Loading ...
Sorry, an error occurred while loading the content.

Re: Distribution of phonemes in lexicon

Expand Messages
  • Alex Fink
    ... But NF# is not licit, right? ... This makes me imagine that /o/ is of recent origin in TF, with its instances in /ko so no/ being something like a
    Message 1 of 8 , May 2, 2013
    • 0 Attachment
      On Wed, 1 May 2013 23:19:00 -0700, H. S. Teoh <hsteoh@...> wrote:

      >Recently, while trying to decide what kind of writing system Tatari
      >Faran should have, I went through the entire lexicon and catalogued all
      >attested syllables. The idea is that instead of arbitrarily flipping a
      >coin for the type of writing system to adopt, I should look for patterns
      >in the currently-attested syllables and see if they suggest what kind of
      >system the san faran would have invented.
      >
      >TF has a rather simple syllabic structure: all words are of the form
      >(CV(N))*(F) where C = any consonant, N is a nasal (/m/ or /n/), and F is
      >the smaller set of allowable final consonants. V is any vowel or glide.
      >So this makes cataloguing rather simple, if a bit tedious.

      But NF# is not licit, right?

      >And sure enough, some interesting trends arose! The most prominent
      >"discovery" is that the most frequent syllables are of the form Ca, Ci,
      >Cu, that is, a consonant paired with an apical vowel. Ce and Cai are
      >less frequent, but comparable (though the distribution varies greatly
      >depending on C). /o/ (which is [O]) is in fact extremely rare as far as
      >occurrences in words are concerned, though it crops up in the neuter
      >case particles that in any non-trivial text would occur everywhere (IOW
      >it rarely occurs in words, but the few words containing it are very
      >frequent).

      This makes me imagine that /o/ is of recent origin in TF, with its instances in /ko so no/ being something like a reduction of *u in (certain) unstressed positions, or maybe of *au or *ua if there're too many nonreducing *u for that to be comfortable. (Maybe pre-TF had not just /kei ?ei/ > /ki?ei/ etc. but /kau ?ei/ > /kO?ei/, and in the latter cases the reduced allomorph generalised.)

      How is /o/ distributed in words otherwise?

      >There are also some interesting distribution asymmetries: /ai/ following
      >a glottal stop has a lower frequency than /u/, but following /n/ it is
      >almost as common as /i/, and following /t/ it is *more* common than /i/.

      You said /-itai/ was a verbalising suffix, IIRC. Controlled for its occurrences, is /tai/ still more common than /ti/?

      >In any case, this distribution suggests that TF's native writing system
      >should be a kind of abugida-like system, in which Ca, Ci, and Cu have
      >dedicated glyphs, and Ce, Cai, Co, and others should use diacritics to
      >modify Ca, Ci, or Cu. The writing will be vertical, so there will be
      >left-diacritics (dextrocritics) and right-diacritics (aristocritics --
      >technically "aristerocritics" but that's too ... bureaucratic :-P).

      Aw, if it were me I'd avoid arbitrary manglings and keep _aristero-_. And _dextro-_ is only a Latin form, isn't it? Greek had _dexios_.

      >Currently, I'm thinking something like:
      >
      >- Caa = Ca + length_mark
      >- Ce = Ca + e-diacritic
      >- Cei = Ca + e-diacritic + length_mark (/ei/ is considered to be long /e/)
      >- Cii = Ci + length mark
      >- Co = Ca + o-diacritic
      >- Cue = Cu + ue-diacritic (/ue/ = [M])
      >- Cuu = Cu + length mark

      Building on Roger's idea, you wouldn't have to have a separate diacritic for each of the /e/, /o/, and /ue/ elements; the diacritics could double up. For instance, you could have a "make more i-like" diacritic which on Ca gives Ce, and on Cu gives Cue. That leaves just Co unaccounted for; but since you said Co is marginal, it may well do something funny.

      >- Cau = Ca + w-aristocritic (since /au/ is pronounced [ao] or [aw])
      >- Cua = Ca + w-dextrocritic (since /ua/ is pronounced [wa])
      >- Cui = Cu + i-aristocritic (/ui/ = [uj])
      >- Ciu = Cu + i-dextrocritic (/iu/ = [ju])
      >- Cai = Ca + i-aristocritic (/ai/ = [aj])
      >- Cia = Ca + i-dextrocritic (/ia/ = [ja])
      >
      >Syllable-final /m/ and /n/ will have their own diacritics as well,
      >perhaps the same glyph as dextrocritic for /m/ and aristocritic for /n/.

      I too think this is a false symmetry. /m/ is not to /n/ as preceding is to following.

      Do /m/ and /n/ *contrast* word-internally, or do internal nasals always assimilate to the following consonant? In the latter case, the more parsimonious phonemic analysis, and thus the more parsimonious writing system, would make them the same.

      On Thu, 2 May 2013 08:20:16 -0700, H. S. Teoh <hsteoh@...> wrote:

      >On Thu, May 02, 2013 at 06:59:52AM -0700, Roger Mills wrote:
      >> OK on those; but: how common is "ue" [M], is it phonemic?? Not worth
      >> having its own diacritic???
      >
      >Yes, [M] is phonemic. It's not particularly common, but does occur in
      >certain common words like _asuen_ [?asMn] "younger brother", _kuen_
      >[kMn] "tree", etc..

      Does it correlate with the position before /n/ in general, like in these two cases?

      >> Maybe for the iu, ia, ua 's you should consider a
      >> "palatalize/labialize" modification of some sort to the C character?
      >> That's what I did in the Gwr system.
      >
      >Yeah, the w-critics and i-critics could become ligatures of some sort,
      >perhaps.

      Eh, they could, but it seems less in the spirit of the TF phonology to do that. (For Gwr the case is probably different.)

      >> If you haven't already, you might take a look at my Gwr writing
      >> system, which is not dissimilar.
      >> http://cinduworld.tripod.com/prelim_gwr.htm Go to _section 3_ (The
      >> writing system) which has a link to the pdf.
      >
      >I like your idea of having final consonants represent VC instead of CV!
      >I think I'll st... adopt that idea. ;) So a final syllable like _kan_
      >would be written as _ka_ + _an_. The vowels have to match, as a
      >convention. In the case of modified vowels, as in _kuen_, the vowel
      >diacritic could sit between the two glyphs rather than clearly belonging
      >to one.

      Well, how then to write actual final vowels which agree with the previous one? Would /kana#/ also be written _ka na_, like /kan#/? Nothing wrong with such an ambiguity, of course, it's just something to bear in mind.

      Alex
    • Alex Fink
      ... Actually, hey, you know what would be a beautifully tidy and symmetric system? Allow not just aristerocritics and dexiocritics but _endocritics_, meaning
      Message 2 of 8 , May 2, 2013
      • 0 Attachment
        On Thu, 2 May 2013 13:46:16 -0400, Alex Fink <000024@...> wrote:

        >>Currently, I'm thinking something like:
        >>
        >>- Caa = Ca + length_mark
        >>- Ce = Ca + e-diacritic
        >>- Cei = Ca + e-diacritic + length_mark (/ei/ is considered to be long /e/)
        >>- Cii = Ci + length mark
        >>- Co = Ca + o-diacritic
        >>- Cue = Cu + ue-diacritic (/ue/ = [M])
        >>- Cuu = Cu + length mark
        >
        >Building on Roger's idea, you wouldn't have to have a separate diacritic for each of the /e/, /o/, and /ue/ elements; the diacritics could double up. For instance, you could have a "make more i-like" diacritic which on Ca gives Ce, and on Cu gives Cue. That leaves just Co unaccounted for; but since you said Co is marginal, it may well do something funny.

        Actually, hey, you know what would be a beautifully tidy and symmetric system? Allow not just aristerocritics and dexiocritics but _endocritics_, meaning "the nucleic vowel's quality is coloured by this one". Thus

        >>- Cau = Ca + w-aristocritic (since /au/ is pronounced [ao] or [aw])
        >>- Cua = Ca + w-dextrocritic (since /ua/ is pronounced [wa])
        Co = Ca + w-endocritic

        >>- Cui = Cu + i-aristocritic (/ui/ = [uj])
        >>- Ciu = Cu + i-dextrocritic (/iu/ = [ju])
        Cue = Cu + i-endocritic

        >>- Cai = Ca + i-aristocritic (/ai/ = [aj])
        >>- Cia = Ca + i-dextrocritic (/ia/ = [ja])
        Ce = Ca + i-endocritic

        The distributional symmetry is *perfect*! Ca can occur with i- and u-critics in all positions; Cu can occur with i-critics in app positions, but u- in none; Ci can't occur with i- or u-critics in any position.

        Alex
      • H. S. Teoh
        Argh, ran into the per-day posting limit yesterday, so had to wait until today to post these replies. :-( ... You re right, my formula is wrong. It should be
        Message 3 of 8 , May 3, 2013
        • 0 Attachment
          Argh, ran into the per-day posting limit yesterday, so had to wait until
          today to post these replies. :-(


          On Thu, May 02, 2013 at 01:46:16PM -0400, Alex Fink wrote:
          > On Wed, 1 May 2013 23:19:00 -0700, H. S. Teoh <hsteoh@...> wrote:
          >
          > >Recently, while trying to decide what kind of writing system Tatari
          > >Faran should have, I went through the entire lexicon and catalogued
          > >all attested syllables. The idea is that instead of arbitrarily
          > >flipping a coin for the type of writing system to adopt, I should
          > >look for patterns in the currently-attested syllables and see if they
          > >suggest what kind of system the san faran would have invented.
          > >
          > >TF has a rather simple syllabic structure: all words are of the form
          > >(CV(N))*(F) where C = any consonant, N is a nasal (/m/ or /n/), and F
          > >is the smaller set of allowable final consonants. V is any vowel or
          > >glide. So this makes cataloguing rather simple, if a bit tedious.
          >
          > But NF# is not licit, right?

          You're right, my formula is wrong. It should be (CV(N))*CV(F).


          > >And sure enough, some interesting trends arose! The most prominent
          > >"discovery" is that the most frequent syllables are of the form Ca,
          > >Ci, Cu, that is, a consonant paired with an apical vowel. Ce and Cai
          > >are less frequent, but comparable (though the distribution varies
          > >greatly depending on C). /o/ (which is [O]) is in fact extremely rare
          > >as far as occurrences in words are concerned, though it crops up in
          > >the neuter case particles that in any non-trivial text would occur
          > >everywhere (IOW it rarely occurs in words, but the few words
          > >containing it are very frequent).
          >
          > This makes me imagine that /o/ is of recent origin in TF, with its
          > instances in /ko so no/ being something like a reduction of *u in
          > (certain) unstressed positions, or maybe of *au or *ua if there're too
          > many nonreducing *u for that to be comfortable. (Maybe pre-TF had not
          > just /kei ?ei/ > /ki?ei/ etc. but /kau ?ei/ > /kO?ei/, and in the
          > latter cases the reduced allomorph generalised.)
          >
          > How is /o/ distributed in words otherwise?

          Unfortunately, /o/ occurs in stressed positions too, such as:

          koronta ["kO4Onta] "frog"
          somata ["sOmata] "give birth"
          tso' ["ts)O?] "wet"
          poriaba ["pO4jaba] "medicine"

          There are about 48 words containing /o/ in various stressed/unstressed
          positions, but we *are* talking about a lexicon of about 1100+ entries,
          so it's still quite a rare phoneme.

          Hmmm... could it be possible that /o/ is actually ancient, but the
          *other* vowels' frequencies are the result of recent mergers? Say, if
          /a/ and /i/ include both ancient /a/ and /i/, but also a bunch of other
          vowels/diphthongs that got simplified and merged into them recently?
          Some of these mergers could've involved /o/, thus reducing its apparent
          frequency? One obvious candidate, at least, is /au/, which is pronounced
          [ao] (not too distant from [aO]). But then /au/ only occurs in 10
          words, so that hardly accounts for the low frequency at all. :-(


          > >There are also some interesting distribution asymmetries: /ai/
          > >following a glottal stop has a lower frequency than /u/, but
          > >following /n/ it is almost as common as /i/, and following /t/ it is
          > >*more* common than /i/.
          >
          > You said /-itai/ was a verbalising suffix, IIRC. Controlled for its
          > occurrences, is /tai/ still more common than /ti/?

          Hmm, good point. I think a large percentage of /tai/ occurrences are as
          the verbalizing suffix. So perhaps it's *not* as common as it appears.
          :-)


          > >In any case, this distribution suggests that TF's native writing
          > >system should be a kind of abugida-like system, in which Ca, Ci, and
          > >Cu have dedicated glyphs, and Ce, Cai, Co, and others should use
          > >diacritics to modify Ca, Ci, or Cu. The writing will be vertical, so
          > >there will be left-diacritics (dextrocritics) and right-diacritics
          > >(aristocritics -- technically "aristerocritics" but that's too ...
          > >bureaucratic :-P).
          >
          > Aw, if it were me I'd avoid arbitrary manglings and keep _aristero-_.
          > And _dextro-_ is only a Latin form, isn't it? Greek had _dexios_.

          You're right! So it should be dexiocritic and aristerocritic instead. I
          still don't quite like the length of "aristerocritic" though. Maybe I
          should just drop all pretenses of Greek scholarship and just write
          L-critic and R-critic. (Or should that be D-critic and A-critic? But
          that may be needlessly obscure.) :)


          > >Currently, I'm thinking something like:
          > >
          > >- Caa = Ca + length_mark
          > >- Ce = Ca + e-diacritic
          > >- Cei = Ca + e-diacritic + length_mark (/ei/ is considered to be long /e/)
          > >- Cii = Ci + length mark
          > >- Co = Ca + o-diacritic
          > >- Cue = Cu + ue-diacritic (/ue/ = [M])
          > >- Cuu = Cu + length mark
          >
          > Building on Roger's idea, you wouldn't have to have a separate
          > diacritic for each of the /e/, /o/, and /ue/ elements; the diacritics
          > could double up. For instance, you could have a "make more i-like"
          > diacritic which on Ca gives Ce, and on Cu gives Cue. That leaves just
          > Co unaccounted for; but since you said Co is marginal, it may well do
          > something funny.

          Yeah, I did consider merging some of those diacritics. I like the idea
          of a fronting/heightening diacritic; it's both linguistically sound and
          natively plausible. :)

          And actually, /ue/ is less frequent than /o/, so maybe a *lowering*
          diacritic is in order? So Ci + lowering = Ce, Cu + lowering = Co, then
          something else altogether for Cue.


          > >- Cau = Ca + w-aristocritic (since /au/ is pronounced [ao] or [aw])
          > >- Cua = Ca + w-dextrocritic (since /ua/ is pronounced [wa])
          > >- Cui = Cu + i-aristocritic (/ui/ = [uj])
          > >- Ciu = Cu + i-dextrocritic (/iu/ = [ju])
          > >- Cai = Ca + i-aristocritic (/ai/ = [aj])
          > >- Cia = Ca + i-dextrocritic (/ia/ = [ja])
          > >
          > >Syllable-final /m/ and /n/ will have their own diacritics as well,
          > >perhaps the same glyph as dextrocritic for /m/ and aristocritic for
          > >/n/.
          >
          > I too think this is a false symmetry. /m/ is not to /n/ as preceding
          > is to following.

          True.


          > Do /m/ and /n/ *contrast* word-internally, or do internal nasals
          > always assimilate to the following consonant? In the latter case, the
          > more parsimonious phonemic analysis, and thus the more parsimonious
          > writing system, would make them the same.

          Hmm. Now, this is certainly curious... /mt/ never occurs (it's always
          /nt/), neither does /nb/ (it's always /mb/). It's always /n'/ not /m'/.
          But both /mk/ and /nk/ do occur, though, as do /mp/ and /np/, and /mr/
          and /nr/. Apparently no nasal ever precedes /f/ or /s/.

          I guess they should be treated separately then. Though their
          distribution is rather interesting.


          > On Thu, 2 May 2013 08:20:16 -0700, H. S. Teoh <hsteoh@...> wrote:
          >
          > >On Thu, May 02, 2013 at 06:59:52AM -0700, Roger Mills wrote:
          > >> OK on those; but: how common is "ue" [M], is it phonemic?? Not
          > >> worth having its own diacritic???
          > >
          > >Yes, [M] is phonemic. It's not particularly common, but does occur in
          > >certain common words like _asuen_ [?asMn] "younger brother", _kuen_
          > >[kMn] "tree", etc..
          >
          > Does it correlate with the position before /n/ in general, like in
          > these two cases?

          Not always; there are words like _tsuekin_, _bue'a_, _kueharan_,
          _tueri_, _uesan_, etc.


          > >> Maybe for the iu, ia, ua 's you should consider a
          > >> "palatalize/labialize" modification of some sort to the C
          > >> character? That's what I did in the Gwr system.
          > >
          > >Yeah, the w-critics and i-critics could become ligatures of some
          > >sort, perhaps.
          >
          > Eh, they could, but it seems less in the spirit of the TF phonology to
          > do that. (For Gwr the case is probably different.)

          Oh? Do you think it "feels" better to have a more "analytic" kind of
          writing system?

          Another thought I had was to recycle one of the w-critics for
          representing Co, most likely Ca + w if we regard it as a labializing
          diacritic, but then Ca + w is already taken by Cau. But perhaps Cau can
          be written as Ca + w + length, since /o/ is a short vowel, and [ao]
          could be construed to be long. :-P (And /au/ is also rarer than /o/, so
          having a longer way to write it seems to be justifiable.)


          > >> If you haven't already, you might take a look at my Gwr writing
          > >> system, which is not dissimilar.
          > >> http://cinduworld.tripod.com/prelim_gwr.htm Go to _section 3_
          > >> (The writing system) which has a link to the pdf.
          > >
          > >I like your idea of having final consonants represent VC instead of
          > >CV! I think I'll st... adopt that idea. ;) So a final syllable like
          > >_kan_ would be written as _ka_ + _an_. The vowels have to match, as a
          > >convention. In the case of modified vowels, as in _kuen_, the vowel
          > >diacritic could sit between the two glyphs rather than clearly
          > >belonging to one.
          >
          > Well, how then to write actual final vowels which agree with the
          > previous one? Would /kana#/ also be written _ka na_, like /kan#/?
          > Nothing wrong with such an ambiguity, of course, it's just something
          > to bear in mind.
          [...]

          I was thinking of having dedicated final variants for those consonants
          that *can* appear word-finally, so /ka/ and /ak/ would have different
          (but probably related) glyphs.

          I think the ambiguity between things like _ka na_ and _kan#_ would be
          unacceptable, since there are quite a few minimal pairs that contrast in
          that position. The resulting writing would be needlessly ambiguous.


          On Thu, May 02, 2013 at 02:34:53PM -0400, Alex Fink wrote:
          [...]
          > Actually, hey, you know what would be a beautifully tidy and symmetric
          > system? Allow not just aristerocritics and dexiocritics but
          > _endocritics_, meaning "the nucleic vowel's quality is coloured by
          > this one". Thus
          >
          > >>- Cau = Ca + w-aristocritic (since /au/ is pronounced [ao] or [aw])
          > >>- Cua = Ca + w-dextrocritic (since /ua/ is pronounced [wa])
          > Co = Ca + w-endocritic
          >
          > >>- Cui = Cu + i-aristocritic (/ui/ = [uj])
          > >>- Ciu = Cu + i-dextrocritic (/iu/ = [ju])
          > Cue = Cu + i-endocritic
          >
          > >>- Cai = Ca + i-aristocritic (/ai/ = [aj])
          > >>- Cia = Ca + i-dextrocritic (/ia/ = [ja])
          > Ce = Ca + i-endocritic
          >
          > The distributional symmetry is *perfect*! Ca can occur with i- and
          > u-critics in all positions; Cu can occur with i-critics in app
          > positions, but u- in none; Ci can't occur with i- or u-critics in any
          > position.
          [...]

          Now, that is a cool idea! I can probably even design the glyphs such
          that Ca and Cu have cavities into which the endocritics can be fitted,
          but Ci doesn't. And since Ca can take both i- and w-endocritics, but Cu
          can only take an i-endocritic, perhaps the w-endocritic is larger than
          the i-endocritic, and the cavity of the Ca's are larger than the cavity
          of the Ci's, so they can't contain the "wrong" endocritic by design.

          Wow. I think I'll actually adopt this idea!

          //

          Anyway, in the spirit of not using up my per-day posting limit,
          yesterday while "muted" on list, I decided to do some actual conlanging
          :-P by adding a few more entries to the TF lexicon.

          Here are some tasteful entries:

          _ba'asat nari_ - adj. sweet (taste).
          Cognates: _ba'as_ - edible fruit. Rationale: the san faran's
          environment doesn't have sugar cane, so they don't have the
          concept of sugar sweetness; instead, their primary (only?)
          source of sweet-tasting food is fruits. So sweet == fruity. The
          finalizer _nari_ has the connotation of happiness, which I'm
          sure is how one feels when biting into a sweet, juicy, delicious
          fruit. :)

          _fasat niman_ - adj. meaty (taste).
          Cognates: _fasa_ - meat, carcass. The san faran do not rear
          animals for meat (they regard that as barbaric), but they do
          hùnt for meat. Hence, meat is considered a delicacy, and
          therefore finger-lickin' good: _niman_ is cognate with _nimat_
          "tongue", and has the connotation of licking. (_niman_ also
          occurs with the verb _suru_, to taste, to lick.)

          _bihas kiki_ - adj. spicy, hot (taste).
          Cognates: _bihuun_ - pepper, any powdered spice (esp. with hot,
          sneeze-inducing properties). _kiki_ - excited, agitated: which
          aptly describes the effect of chewing down a mouthful of
          something tainted with crushed habanero seeds and having your
          mouth catch (and perhaps spew) fire. :-P

          _po'at hike_ - adj. bitter (taste).
          Cognates: _poriaba_ - ointment; _poma'an_ - medicine (both from
          _po_ + something meaning grease or oil: _diaba_ "grease" and
          _ma'an_ "liquid extract", "oil"); _poribai_ - to anoint;
          _po'itai_ - to heal/cure. Herbal medicine is known to be bitter,
          in general, hence the derivation of "bitter". :) The finalizer
          _hike_ is one that keeps cropping up, meaning to bring ill, to
          cause trouble, discomfort, discord, and other such unpleasant
          things. Just about sums up bitter-tasting herbal medicine. :-P

          On that note, _hike_ acquired a new cognate:

          _hikas_ (1) masc. n. curse, misfortune;
          (2) adj. _hikas kirue_ - accursed, tainted, diseased (esp. with
          a deadly, infectious contagion).
          Cognates: _hike_ ["hikE] (fin.) generally associated with
          trouble.
          The finalizer _kirue_ has the sense of repulsive, gruesome,
          grotesque, scary. Just about describes something regarded to
          be cursed or infected with a deadly contagious disease.

          Speaking of deadly (or not-so-deadly) things, we have:

          _kuesimani_ - masc. n. (neologism) puppy, dog.
          Cognates: _kuera_ - to laugh; _simani_ - guardian wolves. The
          san faran don't keep dogs for pets; they have "trainers" who
          keep wolves around their property to ward off thieves and
          intruders. (Although "trainer" is perhaps a misnomer -- the
          relationship between the trainer and the wolves is not amicable
          in any sense of the word. It's more of a subjugation by force
          deal than any real "training".) Hence, to them, any canine that
          is less ferocious is considered laughable and diminutive, hence
          _kuesimani_ for dogs. Dogs are also a foreign import to Fara,
          and not native; so this term is a relatively new coinage.

          And while we're on the topic of foreign imports, here's a phrase a
          foreigner might expect to hear rather often when learning to speak the
          local language:

          _tatari bei'aniin_ - "That's not how we say it". Also used to express
          disagreement - "That doesn't make sense!". A morphological
          breakdown may help understand where the sense is coming from:

          tatari bei-'aniin
          speech NEG-speak(FIN)

          Now, _aniin_ is the finalizer for the verb _tsana_ "to speak", and thus
          carries the sense of something being spoken, thus fully expressed and
          coherent. The negation thereof implies that something doesn't compute.

          It is important to note that this is NOT the usual way of negating a
          sentence; it has several peculiar properties:

          1) The obligatory (or not-so-obligatory?) case particle at the end of
          every NP is absent (and indeed, it's arguable whether _tatari_ here is
          even a proper NP);

          2) There is no verb or any of the other usual semantically-rich
          predicative words that might give a clue as to what this curious
          juxtaposition might mean;

          3) This bare-noun + finalizer construction is the hallmark of TF idioms
          like:

          peira ta'an
          rain down(FIN)
          It's raining. (Rain has come down? The rain downs?)

          kakari koko
          nonsense foolish(FIN)
          (interj.) Utter nonsense! Gibberish! The fool speaketh again!

          It is suspected that these two-word idioms are calcified expressions
          from proto-TF, before the case particles came to be an integral
          component of NPs. In any case (har har), the finalizer in such phrases
          seem to carry predicative meaning, and indeed, there is the non-negative
          expression:

          tatari aniin.
          speech speak(FIN)
          It makes sense; That's true; I agree!; Agreed! (Or in Russian,
          договорилось!)

          Here, the idiom conveys the sense that whatever has been spoken is
          finished, completed, and established, and so, derivatively, that an
          agreement has been reached. One may (crudely) transliterate it as "the
          word has been spoken!".

          So the negation thereof, _tatari bei'aniin_, implies that the words
          which were spoken, which should have come together, haven't, so instead
          of reaching an agreement or a completion of meaning, something doesn't
          compute or is otherwise incoherent. And so we have the sense "that's not
          how we say it", "it doesn't make sense", and derivatively, "I disagree".
          Of course, the distinction between correcting a foreigner vs. mere
          disagreement has to be judged from context. One learns to tell by
          hearing the phrase a sufficient number of times while learning the
          language. :-P :-P


          T

          --
          Trying to define yourself is like trying to bite your own teeth. -- Alan Watts
        • Roger Mills
          On Thu, May 02, 2013 at 06:59:52AM -0700, Roger Mills wrote: ... Well, tonight I did a little bit of counting-- not in the dictionary (several thousand
          Message 4 of 8 , May 9, 2013
          • 0 Attachment
            On Thu, May 02, 2013 at 06:59:52AM -0700, Roger Mills wrote:

            Teoh wrote:
            > What does the distribution of phonemes in yall's conlangs look like?
            > =================================
            > RM Aargh. I have no idea, though I've been meaning to try figuring
            > some out, in view of this recent thread. But it takes time.......

            Well, tonight I did a little bit of counting-- not in the dictionary (several thousand words!!), but in just the first two sentences of a translation exercise from some years back.

            There were just 61 words

            the count (in Kash alph. order)
            h [x] 1
            k 18
            ng [Ng] 7
            ç {S] 4
            c [tS] 13
            nj [ndZ] 3
            ñ 6
            y [j] 10 (not always phonemic, required in i,e__V and a__i,e, but phonemic        elsewhere)
            s 11
            t 16
            nd 4
            n 10
            r 20
            l 10
            f 1
            p 7
            mb 4
            m 15
            (w) 2 (non-phonemic, only in u_V or a_u)
            v 2
            a 64
            i 36
            e 14
            u 22
            o15

            The only surprise was the low number of /h/, I think a longer sample would have more.

            Anyway, now what do I do with this info? Haaaalp!!!
          • H. S. Teoh
            ... What do you do with it? Why, make the Kash equivalent of Scrabble, of course! With the tile scores scaled according to the above frequencies! ;-) Jokes
            Message 5 of 8 , May 9, 2013
            • 0 Attachment
              On Thu, May 09, 2013 at 08:37:45PM -0700, Roger Mills wrote:
              > On Thu, May 02, 2013 at 06:59:52AM -0700, Roger Mills wrote:
              >
              > Teoh wrote:
              > > What does the distribution of phonemes in yall's conlangs look like?
              > > =================================
              > > RM Aargh. I have no idea, though I've been meaning to try figuring
              > > some out, in view of this recent thread. But it takes time.......
              >
              > Well, tonight I did a little bit of counting-- not in the dictionary
              > (several thousand words!!), but in just the first two sentences of a
              > translation exercise from some years back.
              >
              > There were just 61 words
              >
              > the count (in Kash alph. order)
              > h [x] 1
              > k 18
              > ng [Ng] 7
              > ç {S] 4
              > c [tS] 13
              > nj [ndZ] 3
              > ñ 6
              > y [j] 10 (not always phonemic, required in i,e__V and a__i,e, but phonemic        elsewhere)
              > s 11
              > t 16
              > nd 4
              > n 10
              > r 20
              > l 10
              > f 1
              > p 7
              > mb 4
              > m 15
              > (w) 2 (non-phonemic, only in u_V or a_u)
              > v 2
              > a 64
              > i 36
              > e 14
              > u 22
              > o15
              >
              > The only surprise was the low number of /h/, I think a longer sample
              > would have more.
              >
              > Anyway, now what do I do with this info? Haaaalp!!!

              What do you do with it? Why, make the Kash equivalent of Scrabble, of
              course! With the tile scores scaled according to the above frequencies!
              ;-)

              Jokes aside, I did the TF analysis in order to decide on how to design
              the TF writing system, but IIRC Kash already has a script. So I'm not
              sure what else you can do with this info.


              T

              --
              Programming is not just an act of telling a computer what to do: it is
              also an act of telling other programmers what you wished the computer to
              do. Both are important, and the latter deserves care. -- Andrew Morton
            Your message has been successfully submitted and would be delivered to recipients shortly.