There are 13 messages in this issue. Topics in this digest:
1a. Re: SOV > SVO From: Alex Fink 1b. Re: SOV > SVO From: And Rosta 2a. Transcribing IPA From: Peter Bleackley 2b. Re: Transcribing IPA From: Alex Fink 2c. Re: Transcribing IPA From: Tony Harris 2d. Re: Transcribing IPA From: Philip Newton 2e. Re: Transcribing IPA From: Herman Miller 2f. Re: Transcribing IPA From: Patrick Dunn 2g. Re: Transcribing IPA From: Michael Everson 2h. Re: Transcribing IPA From: Peter Bleackley 2i. Re: Transcribing IPA From: Peter Bleackley 2j. Re: Transcribing IPA From: Sam Stutter 2k. Re: Transcribing IPA From: Sam Stutter Messages ________________________________________________________________________ 1a. Re: SOV > SVO Posted by: "Alex Fink" 000...@gmail.com Date: Thu May 12, 2011 5:04 am ((PDT)) On Wed, 11 May 2011 18:45:48 -0400, Amanda Babcock Furrow <la...@quandary.org> wrote: >On Wed, May 11, 2011 at 04:45:37PM -0500, Patrick Dunn wrote: >> In Spanish and, AFAIK, Italian, the old SOV word order survives with >> pronominal objects. >> >> Of course, case endings only survive in Spanish on pronouns, so perhaps >> that's why. In fact, that seems likely. >> >> Why the same thing did not happen with English is hard to say. > >Well, we have a similar thing with some seperable verbs: > >"Mommy picked up Jordan!" vs. >"Mommy picked you up!" Similar, but only half comparable -- there's no *"picked up me", but "picked Jordan up" is perfectly grammatical (ob-IML proviso). Perhaps for heavy objects, the separated construction gets awkward enough that one might consider it ungrammatical, but I'm not sure whether one ought to see that as some sort of processing-depth rule like the one about central embedding. Alex Messages in this topic (12) ________________________________________________________________________ 1b. Re: SOV > SVO Posted by: "And Rosta" and.ro...@gmail.com Date: Thu May 12, 2011 8:30 am ((PDT)) Alex Fink, On 12/05/2011 13:02: > On Wed, 11 May 2011 18:45:48 -0400, Amanda Babcock Furrow > <la...@quandary.org> wrote: > >> On Wed, May 11, 2011 at 04:45:37PM -0500, Patrick Dunn wrote: >>> In Spanish and, AFAIK, Italian, the old SOV word order survives with >>> pronominal objects. >>> >>> Of course, case endings only survive in Spanish on pronouns, so perhaps >>> that's why. In fact, that seems likely. >>> >>> Why the same thing did not happen with English is hard to say. >> >> Well, we have a similar thing with some seperable verbs: >> >> "Mommy picked up Jordan!" vs. >> "Mommy picked you up!" > > Similar, but only half comparable -- there's no *"picked up me", but "picked > Jordan up" is perfectly grammatical (ob-IML proviso). Perhaps for heavy > objects, the separated construction gets awkward enough that one might > consider it ungrammatical, but I'm not sure whether one ought to see that as > some sort of processing-depth rule like the one about central embedding. Forgive me if I've missed the point, but I'll pipe up to point out that "picked up me" is (in most lects) possible when ME is nonclitic/emphatic (e.g. "I picked up HIM and then HE picked up ME". But when pronouns have clitic forms they must follow the verb. Hence "pick 'em up" but never *"pick up 'EM". (It's not quite that simple, because 'direct object' clitics can cliticize to nonclitic 'indirect object', e.g. "I'll give Alex 'em", and in lects with possessive aux HAVE, they can cliticize to at least pronominal 'subjects', "Have you 'em now?") As far as I can see, English clitic pronouns are in no way relics of English's SOV past. --And. Messages in this topic (12) ________________________________________________________________________ ________________________________________________________________________ 2a. Transcribing IPA Posted by: "Peter Bleackley" peter.bleack...@rd.bbc.co.uk Date: Thu May 12, 2011 8:25 am ((PDT)) I'm trying to devise an ascii transcription of the IPA for use in the language recogniser I'm trying to create. Unfortunately, restrictions imposed by the speech recognition engine I'm using preclude the use of CXS for the purpose, so I'm having to roll my own. You can see my work so far on http://code.google.com/p/spoken-language-recognition/wiki/PhoneticTranscription I'd appreciate any helpful feedback. Pete Messages in this topic (11) ________________________________________________________________________ 2b. Re: Transcribing IPA Posted by: "Alex Fink" 000...@gmail.com Date: Thu May 12, 2011 9:27 am ((PDT)) On Thu, 12 May 2011 16:21:43 +0100, Peter Bleackley <peter.bleack...@rd.bbc.co.uk> wrote: >I'm trying to devise an ascii transcription of the IPA for use in the >language recogniser I'm trying to create. Unfortunately, restrictions >imposed by the speech recognition engine I'm using preclude the use of >CXS for the purpose, so I'm having to roll my own. You can see my work >so far on >http://code.google.com/p/spoken-language-recognition/wiki/PhoneticTranscription > >I'd appreciate any helpful feedback. My own inclination here would be not to reinvent the wheel, and just devise some sort of mapping from ways characters can be modified in CXS to suffixal characters in Sphinx notation, along the lines of e.g. suffix "c" for capitalisation, suffix "b" for a backslash, etc. so [r r\ R R\] map to "r rb rc rcb". Alex Messages in this topic (11) ________________________________________________________________________ 2c. Re: Transcribing IPA Posted by: "Tony Harris" t...@alurhsa.org Date: Thu May 12, 2011 10:06 am ((PDT)) I'm curious. Wouldn't SAMPA or X_SAMPA be exactly that, an ASCII transcription of IPA? On 5/12/11 12:24 PM, Alex Fink wrote: > On Thu, 12 May 2011 16:21:43 +0100, Peter Bleackley > <peter.bleack...@rd.bbc.co.uk> wrote: > >> I'm trying to devise an ascii transcription of the IPA for use in the >> language recogniser I'm trying to create. Unfortunately, restrictions >> imposed by the speech recognition engine I'm using preclude the use of >> CXS for the purpose, so I'm having to roll my own. You can see my work >> so far on >> http://code.google.com/p/spoken-language-recognition/wiki/PhoneticTranscription >> >> I'd appreciate any helpful feedback. > My own inclination here would be not to reinvent the wheel, and just devise > some sort of mapping from ways characters can be modified in CXS to suffixal > characters in Sphinx notation, along the lines of e.g. suffix "c" for > capitalisation, suffix "b" for a backslash, etc. so [r r\ R R\] map to "r > rb rc rcb". > > Alex Messages in this topic (11) ________________________________________________________________________ 2d. Re: Transcribing IPA Posted by: "Philip Newton" philip.new...@gmail.com Date: Thu May 12, 2011 1:29 pm ((PDT)) On Thu, May 12, 2011 at 19:01, Tony Harris <t...@alurhsa.org> wrote: > I'm curious. Wouldn't SAMPA or X_SAMPA be exactly that, an ASCII > transcription of IPA? Those are an ASCII transcription of IPA, but not an alphabetic-only, case-insentitive transcription of IPA, which is apparently what is needed. Digraphs are OK, non-alphabetic characters or letters differing only in case are not. Cheers, Philip -- Philip Newton <philip.new...@gmail.com> Messages in this topic (11) ________________________________________________________________________ 2e. Re: Transcribing IPA Posted by: "Herman Miller" hmil...@io.com Date: Thu May 12, 2011 6:01 pm ((PDT)) On 5/12/2011 11:21 AM, Peter Bleackley wrote: > I'm trying to devise an ascii transcription of the IPA for use in the > language recogniser I'm trying to create. Unfortunately, restrictions > imposed by the speech recognition engine I'm using preclude the use of > CXS for the purpose, so I'm having to roll my own. You can see my work > so far on > http://code.google.com/p/spoken-language-recognition/wiki/PhoneticTranscription > > > I'd appreciate any helpful feedback. > > Pete Well, it's good to have some kind of systematic rules like your -R for retroflex. Are you limited to 2-character symbols? Distinguishing between the various "r" sounds could be tricky if so. But if you can use 3 characters, you could have e.g. RA for the alveolar approximant [ɹ], RAR for the retroflex approximant [É»], RT for the alveolar tap [ɾ], and RTR for the retroflex flap [ɽ]. Another convention that you might consider is doubling a letter for representing small capital letters, e.g. GG for [É¢], HH for [Ê], RR for [Ê], YY for [Ê]. One thing I don't like much is SD and ZD for the dental fricatives; [θ] and [ð] are not much like [s] and [z], which are sibilants. I think TH and DH would be more recognizable. Also for ease of recognition I'd use SH and ZH for [Ê], [Ê]. How about the glottal stop? Messages in this topic (11) ________________________________________________________________________ 2f. Re: Transcribing IPA Posted by: "Patrick Dunn" pwd...@gmail.com Date: Thu May 12, 2011 6:53 pm ((PDT)) You could approach it differently, by creating digraphs (or better yet, trigraphs) that represent the location of the symbol in the IPA chart. For example, /t/ could be UAS for Unvoiced Alveolar Stop /c/ could be UPS for Unvoiced Palatal Stop /m/ could be VBN for Voiced Bilabial Nasal and so on. Since voicing in usually binary in most languages, you could reduce this to digraphs by swapping order for voicing, so /t/ is AS /d/ is SA /k/ is VS /g/ is SV it'd be pretty easy to come up with such a scheme for any one language or maybe a group of related languages. If you wanted it to work with every single language, I doubt it'll be possible. After all, there are some sounds that do not even fit neatly on an IPA chart, and every sound could conceivably be modified with diacritics. On Thu, May 12, 2011 at 7:58 PM, Herman Miller <hmil...@io.com> wrote: > On 5/12/2011 11:21 AM, Peter Bleackley wrote: > >> I'm trying to devise an ascii transcription of the IPA for use in the >> language recogniser I'm trying to create. Unfortunately, restrictions >> imposed by the speech recognition engine I'm using preclude the use of >> CXS for the purpose, so I'm having to roll my own. You can see my work >> so far on >> >> http://code.google.com/p/spoken-language-recognition/wiki/PhoneticTranscription >> >> >> I'd appreciate any helpful feedback. >> >> Pete >> > > Well, it's good to have some kind of systematic rules like your -R for > retroflex. Are you limited to 2-character symbols? Distinguishing between > the various "r" sounds could be tricky if so. But if you can use 3 > characters, you could have e.g. RA for the alveolar approximant [ɹ], RAR for > the retroflex approximant [É»], RT for the alveolar tap [ɾ], and RTR for the > retroflex flap [ɽ]. > > Another convention that you might consider is doubling a letter for > representing small capital letters, e.g. GG for [É¢], HH for [Ê], RR for > [Ê], > YY for [Ê]. > > One thing I don't like much is SD and ZD for the dental fricatives; [θ] and > [ð] are not much like [s] and [z], which are sibilants. I think TH and DH > would be more recognizable. Also for ease of recognition I'd use SH and ZH > for [Ê], [Ê]. > > How about the glottal stop? > -- I have stretched ropes from steeple to steeple; garlands from window to window; golden chains from star to star, and I dance. --Arthur Rimbaud Messages in this topic (11) ________________________________________________________________________ 2g. Re: Transcribing IPA Posted by: "Michael Everson" ever...@evertype.com Date: Thu May 12, 2011 11:57 pm ((PDT)) On 12 May 2011, at 17:21, Peter Bleackley wrote: > I'm trying to devise an ascii transcription of the IPA for use in the > language recogniser I'm trying to create. Unfortunately, restrictions imposed > by the speech recognition engine I'm using preclude the use of CXS for the > purpose, so I'm having to roll my own. You can see my work so far on > http://code.google.com/p/spoken-language-recognition/wiki/PhoneticTranscription > > I'd appreciate any helpful feedback. Use real Unicode characters. That is what they are for. Michael Everson * http://www.evertype.com/ Messages in this topic (11) ________________________________________________________________________ 2h. Re: Transcribing IPA Posted by: "Peter Bleackley" peter.bleack...@rd.bbc.co.uk Date: Fri May 13, 2011 1:37 am ((PDT)) staving Tony Harris: > I'm curious. Wouldn't SAMPA or X_SAMPA be exactly that, an ASCII > transcription of IPA? > It has to be case insensitive, and non-alphanumeric symbols aren't allowed. Pete Messages in this topic (11) ________________________________________________________________________ 2i. Re: Transcribing IPA Posted by: "Peter Bleackley" peter.bleack...@rd.bbc.co.uk Date: Fri May 13, 2011 1:43 am ((PDT)) On 13/05/2011 01:58, Herman Miller wrote: > On 5/12/2011 11:21 AM, Peter Bleackley wrote: >> I'm trying to devise an ascii transcription of the IPA for use in the >> language recogniser I'm trying to create. Unfortunately, restrictions >> imposed by the speech recognition engine I'm using preclude the use of >> CXS for the purpose, so I'm having to roll my own. You can see my work >> so far on >> http://code.google.com/p/spoken-language-recognition/wiki/PhoneticTranscription >> >> >> >> I'd appreciate any helpful feedback. >> >> Pete > > Well, it's good to have some kind of systematic rules like your -R for > retroflex. Are you limited to 2-character symbols? Distinguishing > between the various "r" sounds could be tricky if so. But if you can use > 3 characters, you could have e.g. RA for the alveolar approximant [ɹ], > RAR for the retroflex approximant [É»], RT for the alveolar tap [ɾ], and > RTR for the retroflex flap [ɽ]. > I'm not restricted to 2 character symbols, and will certainly need symbols of arbitrary length for some purposes. > Another convention that you might consider is doubling a letter for > representing small capital letters, e.g. GG for [É¢], HH for [Ê], RR for > [Ê], YY for [Ê]. > > One thing I don't like much is SD and ZD for the dental fricatives; [θ] > and [ð] are not much like [s] and [z], which are sibilants. I think TH > and DH would be more recognizable. Also for ease of recognition I'd use > SH and ZH for [Ê], [Ê]. > I was thinking of using H as an aspiration diacritic. > How about the glottal stop? That's one of many problems I've got. There's nothing obvious to use for pharyngeals, clicks, or lateral fricatives, and the sheer number of vowels that are possible. This was a lot easier with stribography, which was never meant to be useful! Pete Messages in this topic (11) ________________________________________________________________________ 2j. Re: Transcribing IPA Posted by: "Sam Stutter" sam.stut...@student.manchester.ac.uk Date: Fri May 13, 2011 1:46 am ((PDT)) I was thinking of something on those lines too. If you allocate each letter in IPA a two character code (2 is more than enough, 676); it could be randomly selected, but something intuitive is probably best. Then allocate each diacritic a two character code: can you use punctuation? Then you'd have; YV!! for, say /y/ extra high tone BB.. for /b/ breathy voiced If you can't use punctuation, then just another two letter code; YVeh BBbv (the change in case is just to make it look neat) On 13 May 2011, at 02:49, Patrick Dunn <pwd...@gmail.com> wrote: > You could approach it differently, by creating digraphs (or better yet, > trigraphs) that represent the location of the symbol in the IPA chart. For > example, > > /t/ could be UAS for Unvoiced Alveolar Stop > /c/ could be UPS for Unvoiced Palatal Stop > /m/ could be VBN for Voiced Bilabial Nasal > > and so on. > > Since voicing in usually binary in most languages, you could reduce this to > digraphs by swapping order for voicing, so > > /t/ is AS > /d/ is SA > /k/ is VS > /g/ is SV > > it'd be pretty easy to come up with such a scheme for any one language or > maybe a group of related languages. If you wanted it to work with every > single language, I doubt it'll be possible. After all, there are some > sounds that do not even fit neatly on an IPA chart, and every sound could > conceivably be modified with diacritics. > > > > On Thu, May 12, 2011 at 7:58 PM, Herman Miller <hmil...@io.com> wrote: > >> On 5/12/2011 11:21 AM, Peter Bleackley wrote: >> >>> I'm trying to devise an ascii transcription of the IPA for use in the >>> language recogniser I'm trying to create. Unfortunately, restrictions >>> imposed by the speech recognition engine I'm using preclude the use of >>> CXS for the purpose, so I'm having to roll my own. You can see my work >>> so far on >>> >>> http://code.google.com/p/spoken-language-recognition/wiki/PhoneticTranscription >>> >>> >>> I'd appreciate any helpful feedback. >>> >>> Pete >>> >> >> Well, it's good to have some kind of systematic rules like your -R for >> retroflex. Are you limited to 2-character symbols? Distinguishing between >> the various "r" sounds could be tricky if so. But if you can use 3 >> characters, you could have e.g. RA for the alveolar approximant [ɹ], RAR for >> the retroflex approximant [É»], RT for the alveolar tap [ɾ], and RTR for the >> retroflex flap [ɽ]. >> >> Another convention that you might consider is doubling a letter for >> representing small capital letters, e.g. GG for [É¢], HH for [Ê], RR for >> [Ê], >> YY for [Ê]. >> >> One thing I don't like much is SD and ZD for the dental fricatives; [θ] and >> [ð] are not much like [s] and [z], which are sibilants. I think TH and DH >> would be more recognizable. Also for ease of recognition I'd use SH and ZH >> for [Ê], [Ê]. >> >> How about the glottal stop? >> > > > > -- > I have stretched ropes from steeple to steeple; garlands from window to > window; golden chains from star to star, and I dance. --Arthur Rimbaud Messages in this topic (11) ________________________________________________________________________ 2k. Re: Transcribing IPA Posted by: "Sam Stutter" sam.stut...@student.manchester.ac.uk Date: Fri May 13, 2011 1:52 am ((PDT)) I forgot to say, blank fields would be something like xx: SHxx On 13 May 2011, at 09:46, Sam Stutter <sam.stut...@student.manchester.ac.uk> wrote: > I was thinking of something on those lines too. If you allocate each letter > in IPA a two character code (2 is more than enough, 676); it could be > randomly selected, but something intuitive is probably best. Then allocate > each diacritic a two character code: can you use punctuation? Then you'd have; > > YV!! for, say /y/ extra high tone > BB.. for /b/ breathy voiced > > If you can't use punctuation, then just another two letter code; > > YVeh > BBbv > > (the change in case is just to make it look neat) > > On 13 May 2011, at 02:49, Patrick Dunn <pwd...@gmail.com> wrote: > >> You could approach it differently, by creating digraphs (or better yet, >> trigraphs) that represent the location of the symbol in the IPA chart. For >> example, >> >> /t/ could be UAS for Unvoiced Alveolar Stop >> /c/ could be UPS for Unvoiced Palatal Stop >> /m/ could be VBN for Voiced Bilabial Nasal >> >> and so on. >> >> Since voicing in usually binary in most languages, you could reduce this to >> digraphs by swapping order for voicing, so >> >> /t/ is AS >> /d/ is SA >> /k/ is VS >> /g/ is SV >> >> it'd be pretty easy to come up with such a scheme for any one language or >> maybe a group of related languages. If you wanted it to work with every >> single language, I doubt it'll be possible. After all, there are some >> sounds that do not even fit neatly on an IPA chart, and every sound could >> conceivably be modified with diacritics. >> >> >> >> On Thu, May 12, 2011 at 7:58 PM, Herman Miller <hmil...@io.com> wrote: >> >>> On 5/12/2011 11:21 AM, Peter Bleackley wrote: >>> >>>> I'm trying to devise an ascii transcription of the IPA for use in the >>>> language recogniser I'm trying to create. Unfortunately, restrictions >>>> imposed by the speech recognition engine I'm using preclude the use of >>>> CXS for the purpose, so I'm having to roll my own. You can see my work >>>> so far on >>>> >>>> http://code.google.com/p/spoken-language-recognition/wiki/PhoneticTranscription >>>> >>>> >>>> I'd appreciate any helpful feedback. >>>> >>>> Pete >>>> >>> >>> Well, it's good to have some kind of systematic rules like your -R for >>> retroflex. Are you limited to 2-character symbols? Distinguishing between >>> the various "r" sounds could be tricky if so. But if you can use 3 >>> characters, you could have e.g. RA for the alveolar approximant [ɹ], RAR >>> for >>> the retroflex approximant [É»], RT for the alveolar tap [ɾ], and RTR for >>> the >>> retroflex flap [ɽ]. >>> >>> Another convention that you might consider is doubling a letter for >>> representing small capital letters, e.g. GG for [É¢], HH for [Ê], RR for >>> [Ê], >>> YY for [Ê]. >>> >>> One thing I don't like much is SD and ZD for the dental fricatives; [θ] and >>> [ð] are not much like [s] and [z], which are sibilants. I think TH and DH >>> would be more recognizable. Also for ease of recognition I'd use SH and ZH >>> for [Ê], [Ê]. >>> >>> How about the glottal stop? >>> >> >> >> >> -- >> I have stretched ropes from steeple to steeple; garlands from window to >> window; golden chains from star to star, and I dance. --Arthur Rimbaud Messages in this topic (11) ------------------------------------------------------------------------ Yahoo! Groups Links <*> To visit your group on the web, go to: http://groups.yahoo.com/group/conlang/ <*> Your email settings: Digest Email | Traditional <*> To change settings online go to: http://groups.yahoo.com/group/conlang/join (Yahoo! ID required) <*> To change settings via email: conlang-nor...@yahoogroups.com conlang-fullfeatu...@yahoogroups.com <*> To unsubscribe from this group, send an email to: conlang-unsubscr...@yahoogroups.com <*> Your use of Yahoo! Groups is subject to: http://docs.yahoo.com/info/terms/ ------------------------------------------------------------------------