Re: Telona grammar, part 2

> > Actually, I think the better question is: is this possible?  How
could
> > one conceivably make a language in which ANY sequence of morphemes is
> > well-formed?  If such a language were to exist, it would have to
accept
> > sentences like this just-created example:
>
> > pa       toko rana    but    gaw     dom pes  hama    zeju.
> > PAST PL    SUBJ DAT QUOT fish  take bubble bite.
>
> %%     Past and subjunctive markers could occur anywhere in a
> sentence without impairing understanding.    Also, the target language
in
> question would not necessarily have dative, plural, or "QUOT"
(referring to
> direct quotes?) morphemes.
Good points, all.  And yes, the QUOT particle is used to introduce direct
quotes (my conlang, Yivríndil, has one, which is why I probably thought
of this.  Yivríndil, oddly enough, doesn't have a way to do what we call
indirect discourse.)  Note, however, that even if a language has no
morphemes for dative and plural, it must have ways of expressing these
ideas.

>   So let's look at what's left, which I'll
> render strictly in glosses, since this is a nonce language.
>
> FISH    TAKE    BUBBLE     BITE
>
> As you probably realize, we're talking about sentences that are
> *syntactically* well-formed, not necessarily semantically
> interpretable.
I overlooked this initially.  An important point.  And in fact, once I
look at this sentence with an open mind, I *can* interpret it: "The fish
takes a bubble and bites it."

> Also, we're not excluding the possibility that word order could
> change the meaning of the sentence;   we're merely saying that all the
word
> orders would be grammatical.   Also, we are not excluding a lexicon
divided
> into two or more word-classes (like nouns vs. verbs).
I think a divided lexicon would provide you with the most problems,
actually--even though I think every natlang has more than one lexical
category.  This may be the intractable problem--how does this language
interpret strings of words in the same class?  If I say:

FISH BUBBLE ABSOLUTION COW CALCULUS

 . . . what relationship can these elements hold?  The easy answer, which
I consider a cop out, is to say that it's a conjoined set of existential
sentences: "There exist fish, a bubble, absolution, a cow, and calculus."
 You can obviously deal with verb strings the same way:

UNDERTAKE CARRY SCRIBBLE BLEND CONFUSE

 means "Someone undertakes to carry and scribble and blend and confuse."
But this isn't interesting, or even all that unusual.  If a friend walked
up to me and simply named a bunch of nouns, I'd probably say "What are
you talking about?", but I wouldn't consider his sentence ungrammatical.
The real challenge would be to come up with something in which long
strings of nouns have an interesting structure and still are
interpretable.

In fact, there *is* a class of natlangs in which any or nearly string of
words is grammatical--pidgins.  Early stages of pidgins are known for
their loose word-order principles and lack of grammatical operators,
which is strikingly like this.  A lot of pidgin sentences do essentially
boil down to a couple of nouns and a couple of verbs thrown together at
random, and with enough context they're still understandable.  But this
feature is what makes pidgins unstable and primitive--and most
importantly, this property is never retained in the creole.  When pidgins
are nativized, they suddenly start rejecting some sentences as
ungrammatial.

So I retract my earlier objection, and offer another one.  It's perfectly
possible to create a language in which every sentence is syntactially
well-formed, but it would be an unstable and non-native grammar.  You're
still welcome to try, though!

Jesse S. Bangs     Pelíran
jaspax@ juno.com
"Skin and tragedy always attract a crowd."
    --Pedro the Lion

Reply