R A Brown, On 30/06/2006 11:37:
> Eldin Raigmore wrote:
> [Replying to me]
>>>> FWIW, I think it's worth putting aside the question about whether
>>>> particles are "real words" or not, unless we're to talk about a
>>>> language that would somehow be entirely composed thereof. (What would
>>>> it be like?)
>>  
>> Are you saying And Rosta's conlang is _not_ worth considering in this 
>> thread -- at least not yet? 
> 
> I am very puzzled by this question. AFAIK And has not yet given a 
> comprehensive description of Livagian, but there is an example given on:
> http://www.valdyas.org/irina/relay5/livagian.html
> 
> This is clearly not all particles in the sense I was using the term. 
> gghoekhg = 'is foul' and gkhnqehr = 'originates at' look like lexical 
> items to me.

Ray is quite right that I've not given a comprehensive -- or even a patchy -- description of Livagian: it's because the language keeps on getting demolished and rebuilt before I ever have time to produce publishable documentation. So my conlang is probably not worth considering in any thread at all, unless the thread is about conlangs that have been laboured over for decades without yet being published. And certainly Eldin is being terribly overoptimistic in thinking it reasonable to ask somebody other than me for their views on some technical aspect of Livagian...

The relay text Ray finds (and Google finds little else) is in a subsequently demolished version of Livagian. 

As for whether Livagian is, as Eldin implies, constituted entirely of particles, that's not a claim I've made, and the picture (as of June 2006) is as follows.
1. Parsing formally proceeds incrementally pretty much segment by segment (-- phonological segment), which is probably why Eldin cites Livagian as relevant. Also, and perhaps also relevant to Eldin's citation, the terminal nodes in the parse tree do not necessarily correspond one-to-one to chunks of the sentence's phonological form: the correspondence can be many-to-one or one-to-none or many-to-none.
2. Nevertheless, it is pretty straightforward to divide a sentence up into phonological words that consist, morphologically, of a bound stem, a bound suffix and sometimes a series of enclitics (and perhaps sometimes a proclitic particle).
3. In a pattern familiar from, say, many Indo-European languages, the typical lexical item consists (in its phonological exponence) of a stem plus a 'declension' (i.e. which of various paradigms the suffix is taken from). (There are 18 declensions, so 18 lexical items per stem.)

So as with natlangs, there is no obvious single candidate for what to call a 'word', and I don't think Livagian adds to the debate in any significant way, gratifying though it is that Eldin might think otherwise.

Ray continues:
> What do you mean by 'particle'? (Another ill-defined word) Do you simply 
> mean 'bound morphemes'?
> 
> 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?"
> 
> 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.
> 
> That leaves only meaning (3). If Livagian consists only of particles it 
> means that Livagian has only one part of speech and, as no obvious label 
> can be attached to this part of speech, then the term 'particle' is 
> given it. It may be that And has actually said this; but I have no 
> recollection of his doing so. However, I have no doubt And himself can 
> tell us    :)

I think a linguist doing fieldwork on 2006 Livagian would say that it has one part of speech and that it is inflected. There are natlangs a bit like that, I think.

>>> As far as I can see, it would convey no meaning. If we have particles
>>> only, there are no lexical words. 
>>
>> AIUI that applies to And Rosta's conlang.  
> Not AIUI - see above.
> 
> (Of course we should ask him.)
> Indeed.

It is certainly the case that Livagian has a lexicon.

In the context of the present discussion, the only way in which Livagian is egregious from the common flock of natlangs is in the way that sentences are parsed segment by segment. 

>> I don't think And Rosta's conlang is actually "unsatisfactory".
> 
> No one has said it is "unsatisfactory".

Its appalling incompleteness and lack of public documentation is profoundly unsatisfactory. I would have every sympathy with any reader of the list who was irked and frustrated to see public discussion devoted to a language so thoroughly unavailable to public scrutiny.

>> But if it had native speakers, I'm pretty sure they'd think it had words.
> 
> The text given on the web page I cited above is clearly broken up by 
> white space interpuncts.

The orthography has changed since then, and white space interpuncts are not used, except, optionally, between sentences. 

As for native speakers of Livagian, they are (I declare) conscious of morphophonological words, but not of syntactic words. IMO that picture is true of natlangs too: the word is a morphophonological entity, not a syntactic one.

--And.