Re: How to minimize "words" (was "Re: isolating conlangs")

---------- Forwarded message ----------
From: Jeff Rollin <jeff.rollin@...>
Date: 23-Feb-2007 02:47
Subject: How to minimize "words" (was "Re: isolating conlangs")
To: Constructed Languages List <CONLANG@...>

Hi David

On 23/02/07, David J. Peterson <dedalvs@...> wrote:
>
> Leon wrote:
> <<
> I think there is a subtle difference. Morphology only affects a word
> or  root neighboring it (or neighboring neighbor affix), while
> certain  words with grammatical usage may affect a word at the other
> end of the  sentence. Surely no one defines a word as being separable
> into  scattered parts all over a sentence?
> >>
>
> That's part of the problem.  No one *has* put forward a satisfactory
> definition of a word.  Even if one comes up with a satisfactory
> definition in one language, it probably will not apply to a different
> language.  Indeed, sometimes a single language will have different
> definitions of "word" within the same word.  Take "that's" in English.
> Most phonologists will agree that "that's" is a single phonological
> word, but that "that" and "'s" are two separate morphological
> words.  Are we to be satisfied with that?
>
> It might turn out to be the case that the notion of "word" isn't
> useful to linguistic analysis, along with the notion of "morpheme".
> At the very least, they are assumptions that should be questioned.
>
> -David
> *******************************************************************

I read somewhere that the notion of "word" only applies to literate
societies that attempt to segment written communication into definable
chunks,
insteadofputtingthemalldowninonelongsequenceliketheromansandgreeksusedtodo.
In that context, so-called "polysynthetic languages" would exist as a
distinct subset of human languages
onlybecauseAmericanistlinguistschosetoseethemthatway. (I don't know of any
language outside the Americas that is widely held as being polysynthetic to
the degree of languages like Lakhota.) - French is an example of a language
which retains an inflectional structure in writing, (inherited from Latin),
but which is /beginning/ to be widely regarded as polysynthetic.

----

My own conlang/artlang, provisionally called Vallian (Vn), attempts to
minimize the number of independent "words" in a sentence by using as many
suffixes as possible (the language is exclusively suffixing).

It's sketchy at the moment, but a rough outline would be as follows:


Phonetic structure
---------------------------

Phonetically, the language is inspired by Quenya and Finnish and some of the
Bantu languages - and to some extent by Hungarian, Turkish, Latin, and
Greek. It has a restricted set of word-final consonants, does not use
word-initial clusters, and has prenasalised, palatalised, labialised, and
aspirated variants of most consonants.


Morphosyntactic structure
----------------------------------

The language is (poly)agglutinative and polysynthetic and aims to be
exclusively suffixing; most nominals consist of a root and a gender affix;
most verbs consist of a root and various suffixes as detailed below; those
which do not either end in a restricted set of consonants, or add an
epenthetic -i or -a/a: (a colon after a letter represents an umlaut over
that letter).

The language uses double agreement where possible (verbs show agreement with
agent, patient, indirect object, and beneficiary; nouns and adjectives agree
in gender, number, and case; dependent nominals agree in GNC with their head
noun; in possessor-possessive constructions, the syntax of "the king's
horse" is "of-the-king his-horse"; in "compound verbs" composed of a root
and locative/directional postpositional suffix, the object of the verb also
takes an appropriate case (as in Hungarian).


Nouns
----------

Nouns have gender, a singular, dual, and plural, several cases (including
ergative, accusative, dative, and several locatives) formed by adding
suffixes to the nominative singular, the oblique singular stem, or the
plural stem as appropriate), suffixes for possession (my, our, their) (PN)
and predestination ("my book for him", as in Tundra Nenets) (PDN).

Nouns can take several case suffixes at once, either as a result of
"polyagglutination," as in Dravidian, where a phrase such as "[he walked] in
towards the source of the dreadful noise" might be expressed as "dreadful
noise-of source-towards-in", and multiple agreement/case stacking, where a
noun that depends on another noun takes GNC suffixes not only in accordance
with its own syntactic requirements, but also in agreement with its head
noun.

Prepositional phrases (such as "in the back of the car") are replaced with
either case-like suffixes or some sort of locative-noun---head-noun
compound.

I'm leaning towards agreement in GNC not being a requirement when the phrase
is contiguous, as in Korean. Finally, I'm thinking of adding "pronominal
affixes" (this-, that-, who-/what-/which-), and definite, indefinite, and
construct states, and nominal tense/aspect/mood and/or TAM suffixes to agree
with the verb in the clause.
("This-former/ex-wife-of-mine"/"One-of-these-former-wives-of-mine")

Whilst I'm not trying specifically to break language universals, IIRC I have
read that so-called "nominal TAM" is really only "nominal TM" or "nominal
AM", as nominals typically do not inflect for tense (or aspect, depending on
your point of view); so it would be nice to be able to explore doing just
this in Valian.


Adjectives
---------------

Adjectives agree in (most) particulars with their head noun. I haven't yet
decided whether this should include agreement for PN/PDN/TAM, (Tundra Nenets
apparently includes adjectival agreement for (PN), or whether the GNC/PN/PDN
suffixes should differ from those of nouns. (This would pose a significant
problem for the small(ish) phoneme inventory of the language, so I may have
to use multi-syllable suffixes for either adjectives or nouns.) Predicative
and "emphatic adjectives" ("The house is white"/"The house which is white")
are like stative/intransitive verbs (in main and relative clauses,
respectively).


Verbs
---------

Verbs inflect for {(in)transitivity, causativity, reflexivity, reciprocity,
evidentiality} (I have not yet decided which of these should be true
inflectional suffixes, and which derivational), {mood, aspect, tense}, and
{gender, number, exclusivity, politeness} in {four or five persons} for each
of {patient, agent, indirect object, beneficiary}. Other inflections/clitics
might include question, emphasis, and identity-of-/attitude-to-addressee
particles.

Verbs can incorporate arbitrary nouns, or be modified by adding affixes
(preferably suffixes) corresponding to locative nouns. A beneficiary or
other argument can be "promoted" by a series of applicatives. I am toying
with the idea of adding suffixes to express the number, gender, and case of
arguments to verbs, either as main verbs or in participles and gerunds, or
both.

As the language is inspired by Turkish/Finnish/Uralic (though hopefully it
does not take *too* much inspiration from them), nouns take predicative
suffixes to turn them into verbs ("he/she/it is a Vn", etc.), and there is a
negative verb ("Does he speak Vn? - He nots. [=He does not]"), which also
does the work of "am/is/are not" in "I am/{we,you,they} are/{he,she,it} is
not/there {is,are} not"-type sentences.

Adverbs
------------

I'm trying to get rid of adverbs by using "adverbial phrases", e.g. "derived
adverbs" such as "in a careful way" for "carefully", and even "at this
(present) time" for "now". (The former is even how adverbs in the modern
Romance languages got started.)


Unresolved Questions
--------------------------------

There are a few niggles I have come across.

I'm trying to get rid of pronouns; does anyone know of a language which uses
no pronouns (not even nouns instead of pronouns, like Japanese) at all?

I'm aware that (independent) pronouns are used for emphasis: Perhaps this
could be accomplished by using a set of "emphatic pronominal suffixes", or
an "emphatic suffix" which comes after or before the pronominal suffix? I'm
aware this poses a problem for "prepositional phrases", but I propose to get
around that by using (at least some of) the nominal case suffixes (or a
modification thereof) as bases for possessive affixes, as in Irish, Welsh
and Hungarian.

If I used a general "root" meaning "thing", or "object" (as in Bantu, -ntu,
where "ubu-ntu" means something like "thing for humanity") which could take
possessive, predestinative, gender, number, and case affixes, I could get
rid of possessive and demonstratives (so instead of "that book is mine for
him", you might say "that-book my-papery-object-for_him-is" or in Valian
order, "book-that thing-papery-my-for_him-is" - don't forget there's a
predicative ending!).

Gender/noun-class: I'm trying to move beyond the traditional
masculine/feminine(/neuter) gender distinction into a Bantu-inspired
noun-class system for things like professionals, one for languages, and one
for inhabitants ( e.g. as if one said "the wise-r bak-er", "the
passionate-an Itali-an", "broken-ish Engl-ish". Has anyone designed a
language like this? What do you do about the alternation between
roots/suffixes ending/beginning in a consonant or a vowel?

Orthography: I'm a little stymied by orthographical problems, too. As I
mentioned, I have prenasalised, palatalised, labialised, and aspirated
consonants. Writing those themselves is no problem: As a quasi-Uralic
language, Valian has no initial consonant clusters or voiced consonants
(with the exception of the prenasalised consonants, which have voiced
allophones, liquids and /v/, which is an approximant as in Finnish), so I
favour the following system for the prenasalised consonants:

1. *Initial* PNC is represented by the homorganic nasal (HN + the voiceless
consonant (VLC), e.g. stops (S) _mp-_, _nt-_

2. Since Vn has medial (M) consonant clusters (CC) of HN + VLC,  MPNC is
represented by HN + voiced consonant (VC) (this is convenient since MPNC
tend to be voiced anyway): e.g. stops _-mb-, -nd-_.

(In dia-/idio-lects in which the PNC are voiceless, they are distinguished
either by (a) the voiceless quality of HN; (b) the fact that in a CC, the
first consonant in the cluster belongs to the preceding syllable, and the
second to the following, whereas in a PNC, both consonants belong to the
same syllable; or (c) both (a) and (b)).

3. Since there is a single final consonant cluster incorporating a nasal
(viz.) -nt, representing PNC by HN + VLC (e.g. _-mp_) would be ambiguous; to
avoid ambiguity, I could represent final nasalised stops by either HN + VC (
e.g. _-mb_ or simply by VC, e.g. -b. However, to me this seems ugly; I'd be
interested in others' opinions.

4. Whilst using HN + VC presents no problem for stops and fricatives
(_-nd-_, _-nz-_, etc.), a problem arises when transcribing liquids, nasals,
and rhotics, particularly in HTML, since combinations like [nr] and [rm]
readily appear in the language anyway, and the Roman alphabet/HTML/ASCII.
Have others solved this problem?

5. Palatalised consonants, even in the face of words like "atja" /atja/
(bird) are easy, since the Roman alphabet has both j and y, which can be
used for either palatalisation or a [j] phoneme, and "j" is not used for
anything else (such as Z, the "s" in "pleasure"). However, combinations such
as "nyk" and even "nyj" are ugly and are apt to be pronounced by English
speakers as "nick" and "nidge" anyway - any thoughts? (Perhaps palatalised
consonants in clusters (and at the end of words) should be denoted by -j-
instead of -y-?)

6. Labialised and aspirated consonants also present a problem, since the
language can have both aspirated "t" and "t" followed by "h" (and other
combinations), and the Roman alphabet has no variations on h or w analogous
to the j/y split. Any suggestions? (Maybe I could make a rule that, say,
"lh" represents aspirated /lh/ and that an /l/ followed by a ("full") /h/ =>
"lk"?)

TIA

Jeff


--
Now, did you hear the news today?
They say the danger's gone away
But I can hear the marching feet
Moving into the street

Adapted from Genesis, "Land of Confusion"

http://latedeveloper.org.uk

Replies