Re: Vulcan Word Generator (VWG) trials and tribulations Rob Zook Thu, 25 Mar 1999 18:48:48 -0600 At 05:43 PM 3/25/99 -0600, Saul Epstein wrote: >(Quotes Rob Zook: Thursday, March 25, 1999 4:32 PM) > > >>At 03:20 PM 3/25/99 -0600, Steven Boozer wrote: >> >>>Refresh my memory, I've forgotten where we left the approved >>>transcription. (Marketa did approve it, yes?) > >She did, and therefore I REALLY, REALLY hope we can abide by it. Does that mean you'd object horribly to a little tweeking here and there :) >Though Marketa seems to pronounce two major variants each of A, E, I, >and O, the distribution does not seem to bear any relation to the use >of the <:> diacritic, but rather to the context of the sounds, which >suggests that each of the variants are allophones of single phonemes, >(except in the case of A). For this reason, I suggested that no >distinctions be made in spelling to represent the variations, either >through diacritics or two-letter signs (except in the case of A). This >suggestion informed the transcription that was approved. Ok, I'll by that. It would be simpler in the program to work with phonemes anyway (since that's what it was originally working with). >With regard to the sounds specifically asked about, the approved >represents both the [e] in "bet" and the first vowel in the diphthong >in "bait." It doesn't represent the diphthong in "bait;" that is >written [ei]. And I'll recommend right here that any occurance in the >original wordlist of apparent dipthongs consisting of be >converted to , because I would guess that is what whoever coined >those words meant in almost every case. Yeah, that's what I was thinking too. > and were alternatives offered for transcription of what had >been , and which Marketa does seem to use phonemically. It is the >vowel in "ah-hah" in many dialects. I don't remember Marketa >pronouncing any differently from the approved diphthong , so >it may be that should become . Have to listen to the tape again. I don't know if she got very far into the Ks or not. >>e: and a: are transliterations for a couple of sounds which appear in >the >>ZC Lexicon, but which I think we all agreed to dispense with because >the ZC >>Lexicon does not adequately define how they sound. > > >There were several different problems involved, and that was one of >them, mainly with regard to ,,<^>, and <^:>. Well, yeah, I guess e: was defined in the ZC Grammar. I just don't fathom what a "emphatic" French prenez means :) >>>: ur-sev-eh >>>: va'-ne: >>> >>>ur-seveh, va'neh >>> >>>/eh/ and /e:/ are the same vowel? >> >>Possibly, that's what I'm asking about. > >The H was probably included in to prevent English speakers >from pronouncing it as a "long E," a "long A," or a "silent E." >Fortunately, it doesn't matter all that much if they're phonetically >the same or not; phonemically, their identical with each other, and >with plain old . Probably, and except for ah, we can probably change all the /eh/'s to /e/'s. >>>: kai-idth >>>: lai-la-ra >>>: nai-la-ra >>>: rel-dai >>> >>>kyidth (ky'idth), lylara, nylara, reldy (yech! reldai looks better) >>> >>>Can we retain /ai/ or /ay/, at least for word finals? >> >>Personally, I think we should make the use of the /y/ or /ai/ >>completely optional. I prefer the /ai/, after all, why make a special fuss >>over one dipthong? > > wake me when it's over... Well, for at least the input to my program, I'll have to convert /y/ to /ai/. Otherwise I wouldn't have the words in phonemic form. Here's a new list based on the revisions we've suggested (feel free to object again Steven if you don't like these changes either :). I've elimnated all the words which passed without protest, and ones which don't have any letters which conflict with the approved transliteration. a: -> ? an'-kharh -> an'-khar ahn'-woon -> ahn'-wun da'-Nii-khirch -> da'-Ni?-khirch droy -> droi farr -> far? or keep it rr? fort'-eh -> fort'-e imroy -> imroi Ka'-ath-y-ra -> Ka'-ath-ai*-ra kae'-k'-ak-ka-yam -> kae'-k'-ak-ka-jam kahr-y-tan -> kahr-ai-tan or kahr-i-tan? It's a TOS book word. Kh'-ask-peth-e-ya'-th -> Kh'-ask-peth-e-ja'-th Kh'-ask-ey-ra-lath-a -> Kh'-ask-ei-ra-lath-a Kh'-spark-ey-ra-lath-a -> Kh'-spark-ei-ra-lath-a krey-la -> krei-la kroy-kah -> kroi-kah k'-tveh-i -> k'-tve-i or k'-tve-hi? le'-mat-ya -> le'-mat-ja ma-toy -> ma-toi na'-Tha'-thhja -> thhja = a aspirated dental fricative+y? ni-droy -> ni-droi say-a -> saj-a shi-ka'-ree -> shi-ka'-ri shroy -> shroi tal'-shay-a -> tal'-shai-a or tal'-sha-ja? t'-hy'-la -> t'-hai*'-la to-tsu'-k'-hy -> to-tsu'-k'-hai* tow -> tau tr'-ai-yar -> tr'-ai-jar ur-sev-eh -> ur-sev-e va'-ne: -> va'-ne vi-proy -> vi-proi Several times I ran accross a set of consonants which involved /r/ either in the form /CrC/ or /rCh/. I propose that we leave that as a proper spelling and that by default the r is pronounced as /*r/ in both cases (where * represents the neutral voiceless vowel). Another thing we should resolve is whether or not to treat the /'/ as phonemic. I advocate the idea of /'/ as a phoneme. It's allophones can be as a neutral voiceles vowel followed by a voiceless stop when forming a syllable with a lone consonant either before or at the end of a word; or as a voiceless stop in the middle of a word when ending a syllable (which I think covers all cases in the dictionary the way I've divided the words into syllables). The word /t'amtar'am/ shows both cases. I think both of these cases adds a little random "natural" feel to the language of sufficient "alienness" to keep those rules from being identified with any existing human tongue. Rob Z.