Hallo conlangers!

On Mon, 11 Apr 2011 22:50:29 -0700, J. Burke wrote:

> After making my most recent post, I thought that perhaps some might be 
> interested in seeing how a gender system such as that of Hano?tsi or Mohawk 
> originates.  Below is a (fairly rough) description of the evolution of the 
> Hanoa?tsi system, inspired to some degree by Iroquoian.
> 
> To preface it, a little background: Hanoa?tsi is one of the descendants of 
> my (better-known) conlang Noyahtowa.  Noyahtowa has five persons (first, 
> second, third, fourth or obviative, fifth or farther obviative), two numbers 
> (singular, plural), inclusive and exclusive plural first person, animate and 
> inanimate nominal categories, and an indefinite.  Animate and inanimate are 
> not living and non-living as we consider them; rather, they're related to 
> spiritual power or importance (as in Algonquian); thus, e.g., a sacred pipe 
> or medicine bundle may be animate.

Nice and interesting.  I also have an animate/inanimate distinction
in my conlang Old Albic, and there too are some things which are
grammatically animate even though they aren't living beings, but I
haven't found out much of what they are.  (The whole Old Albic
lexicon requires more work yet.)

>       The Noyahtowa verb agreement system marks both subject and object and 
> is active alignment: intransitive participants are designated as agents (and 
> marked with so-called "subjective" finals/pronominals) when they initiate 
> and/or control the action, and as patients or undergoers (and are marked 
> with so-called "objective" finals/pronominals) when they don't.   Objective 
> suffixes appear on the finals/pronominals when the intransitive participant 
> is a patient.

Sure.  Quite common in North America, those beasts.  I also have it
in Old Albic (even though the language is set in Europe).

> The passage below discusses the evolution of the pronominal system from 
> Noyahtowa to Hanoa?tsi.  The context is the reshaping of the language's 
> inflectional system in the wake of a number of very destructive sound 
> changes (such as the loss of voiceless vowels) that chopped off and merged 
> many of the personal endings (finals) on the verbs.  I always imagined that 
> between the destruction of the old inflectional system and the development 
> of this new one that separate pronoun words became common to disambiguate 
> (whose use, of course, receded with the implementation of the new system);
> formerly, separate pronoun words served largely empatic functions.

Sure.  When a vital distinction collapses due to sound change,
some kind of replacement has to be developed.

> (A reference is made below to "pPCM," which is pre-Proto-Central Mountain, 
> the ultimate ancestor of all my Central Mountain languages, the family of 
> which Noyahtowa and Hanoa?tsi are members; it's an internally reconstructed 
> language which, owing to that fact, is more agglutinating and regular than 
> any natural language has a right to be; it's pure CV in form, with no vowel 
> length and only the most basic sounds.)

As reconstructed languages tend to be, especially internally
reconstructed ones.
 
> --------------------------------------------------------------------------
> 
> Both the pronominal system of Hanoa?tsi (H) and the means by which 
> pronominal ideas are expressed are radically altered from post-Classical 
> Noyahtowa (pCN).  First, pronominal marking by suffixes (finals) is 
> abandoned, and a new set of pronominal prefixes are innovated to replace the 
> finals.  The basic components from which these prefixes are built are (save 
> in a single case, the masculine, as described below) drawn from the old 
> intransitive objective finals, thus:
> 
> -γ˜ã- first person
> -ya- exclusivity
> -ž˜ẽ- inclusivity
> -wa- second person
> -nã- third person animate
> -hʔə- third person inanimate

How do these finals end becoming prefixes?  Perhaps a kind of copula
that is prefixed to the verb and then eroded by sound changes, as in

COP-S VERB-S > COP-S-VERB-S > S-VERB ?

> Additionally, the element -ʔã- was drawn from the old third person animate 
> plural intransitive objective final, where it was interpreted as a 
> pluralizer; and the elements -ł- and -š- were drawn from the old 
> intransitive objective finals where they functioned as objective suffixes.  
> The latter element was made the objective suffix for inanimate third 
> persons, while the former was made the objective suffix for all other 
> persons, following the patterns present in the old intransitive objective 
> finals.  

What are the functions of -ł- and -š- in Noyahtowa?  I guess they
are connected to animacy, if that is how they are used in Hanoa?tsi.

> The obviative, farther obviative and indefinite are lost in H; but the 
> singular vs. plural distinction of number becomes augmented by a third 
> number, the dual (=D), which refers specifically to two persons or things, 
> while the plural comes to refer to three or more persons or things.  The 
> element -ʔã- could mark either the dual or plural number: when it preceded 
> the element for person or clusivity, it functioned as a pluralizer (in 
> keeping with the pattern in the old third person animate plural intransitive
> objective final, where -ʔã- preceded -nã-, with the latter marking the third 
> person animate); when -ʔã- followed the element for person or clusivity, 
> however, it functioned as a dualizer.

This sounds weird to me.  The same element as dual or plural marker,
depending on whether it precedes or follows the person marker?  Not
impossible, but weird.  Is there a natlang precedent for that?

> The new pronominal prefixes were built up on this template:
> 
> plurality + person or clusivity + duality
> 
> The element -γ˜ã- marking first person is omitted from the non-singular 
> (dual and plural) first person forms; since there are elements that
> specifically mark exclusivity and inclusivity, and these distinctions only 
> exist in the non-singular first person, the presence of -γ˜ã- in those forms 
> was redundant and therefore unnecessary.

I see.  I could imagine *-γ˜ã-ya- and *-γ˜ã-ž˜ẽ- losing the -γ˜ã-
element, perhaps due to some sound change; and as the category of
clusivity is found only in the first person, no replacement for the
lost 1st person element is necessary.  A nice way to arrive at
irregular 1st person plurals.

> The new intransitive pronominal prefixes are thus:
>
> [paradigms snipped]
> 
> Among the intransitive objective forms above, the objective suffixes (which 
> immediately follow the subjective prefixes) -ł- and -š- appear as -la- and
> -ž˜ã-, respectively.  The reason for this is thus: in keeping with the 
> principles of word-formation, a connective vowel -a- was inserted between 
> the (originally) consonant-final intransitive objective prefixes and a 
> following element, and this connective became analyzed as part of the 
> intransitive objective prefixes.  Further, ł is voiced to l and š voiced and 
> nasalized to ž˜ by virtue of the phonological rules governing intervocal 
> obstruents present in the language at the time; likewise, the nasalization 
> of the connective -a- to ã in the third person inanimate forms, and the 
> nasalization of ə to ə˜ in the singular and plural third person inanimate 
> forms follows from phonological rules governing vowel nasalization.

So far, so good.

> Into the foregoing pronominal system was introduced the masculine gender 
> (=M) (see below for a description of typical gender referents). The 
> phonological element marking the masculine was -sẽž˜ə˜-, from an initial of 
> the same form that had originally meant 'customary, ordinary, plain' (pPCM 
> stem sasʊŋɔmɩ-, a detransitivized and stative derivative of the complex 
> unproductive initial -saŋɔmɩ- 'become accustomed to someone/something'), but 
> which had come to mean 'human being, person';

An interesting semantic shift.  Perhaps via the intermediate step
'mortal' (as opposed to immortal higher beings)?  'Ordinary' >
'mortal' makes sense, and 'mortal' > 'human' is attested in several
Indo-European languages (e.g., Persian).

>       with the introduction of the masculine gender, this initial came to 
> refer specifically to male human beings.

Has happened in many languages.

>       Three further pronominal prefixes were innovated containing the 
> masculine element (singular, dual, plural), on the model of the third person 
> animate prefixes.  Co-occuring with this development, the long-established 
> distinction of animacy faded from the language; the inanimate third person 
> forms came to express a neuter gender category (=N), while the
> animate third person forms came to express an indefinite gender (=I), 

From the lists of referents you give below, I conclude that the
neuter gender roughly compares to the Old Albic inanimate, while
the indefinite compares to the Old Albic common gender (Old Albic
subdivides animate into masculine (male beings), feminine (female
beings) and common (beings of generic sex, sexless beings,
collective entities, etc.)).

> filling the semantic hole that resulted from the previous loss of the 
> indefinite pronominal.  The addition of the masculine, neuter and indefinite 
> dimensions to the language created a semantic hole for referring to the 
> feminine (=F).  This hole was filled by using both of the original third 
> person prefixes (for animate and inanimate third persons, which by now 
> encompassed indefinite and neuter genders, respectively) to refer to the 
> feminine, with each prefix holding distinct connotations and having distinct 
> uses in that capacity, in addition to and alongside of their indefinite and 
> neuter meanings and uses; this re-analysis occured such that the indefinite 
> gender (originally animate third person) became a feminine/indefinite gender 
> and the neuter gender (originally inanimate third person) became a 
> feminine/neuter gender.  Thus, the four semantic gender distinctions of the 
> language (masculine, feminine, neuter, indefinite) are expressed 
> morphologically by only three sets of third person prefixes; those prefixes 
> are thus:  
> 
> [paradigms snipped]
> 
> Among the third person masculine forms, the masculine element -sẽž˜ə˜- 
> appears as -z˜ẽž˜ə˜- in the third person plural masculine prefixes 
> (subjective and objective); the s there is voiced and nasalized to z˜ by 
> virtue of the phonological rules governing intervocal obstruents present in 
> the language at the time.
> 
> Transitive pronominal prefixes, which express the co-occurence of an agent 
> and patient, consist of the intransitive subjective prefixes agglutinated, 
> thus:
> 
> agent + patient

Is it a typological rule that patient markers are placed closer
to the verb stem than agent markers?  I dimly remember such a rule
from somewhere, and follow it in Old Albic (where the patient
marker precedes the agent marker because they are suffixed).

> In all transitives, there is an inherent third person singular 
> feminine/indefinite patient (originally an inherent third person singular 
> animate patient); in such cases, the patient is not overtly expressed, and 
> thus the element expressing third person singular feminine/indefinite never 
> appears in the patient position of a transitive prefix (only when a third 
> person feminine/indefinite participant is dual or plural is it overtly 
> expressed in a transitive prefix).  This inherent patient common to all 
> transitives is changed or overridden by the presence of an element 
> expressing some other, different pronominal category combination in the 
> patient posistion of the transitive prefix.  

Old Albic has no inherent "default" patient in transitives; leaving
the slot empty detransitivizes the verb (functionally an antipassive;
a functional passive is formed by leaving the agent slot empty).
Does your language have a morphological antipassive?

--
... brought to you by the Weeping Elf
http://www.joerg-rhiemeier.de/Conlang/index.html