Theiling Online    Sitemap    Conlang Mailing List HQ   

Re: Wordless language (WAS: NonVerbal Conlang?)

From:Eldin Raigmore <eldin_raigmore@...>
Date:Monday, July 3, 2006, 22:05
On Sun, 2 Jul 2006 07:48:49 +0100, R A Brown <ray@...>
>>The content-morphemes include many whose semantics would strike a speaker >>of a natlang as "particles". > >Maybe - but while 'linkers' may be considered 'particles', I do not >think content-morphemes would be considered by most to be particles. I >would certainly not call them that. > >"Linkers" & "content morphemes" reminds me of the traditional Chinese >distinction between 'empty' and 'full' words. Whether one calls them >'linker words' and 'content words', or whether one would consider >'words' in that language to be composed of content & linker morphemes, >there will surely be elements that can be called 'words'.
Well, as I recall, in this particular conlang, its creator had such a small closed class of "linkers" or "operators" or whatever he/she called them, that very many of the words in English or other languages you would think of as "empty", got translated into the other class of words -- not the operators or linkers, but the class that contained all the "full" words. Without pointing you to the specific conlang it will be difficult to make you understand unless you are willing to help me explain it to you. I am sorry I don't have the particular document's URL available. I think it might have been an earlier version of Knibb's "T4"; but I could be wrong about that. But the secondary point I was making was that this conlang's idea of what were "full" words was much larger, and of what were "empty" words was much smaller, than we are used to in most natlangs or most other conlangs. The primary point was that all of its "full" words were 0-valent; all of its 2- valent words were "empty". And of course I can't say all of that without saying the language had "words". But it only had "words" according to some definitions, not according to others. It had no syntagma intermediate between the morpheme and the sentence. What I have called "words" above, were morphemes; and when one combined them, one couldn't stop until one had a sentence; else what one produced wouldn't be meaningful. --- On June 29, R A Brown <ray@...> wrote:
> Yes, I've been reading Trask again (never a bad idea), and he gives > three meanings to 'word' in _grammatical_ use (he explicitly does not > deal with 'phonological word' or 'orthographic word'). His third > definition is: > "In formal language theory, a string that is a member of a language; a > sentence." > > The other two definitions define more precisely what such a string is > grammatically; but I'll not bother with those at this point. > > The thing is that any sequential form of language is ipso_facto a > string; and we can thus have strings which are members of any utterance > or piece of writing. As I said, any ID language, and that surely > includes _all_ natlangs, must therefore have elements which may be > called 'words'. The question is the delimiting of words in a particular > language. > > As I see it, if a wordless language is possible then it must be fully > multi-dimensional (i.e. 2 or more _real_ dimensions - not a 1D language > cunningly chopped up and re-arranged in 2 or 3 dimensions, as we had > suggested in one of the earlier NLF2DWS threads). So it seems to me that > to answer the question whether such a beast can exist is to answer these > two questions in the following order: > 1. Is a non-linear fully multi-dimensional language possible? > 2. Does such a language have elements that can be called words, or are > its elements of a different kind?
I think _for_purposes_of_this_thread Trask's _third_ definition of "word" is not worth using more than once. It is worthwhile in its own right; and it was worth mentioning once, thank you, Ray; but the definition is such that every language has words by definition. This is analytic knowledge, not synthetic knowledge, in the terms of Kant's "Critique of Pure Reason" and "Prolegomena to Any Future Metaphysics". In other words, the statement that "every language consists of words" is true a priori, because of the definitions of the terms in which the statement is framed. Trask's other two definitions are more useful for purposes of answering the question "Can there be a language in which every or almost every utterance doesn't consist entirely or almost entirely of words?". Whatever definition of "word" we use in this thread -- and I hope we use more than one of them -- not much useful can be said if we use a definition that guarantees the answer must be "no" even before we begin any investigation (beyond pointing out that we shouldn't use that definition on this thread, no matter how useful it is in other discussions). ---On July 2nd --- In, Jonathan Knibb <j_knibb@...> wrote:
> I am still lurking, but haven't been following this thread very closely, > I'm afraid, interesting though it is. I can answer the queries about T4 > (well remembered!) though. It's moved on significantly if fitfully since > the 2003 description Ray found, but Ray's point that T4 was never > designed to be wordless is well made.
It may be, nevertheless, without "words" according to some definitions of "word" that better apply to other languages -- even to quite a few of them. (Of course _every_ language has "words" according to Trask's third definition; it is not necessary to actually know any languages to see that that must be true, only to understand the definition.)
> Eldin: > >>>The conlang I meant to mention ... had just two "parts of > speech" -- "linkers" (a small closed class) and content-morphemes. ... > [I]n this conlang there is no difference between morphotactics > and syntax. The content-morphemes include many whose semantics > would strike a speaker of a natlang as "particles". <<< > >That certainly describes T4 well. I didn't find it possible to reduce *all* >semantic elements to a single syntactic class without unacceptable >ambiguity, so T4 has two classes of element on the syntactic level. One >is closed ('linkers' or 'operators') and consists of three elements: an >element indicating the verb-object relation, its reverse (object-verb), >and a copula (realised as zero). The open-class words correspond to all >the content words, many of the function words, and many of the >bound morphemes of English. For example, most sentences begin with >a group of markers indicating number, tense, aspect and mood, but >the syntactic behaviour of each of these markers is identical to that >of any other open-class word. (DA-markers and delexical clitics have >been abandoned since the 2003 description.) > >As Eldin says, the syntax is not divided into morphology and phrasal >syntax; there is a single rule for combining elements syntactically, and >all >syntactic words are single morphemes. There are rather complicated >sandhi-type rules on the phonotactic level, and these operate on a unit >larger than the syntactic word but smaller than the sentence (and >which is not a syntactic constituent), which could be called the >phonological word if it weren't so discrepant with the syntactic word.
This makes it sound like the conlang I had in mind was indeed J. Knibb's T4. I might have been thinking of a version earlier than the current one. These are the features I had in mind; o Only two "word-classes", a very small closed one and a very large open one. o The "open class" contains many of the "bound morphemes" of other languages. o No division between morphology and phrasal syntax. o All "syntactic words" are single morphemes. I was not aware that there were any units larger than the "word" (= "morpheme" in T4, if "syntactic word" is meant, according to Jonathon's remarks above), but smaller than the sentence. Either this wasn't clear in the document I read -- sorry I can't find it -- which wasn't necessarily the complete description of T4 anyway; or it didn't show up in that version of T4, and Jonathon re-introduced it later.
> I hope this is relevant to the thread! Apologies if not. I have been > working on a T4 grammar sketch, but I'm entering the final year of my > PhD and have little time for such matters :(
I never got mine. I hope you finish yours. We'll miss you 'til you come back, but the degree is more important, I think. --- On July 2nd, And Rosta <and.rosta@...> wrote:
> That sounds like Jonathan's ("T4"?).
Thank you , And.
> I suppose it is pertinent to remark that one could without difficult > design a conlang in which there was perfect homology between > morphophonology and syntax, and only a single, uninflected, class of > content words, plus various uninflected function words. -- Not a > particularly groundbreaking idea, but it constitutes, I believe, the > maximum degree of simplicity that conlangers have so far managed to > conceive.
Has it been done? Who did it? Which conlangs? Is it reasonable to guess that they weren't very popular because most conlangers didn't find them sufficiently naturalistic nor sufficiently challenging? --- On July 2nd, "David J. Peterson" <dedalvs@...> wrote:
> > Eldin wrote: > << > Do Sign Languages constist of phones? > Are Sign Languages natlangs? > >> > > Yes and most definitely yes. There's been lots of work done > (fairly) recently on the phonology of sign languages. As an > analog, the place of a sign (its location in space and/or in relation > to the body) is similar to a consonant in spoken language; the > movement of a sign is similar to a vowel; and the handshape > one uses is similar to a tone (according to at least one theory). > I wrote an IPA for signed languages which may be a useful > introduction: > >
I'm going to look that up. --- On July 2nd, Kalle Bergman <seppu_kong@...> wrote:
>>Yes, of course - I had forgotten about sign >>languages. Sorry. >>AFAIK most (all?) are natlangs > > They most certainly are. > >>>Do Sign Languages constist of phones? >>> Are Sign Languages natlangs? >> >>They won't consist of phones - but signs languages >>are not things I'm familiar with. I do not know how they are analyzed. > >When discussing sign languages, the terms "phone" and >"phoneme" are used, although they - of course - don't >refer to anything sound-related, but rather to >different ways to align ones fingers, to move ones >hands and to place one hands in space. > >On the topic of sign languages: For a non-technical >description of sign language in general and its >history in particular, I can strongly recommend the >book "Seeing Voices" by Oliver Sacks.
Thanks, Kalle. Thanks, David. --- On June 30, John Quijada <jq_ithkuil@...> wrote:
>Interesting thread. As for Ithkuil, it is true that "word" is definitely a >secondary consideration and that the morpheme is king. However, I would say >Ithkuil retains some concept of "words"
It sounds like this could all also be true of Knibb's T4.
>in that word boundaries are >important in Ithkuil for parsing purposes so that a speaker/listener can >correctly interpret homophonous morphemes, e.g., a word-final vocalic >suffix -/a/ has an entirely different meaning than a word-initial >prefix /a/-. The only way to know which you're hearing is via stress/tone >patterns which indicate word boundaries. So in that sense, I would say >Ithkuil definitely retains "words".
These reasons why Ithkuil still not only has, but "needs", "words", sounds different from the reasons why T4 has them. What do John Q. and Jonathon K. think? --- On June 30, R A Brown <ray@...> wrote: [snip]
> What do you mean by 'particle'? (Another ill-defined word) Do you simply > mean 'bound morphemes'?
Jonathon Knibb's most recent post on this thread about his T4, manages to say what I meant to say, without using the word "particle".
> I note that Trask in fact gives three meanings of 'particle': > 1."Traditionally, any lexical item which exhibits no inflectional > morphology and hence is invariable in form; the term is used only in > connection with languages in which the open classes Noun, Verb and > Adjective do inflect." > 2. "In the grammar of English, one of the preposition-like items which > occur in *phrasal verbs*, such as..........." > 3. "A label typically applied to some more-or-less well-defined class of > uninflected words in the grammar of some particular language when no > more obvious label presents itself." > > Clearly meaning (2) is irrelevant here, as it is English-specific and it > just would not be sensible to ask the question: "What would a language > consisting only of particles be like?"
I see what you mean, and agree.
> Nor does the question make sense with meaning (1), as that has meaning > _only_ if a language also has flectional elements, i.e. by definition > such 'particles' cannot be the only elements in the language.
I see what you mean, and agree. Most definitions of which one can say " ... i. e. by definition ... etc. " are probably not useful in the discussion.
> That leaves only meaning (3).
> If Livagian
(It wasn't Livagian, as And and you and others have shown. Let me call it "Conlang X" for now.)
>consists only of particles it means that
[Conlang X]
>has only one part of speech
I don't see why there can't be two parts of speech both of which are particles. But I suppose what I "really meant" was that there were two parts of speech and both consisted of bound morphemes.
>and, as no obvious label can be attached to this part of speech, then the >term 'particle' is given it.
What would be a better label? How would it be obvious? Would "function words" vs "content words" be the useful division into two "parts of speech"? If all of them are "bound morphemes" then is it not true, in a sense, that every sentence consists of a single "word", and that there are no "words" except for such sentences?
> But I am at fault, I guess, in seeking to define 'particle' in this > context. In all three of Trask's definitions "particle" is a word,
Probably only due to the fact that Trask wasn't thinking of conlanging. If he had been considering the possibility of a language without words while writing his definitions of "particle", he might have phrased some of them so that they did not require a "particle" to be a "word". But I think even if he had been thinking of clitics, if he had agreed with your later remark in this post that a clitic isn't a word, he wouldn't have required particles to be words; because in many languages with "particles", some of the particles are usually "clitics". Not even as great an authority as Trask writes his dictionaries for the benefit of science-fiction creators and computer programmers. If he had kept such an audience in mind, he probably would never have finished his dictionary; but if he had finished it, it probably wouldn't have assumed that every particle is a word.
> therefore it does not make sense to question whether particles are > words. However, it was being questioned; therefore I assumed what was > meant was that category known as: empty words/ grammatical words/ > function words/ form words.
That's right. And of course I wasn't requiring that that category be restricted to "words". [snip]
>>>But the question of whether particles ('empty words') are "real words" >>>or not is surely just asking whether they are free or bound morphemes >>>and/or whether bound morphemes should ever be considered as words.
Whether or not that's all there is to it, clearly you're right that that is the main thrust of the question.
>>And what about clitics? Are they free or bound? Either way, are they >>words or not? > >Clitics, of course, are bound morphemes. The whole point of a clitics is >that it cannot occur unless affixed in pronunciation to another word. At >the phonological level they are, therefore, not words.
>But the way they are treated in writing varies.
[snipped some interesting stuff relevant to how clitcs are written]
>>>The point, as I understand it, is that these North American languages >>>have few, if any, unbound morphemes. They are bound in whole phrases, so >>>that 'word' and 'phrase' are more or less synonymous (at least by some >>>people's analyses). >> >>Perhaps even, "word" and "sentence" are more-or-less synonymous. >>John Q's Ithkuil could be an example. > >More likely, I think, 'word' and 'clause' or 'word' and 'phrase' in some >languages.
Yes, you are right. I meant to say "clause" instead of "sentence". I think in some of the polysynthetic languages -- at least in the most "extremely polynthetic" of them -- every clause is, or at least can be, a single "word". In languages like modern spoken French, as you say below, it is the phrase rather than the clause which appears to be a single "word", according to appropriate definitions of "phrase" and "word".
>Also we should remember that written & spoken forms of the >same language exhibits different behavior in this respect. For example, >in _spoken_ French phonological words are co-terminous with phrases and >sometimes whole clauses;
>>>If a language can be >>>morphologically analyzed then it seems to me that it must ipso_facto >>>have words. >> >>That could be true; IMO it remains to be proven, at least on this thread. > >Surely a string A must have at least one string B which is an element of >string A, even if in certain circumstances A and B are co-terminous. >What is there that needs proof?
If you so define your terms that you are guaranteed to be right before the investigation even begins, then, of course, you need no proof.
>>However general consensus among professionals seems to be that all >>natlangs do, in fact, have words. > >As I have said, both in this mail and a previous one,
>I do not see how it can be otherwise.
Does that mean you can't believe that anyone will ever show you how it could be otherwise? Nor that perhaps you will discover it on your own? [snip]
>As trees can be decomposed into strings, we are not IMO dealing with >_fully_ >1D languages.
That seems to be the consensus.
>>I think if you take any major definition of "word" you can find a natlang >>which doesn't have "words" by that definition, or, at least, one most of >>whose utterances are not mostly composed of such "words". > >I do not think this true. Can you please give clear examples? > >Trask, who is concerned only with _grammatical_ terms in linguistics, >gives: > 1. "A *lexical item*; a single item belonging to some *lexical > category*, having an identifiable meaning or grammatical function and > typically fairly consistent phonological shape, though possibly > exhibiting a certain amount of *inflectional* variation reflecting its > grammatical environment in particular senses. In this sense of the term > 'word', such forms as _go, goes, went, going_ and _gone_ are all forms > of the single word (lexical item) _go_." > > 2. A *word form*; a particular morphosyntactic form of a lexical item > occurring in a particular grammatical environment. In this sense of the > term 'word', the forms _go, goes, went, going_ and _gone_ are all > different words; in fact, it is usually regarded as representing more > than five word forms, since the _go_ of _I go to Rome on Tuesday_ and > the _go_ of _I want to go to Rome on Tuesday_ are regarded as > syntactically distinct in spite of their identical form, and much the > same is true of _going_ and _gone_, which also represent two > syntactically distinct forms." > > Trask also adds that it is essential to distinguish between these two > quite distinct senses of the term 'word', to which I say "Amen." He also > says quite explicitly that he is not defining 'phonological word' nor > 'orthographic word'. However, he does give a third definition, which I > have quoted before, but it seems to me needs to be quoted again:
> 3. "In formal language theory, a string that is a member of a language; > a *sentence*."
I don't think the definition "In formal language theory, a string that is a member of a language" has any use on this thread other than as a means to persuade the rest of us to quit contributing to it. If we adopt that definition, we have to admit we are talking about a "possibility" that is _by_definition_ impossible. [snip]
>>I am willing to bet that you could take any _two_ major definitions >>of "word" and find a natlang not composed of "words" by _either_ of them. > > Sorry, this seems to me - excuse the pun - just playing with words.
As indeed is all conlanging, unless the title question to this thread can be answered "yes".
> Because it may be possible to find a natlang where there is no element > corresponding to X's definition of 'word' does not mean, surely, that > the natlang does not contain elements that can be meaningfully called > 'words'.
Yes, you are right, and this was exactly my point. There is no natlang whose speakers don't think it has words. If we want to find a natlang that doesn't have "words", we have to use a definition of "word" that is not appropriate to that particular natlang. My point was there may be plenty of such definitions that are appropriate to plenty of other natlangs, but not to that one.
> It might, however, suggest that X's definition is at fault - it > will be if X attempts to give one grand definition to a term that > clearly has more than one meaning, even in linguistics.
Even if the definition is not at fault, it will not cover every natlang; that was what I was trying to say.
>>I imagine the same is true of most conlangs; especially of those intended >>to be spoken and heard. I think And Rosta's conlang, for example, misses >>several definitions of "words" but hits a few. > > Yes, but if it hits a few, then we cannot describe it as wordless!
Oh yes we can. We can also simultaneously describe it as "wordy". That was my point; a language can be "wordless" by some definitions, and not "wordless" by others. (More likely, it can be described as not having nearly every utterance composed nearly entirely of words.) Indeed many natlangs are. But every natlang is "made of words" by _some_ definition. Is it possible to make a conlang which doesn't consist of "words" by _any_ definition? No, because Trask's third definition quite cleverly closes off that possibility, by defining "word" to mean "something in a language". Thus only an empty conlang could have no words.
> It seems to me that in fact this thread is discussing two quite > different things, which to some extent is causing confusion: > > 1. What do we mean by 'word'? (We mean more than one thing IMO)
> 2. Is it possible to have a language in which there is no elements that > could be considered a 'word' in any of the normally accepted senses?
Right. But we'd have to consider each normally accepted sense separately; or perhaps to consider them in groups. We couldn't consider them all at once, because then we would have to include such definitions as Trask's third; definitions which would make it impossible by definition to have a language without the thing defined. [snip] ----- eldin


R A Brown <ray@...>