Despite common belief, there are pragmatically interesting phenomena in electronically mediated communication (EMC) deep below the level of meaning, syntax or discourse. One such phenomenon is that users of non-Roman alphabets must decide how to represent their spelling in EMC. While there are transliterating schemes for most such languages, many e-mail users may find them too inaccessible, complex and difficult for everyday use, and switching to English as lingua franca is not an option open to monolingual speakers. In this situation (quasi-)phonetic transcription, although it tends to be erratic and requires a dramatic change of metalinguistic perspective, appears to be a viable suggestion.

Polish spelling is based on the Roman alphabet but nine letters carry 'accents' (diacritics) which do not transmit over electronic networks. Only one of these symbols appears among the upper ASCII set as #162. An identical right-slanting superscript dash appears over Polish <s, c, z, n> and cuts through <l>. <z> can also carry a superscript dot, and <a> and <e> can be rightward-hooked (cedilla-like style). The phonetic value of these symbols is as follows (in alphabetical order; capitals will henceforth be used for these accented characters):

A nasalized back mid vowel, like in French 'son',

C voiceless pre-palatal affricate, somewhat like in English 'cheat',

E nasalized front mid vowel, like in French 'main',

L bilabial semi-vowel, like in English 'one',

N pre-palatal nasal, somewhat like in English 'knee',

O (ASCII 162) back high vowel, somewhat like in English 'put',

S voiceless pre-palatal fricative, somewhat like in English 'sheep',

Z (dotted) voiced apico-postalveolar fricative, somewhat like in English 'measure',

Z/ (slashed) voiced pre-palatal fricative, somewhat like in French/English 'gigolo'.

In Polish EMC all these symbols are normally stripped of their diacritics, which is communicatively unobtrusive in most cases due to the disambiguating effect of the sentential context. It is not entirely satisfactory, either, for a number of reasons. First, every third word on average in a continuous Polish text is graphemically changed, which can make for rather difficult reading at times. Second, while the phonetic value of some of the 'de-accented' (diacritic-less) symbols is close to their diacritic-full variants (E, C, N, S, Z), A, L and O are phonetically very different from <a, l, o>, which may affect lexical access in those readers/cases/circumstances where it proceeds (partly) through the phonetic channel.

In this situation, even for this Roman-based spelling, quasi-phonetic transcription could be used to avoid both using diacritics and creating new grapho-phonemic problems. I have myself successfully used this transcription a number of times in EMC. After the initial 'getting-used-to' period the reaction of readers was generally favourable. Notice, however, that unlike the diacritic-stripping process this sort of transcription goes far beyond mechanical graphemic manipulation. It is, indeed, a highly metalinguistic activity which involves pragmatically motivated deliberate tampering with Polish grapho- and morpho-phonemics. In the remainder of this abstract I will very sketchily describe some of the changes.

As far as simple graphemic substitutions are concerned, <u> can be used instead of O as it is simply the other (complementarily distributed) spelling variant of the back high vowel as well as its IPA rendition. Notice that we have already broken one spelling norm of Polish regulating the 'correct' distribution of O and <u>, the norm that Polish children painfully acquire in school. Another case like this is the substitution of <rz> for Z, which is of course not a case of orthodox phonetic transcription because the <rz> digraph has this phonetic value in Polish spelling only, and not in any accepted phonetic alphabet.

The symbols standing for the 'soft' palatal consonants of Polish (C, N, S, Z/) have no obvious substitutes. Polish palatals are also spelled as <ci>, <ni>, <si> and <zi> before vowels, but using these digraphs pre-consonantally and word-finally would lead to wholesale grapho-phonemic contrast neutralization, especially as the then pronounced <i> is very salient inflectionally, e.g. <SmieC> vs <Smieci> ('rubbish', sg. vs pl.). No real -- even computer-readable -- phonetic transciption would help in this case as the symbols are rather impractical for a casual Polish user of e-mail: /ts', J, s', z'/, respectively, in the recent suggestions for a revised SAM Phonetic Alphabet. The best solution probably is to simply use capitalized letters, as I did above (with the slash after Z rendered useless by representing Z as <rz>).

Finally, the two most interesting cases: nasalized vowels and the labial semi-vowel. The former are only pronounced as such in natural Polish before fricatives (and O word-finally); before other consonants they are dissolved into an oral vowel and a nasal homorganic with the following consonant, and word-finally E is denasalized. This phonological behaviour makes possible consistent avoidance of diacritics by transcribing the two vowels phonetically as <e, en, em, on, om>, depending on context, especially as they are so spelled in words of foreign origin: 'abonent', 'membrana', 'blond', 'kompletny' (which graphophonemic inconsistency is source of many a headache in schoolchildren and adults alike). This manipulation obviously requires a fair degree of metaphonological sophistication, but can be easily performed with a bit of practice. Notice, additionally, that this time we are not only flouting spelling rules, but also pragmatically defined careful pronunciation norms which militate against 'excessive' vowel denasalization of, for example, word-final E. Such rules are (most expectedly) invoked in the context of written EMC communication, as writing is of course perceived as more 'formal' than speech, even if it is e-mail writing. The pragmatics of casual EMC in Poland is only now developing.

Once informal speech is taken as the basis for graphemic simplification in EMC the path is open for even more drastic types of phonostylistic manipulation. In natural spoken Polish the labial semi-vowel is often deleted, especially when absorbed by the adjacent /u/: <sLuZyC> 'serve', or <puLkownik> 'colonel', <kaLuZa> 'puddle'. It is possible to simplify the EMC representation of such words by simply dropping the L. The resulting 'casual speech' will fit very well the usual style of EMC, with the more formal styles served by accent-stripping or capitalization (the IPA /w/ has the phonetic value of /v/ in Polish spelling, and as such would be very confusing to Polish EM communicators).

All such grapho-phonemic manipulation in EMC is clearly motivated and constrained pragmatically, or -- to be more precise -- phatically, by the desire to clear obstacles in the channel of communication.