Re: Question Re: Reduplication

On Tuesday, September 23, 2003, at 02:58  AM, BP Jonsson wrote:

> At 10:22 22.9.2003 -0600, Dirk Elzinga wrote:
>
>> On Sunday, September 21, 2003, at 12:43  AM, David Peterson wrote:
>>
>>> Hi All,
>>>
>>> If you had a language with a pretty standard intervocalic voicing
>>> rule
>>> (let's say, /s/ > [z] / V_V), yet you had reduplication, how would
>>> that affect the voicing rule?
>>>
>>> I realize that it's simply a matter of rule-ordering and cyclicity,
>>> but I was just wondering what's more common among natural languages
>>> that feature reduplication and intervocalic voicing (or intervocalic
>>> anything, for that matter).
>>>
>>> Here's an example:
>>>
>>> Phonemically: /sopo/ > /sosopo/
>>> Phonetically: [sopo] > [sosopo] or [sozopo]?
>>
>> This was one of the hot topics in phonology when I was in grad school.
>> There are three possibilities: 1) overapplication, where phonological
>> processes applying to the base of reduplication also apply to the
>> reduplicant, even though the conditions for the application of the
>> phonological process have not been met ([sopo] -> [zozopo], showing
>> intervocalic voicing in both base and reduplicant), 2)
>> underapplication, where phonological processes which might otherwise
>> apply to the base don't because they can't apply to the reduplicant as
>> well (your example above of [sosopo]), and 3) normal application,
>> where
>> processes apply "normally" (your example of [sozopo]). What seems to
>> be
>> at issue here is the relative importance of making the reduplicant and
>> the base match. When this requirement is valued highly by the grammar,
>> you get either overapplication or underapplication; in both of those
>> cases, the reduplicant matches the base exactly with respect to the
>> specific property under investigation. There are a surprising number
>> of
>> languages which show either over- or underapplication; look at John
>> McCarthy and Alan Prince's papers "Correspondence and Reduplicative
>> Identity" (1995) and "Faithfulness and Identity in Prosodic
>> Morphology"
>> (1997); they provide several examples of each type and generate a
>> typology of over-/underapplying languages within Optimality Theory.
>
> Wouldn't it be of some importance whether there is a /z/ phoneme or
> if [z] is merely a contextual allophone?
It doesn't seem to be. One of the clearest examples of overapplication
is found in Madurese. In Madurese, nasality spreads rightward from a
nasal segment until it reaches an oral obstruent. It spreads to vowels,
glides, and through /?/ and /h/. Nasal vowels and glides are only found
as the result of nasal spread, and are thus predictable alternants of
oral vowels and glides.

In reduplication, nasal glides and vowels are also found, where the
corresponding portions of the base are nasalized:

[j~a~t-ne~y~a~t]  'intentions'
[on-so?on]  'request (noun)'

The second example shows that nasality really doesn't spread leftwards.
The only reason that the first example shows a nasal glide and vowel is
that they are nasalized in the base. By overapplication, the
reduplicant now matches the base with respect to this property.

In Tagalog, there is a well-known alternation by which a nasal-final
prefix and a stem-initial voiceless stop merge into a nasal consonant
with the same place of articulation as the stop:

pan-pu:tul -> pamu:tul (no gloss given)

Reduplication copies the /m/ of the base in reduplicated forms:

pan-RED-pu:tul -> pamumu:tul (no gloss given)

In Tagalog, /m/ is distinctive; in Madurese, nasalized vowels and
glides are not. So overapplication can target both distinctive and
non-distinctive segments.

Dirk
--
Dirk Elzinga
Dirk_Elzinga@byu.edu

"I believe that phonology is superior to music. It is more variable and
its pecuniary possibilities are far greater." - Erik Satie