fiction (e.g. Quenya and Sindarin). It is clearly too subjective a choice for our purposes.

If our aim were only to maximize brevity or conciseness, then Choice #III would be the obvious one. This indeed is exactly what we do find in R. Skrikanth's constructed language, Lin. The language uses upper and lower case symbols as separate letters, together with digits, symbols such #, +, =, / as well as various brackets; they are all given phonetic realization. This does indeed produce a very concisely written language; but the large phonemic inventory necessarily contains many sounds that are fairly rare and that many will find difficult to pronounce. It conflicts with DP 11. We cannot, therefore, simply opt for Choice #III. But we shall have to bear in mind that we may need as large an inventory as possible which does not seriously conflict with Choice #IV.

Choice #IV is our Design Principle 11. This must be our main choice.

Choice #V is also relevant in that DP 1 limits morphemes to one, two or three letters and DP 3 requires morphemes to be self-segregating.

The Briefscript project choices

As the Briefscript project has developed over the years it has evolved three possible systems which are desribed on other pages. They are:

The Classical Briefscript (CBS) system that was first developed in the early 1960s and, although modified in recent years, has remained the same in its essentials.
The Roman letter syllabary (RLS) system that was first mooted in the late 1950s and has been revived in recent years.
The alternative Roman letter syllabary (ARS) system that has been developed only between 2000 and 2004.

ARS, although interesting and certainly conforming to DP 11, uses both traditional consonant and traditional vowels symbols in unusual ways and its rules for determining high and low vowel sounds as well as front ~ back vowel harmony are probably more complicated than is desirable in a potential IAL. It does not meet DP 5 as well as the other two systems do since, although it will contain no exceptions to its rules, there are rather more than just one or two simple rules. With some regret, therefore, I feel this system must be disregarded and that our choice is between CBS and RLS.

Of these two systems, both fully meet DPs 1, 2, 4 and 5. RLS unquestionably meets DP 8 and, although CBS cannot generate so many possible three, two and one letter morphemes as RLS, it arguably generates more than sufficient to meet DP 8. This leaves us DPs 3 and 11 to consider.

DP 3: It should be clear how morphemes are divided; the learner should not be presented with words that could be analyzed in more than one way

CBS fully meets DP 3 as in any string that is generated, there is never any ambiguity regarding the division of morphemes; this applies both in the written form and in the spoken form.

RLS, however, is less successful at meeting this DP. In writing, segration of morphemes is achieved not only by using white space but also using hyphens and/or dot notation. In the spoken form it would be difficult and, without artificial gaps in the manner of Lojban, it would in fact be impossible to achieve without some degree of ambiguity. For example, if we stress the first syllable of three letter morphemes, how is one to differentiate in speaking between "btr" /'putalu/ and "bt r" /'puta lu/? Or if the stress is on the final syllable, how in speech do we distinguish between "btr" /puta'lu/ and "b tr" /pu ta'lu/? Or if we adopted the penultimate stress of Esperanto, then we have similar problems distinguishing in speech between "btr" /pu'talu/ and "b tr" /pu 'talu/.

Whatever stress system is used for polysyllabic words, it will be found that in order to o achieve non-ambuguity in speech in RLS we would have to have:

either all lexical morphemes of two letters (two syllables) and 28 one letter (one syllable) functional morphemes. This gives a limit of 784 lexical morphemes. Admittedly, this is more than Speedwords, but such a restricted number will inevitablt lead to the use of compounds which are likely to prove longer than equivalent words in many natural languages.
or no one letter (one syllable) morphemes, restricting our morphemes to two and three letter ones.

Neither of these solutions are desirable in a briefscript.

DP 11: Phonemes should be restricted to those that most people in the world already know or can learn to pronounce easily

In Chapter XII 'Language Planning for a New Order' of the 'Loom of Language' (1944), Frederick Bodmer wrote: " A five-fold battery of vowels, with values roughly like those of Italian and Spanish a, e, i, o, u, suffices for many speech communities. Several of our own consonants are phonetic rarities, and many varieties of human speech reject the voiceless series in favour of the voiced, or vice versa. A battery of consonants with very wide currency would not include more than nine items - l, m, n, r, together with a choice between the series p, t, f, k, s, and the series b, d, v, g, z. Even this would be a liberal allowance. The Japanese have no l."

It would indeed be a liberal allowance since quite a few languages do not, in fact, distinguish between /r/ and /l/. I may not agree entirely in the fine detail with Frederick Bodmer, but I am in full agreement with the sentiment of what he wrote.

RLS (and also ARS) meet this principle very well. It has a restricted range of phonemes, which are common to very many languages; also because of its limited range, a fair tolerance can be allowed in their pronunciation without causing lack of comprehension.

CBS, however, is very much a compromise and while it does not contain rarities like /θ/, it does includes both voiced and voiceless series as well as both /r/ and /l/. Indeed, on the CBS page I discuss the sounds that I am unhappy about.

Conclusion

So CBS fully meets PD 3 but falls down on PD 11, while RLS fully meets DP 11 but falls down DP 3. Could we not have a system that combines the best elements of both?

In this connexion, it is worth noting that all the spoken CV syllables in the Babm syllabary have a short vowel. The single vowels themselves are long. In defining the structure of nouns, Okamoto wrote: "Four letters make up a noun as a rule, commencing with a short-sound letter .... the second or third is a long-sound letter, which is to be pronounced long and strong with accent ... .".

So, for example, Babm is pronounced /bɔ'ɑːbɔmu/.

It may be inferred even from that small extract from Okamoto's book that Babm is a 'classificational' type of AIL where letters repesent categories and subcategories of meaning like the "philosophocal" languages of 17th century scholars or the more recent example of Edward P. Foster's 'Ro' of the early 20th century. Piashi will not be that type of language, but we can use the idea that our morphemes written CVC begin with "a short-sound letter" and the written V is a "a long-sound letter, which is to be pronounced long with strong accent". In other words our language can have the morpheme structure of CBS combined with a restricted phonemic inventory similar to RLS..

We cannot, however, just use RLS as it is, since the vowel symbols a, e, i, o, u, w, y, need to be used as they are used in CBS. Also, it would be better if we did not have recourse to the inclusion of Greek letters or very unusual use of Roman letters such as h = /nu/. The orthography and phonology adopted for Piashi is given on the second of the pages on 'Phonology & Orthography'.