Re: THEORY: Question: Bound Morphemes

dirk elzinga wrote:

-----<snip>-----
>>
>> In that case, I think I'm dealing with affixes. But what if I'm
>> dealing with an accent pattern that is phrasal rather than local?
>
>You are dealing with affixes iff the determiner must attach to a stem =
of
>a particular category. The distinction between phrasal and local accent
>doesn't matter here, unless the determiner carries its own stress, and
>the stress pattern of the word to which it attaches remains unaffected.
>In that case, the determiner is probably a word in its own right.
OK... the determiners do not carry their own stress. They are all=20
light syllables and only heavy syllables carry any significant stress=20
in Boreanesian. Furthermore, they are always attached to a stem of a=20
particular category; nouns (including nominalized verbs). So all in=20
all, they must be prefixes.

>> In Boreanesian, this could especially be a problem in verbal phrases
>> because they tend to be quite long. Many of the morphemes that make
>> up the verb phrase itself is linked to the verb (nominalized verb)
>> via the genitive. For example (with each morpheme separated by a
>> space for the sole purpose of this example):
>>
>>      /t[@ s@k@:h ?@n@Nh kijh m@nuw?/
>>      the  place  eat    1    chicken
>>      lit.: 'my eating-place of chicken'
>>
>> Phrasal stress falls regularly on the final syllable. In the example
>> above for instance, it falls on the syllable /nuw?/. Secondary
>> phrasal stress falls on all heavy syllables before the final, and on
>> all odd minor syllables before a heavy syllable. Does this mean that
>> the above phrase should be written as one word?
>
>I would agree with what has been said by others; this is largely a
>matter of personal taste. One convention you may wish to adopt in
>romanization is to mark affix/stem boundaries with a dash "-", and a
>clitic/host boundaries with an equal sign "=3D"; this has the advantage =
of
>clearly marking the type of morpheme which has combined with a given
>stem. Or if you would prefer not to have your transcription peppered
>with dashes and equal signs, let affixes be tacitly connected to their
>stems, but separate clitics from their hosts with a dash. You don't get
>so many non-alphabetic characters, but still acknowledge the special
>status of clitics.
I think I can do without dashes and equal signs quite safely. Most of=20
the morphemes in Boreanesian are composed of a heavy syllable which=20
is optionally preceded by a light syllable (like Mon-Khmer=20
morphemes). Very few are composed of just a light syllable. I have=20
now established, thanks to you, that these few light syllabic=20
morphemes are either prefixes or infixes. All this makes it possible=20
to separate the various morphemes since most morpheme boundaries will=20
be marked by a heavy syllable.

Alright! I think I'm beginning to understand all this. But just to be=20
really sure of things, I'll try to categorize each of the morphemes=20
in my example (repeated below for reference). Please tell me if you=20
think I got it right.

The phrase is composed of the following morphemes:

     /t[@ s@k@:h ?@n@Nh kijh m@nuw?/
     the  place  eat    1    chicken
     lit.: 'my eating-place of chicken'

...and these morphemes are:

     /t[@/ - Determiner prefix, because it attaches to a nominal=20
     stem.
     /s@k@:h/ - Locational verbal nominalizer prefix, because it=20
     attaches to a verbal root.
     /?@n@Nh/ - Verbal root.
     /s@k@:h?@n@Nh/ - Nominalized verbal stem.
     /kijh/ - Possessive 1st-person clitic, because it attaches to a=20
     position in the phrase but is bound phonologically to some word.=20
     In the example, it takes the position of an agentive possessor.=20
     If I were to replace /kijh/ with /kajh/ 'man', then /kajh/ would=20
     not be a clitic but a separate word. With regards to /kijh/=20
     being phonologically bound to some word, in the example, it=20
     binds phonologically to the nominalized verbal stem. But if, for=20
     instance, an attribute (adjective) was added, it would bind=20
     phonologically to the attribute.
     /m@nuw?/ - Nominal root preceded by a genitive zero-marker.

Thus, the example is compose of two words:

     /t[@s@k@:h?@n@Nhkijh m@nuw?/

>> I have heard of languages, like the Polynesian languages, where
>> authorities cannot agree what constitutes a word because boundaries
>> tend to be fluid. I fear that this may be the case with Boreanesian.
>
>'Word' is a fuzzy term, at best. Do you mean a linguistic object which
>is manipulated by the syntax? Or do you mean the domain of stress and
>accent rules? Or do you mean the end result of affixation? Little =
wonder
>people can't agree on what a word in Polynesian languages is, when we
>can't agree on what a word is!
No wonder I was (and probably _still_ am) confused! But I'm glad a=20
professional linguist couldn't resist jumping into this one.

Thanks!

-kristian- 8)