Download Phonological Theory I: Extrinsic vs. Intrinsic Ordering in Linguistics and more Study Guides, Projects, Research German Language in PDF only on Docsity!
Class 12: Process interaction: intrinsic ordering?
To do
- Think about your project!!
- Weâll take a break from problem sets this week. Thereâll be 2 more total: 1 tone & 1 stress.
- Next reading is Goldsmith, for Nov. 13 1. Before we get started, some stock-taking
Organization of course so far
- Pure-rule theory (Chomsky & Halle: SPE)
- Rules+constraints (Kisseberth, Sommerstein)
- Pure-constraint theory (Prince & Smolensky: OT) For each theory, and its variants, weâre asking how its predictions differ in terms of what phenomena should exist. We want to avoid both underprediction (failing to analyze an actual language) and overprediction (predicting phenomena that we donât think occur). 2. Potential paper topics
Here are some of the phenomena weâve seen where the theories make different predictions:
- feeding vs. counterfeeding: e.g., OT predicts feeding, one-shot simultaneous rule application predicts counterfeeding, SPE predicts both
- bleeding vs. counterbleeding: e.g., OT predicts bleeding, one-shot simultaneous rule application predicts counterbleeding, SPE predicts both
- iterative vs. non-iterative rule application
- interaction (or not) of multiple rule targets
- directional rule application
- myopic vs. fell-swoop derivations (see Walkerâs case below)
- constraint-specific repairs (see Spanish // case below)
- rule-ordering paradoxes; constraint-ranking paradoxes 3. Case from Rachel Walkerâs talk at SCOPHO this past weekend
Recall the basic metaphony pattern
stressed {e,o} [+high] / __C (^0)
+syll +high tense Vs rais e kals-ĂŠt-o kals- Ă t-i âsock (m. sg/pl)â mĂłv-o m Ăş v-i âmove (1 sg/2 sg)â lax or low Vs donât gĂĄt-o g ĂĄ t-i âcat (m sg/pl)â can spread across unstressed V Ăłrden-o Ăş rd i n-i âorder (1 sg/2 sg)â low V blocks spreading lavĂłr-a-v-a lav Ăł r- a -v-i âwork (1 sg perf/2 sg impf)â no spreading if thereâs no point ĂĄngol-o ĂĄ ng o l-i âangel (m sg/pl)â prseg-o prs e g-i âpeach (m sg/pl)â
In other words, spreading is ânon-myopicââit sees all the way to the end of the derivation, and if that derivation doesnât solve the fundamental problem of the unraised stressed vowel, then no spreading is done at all.
Overview of class: Extrinsic vs. intrinsic ordering SPE assumes that a language can impose any order it wants on rules. Many researchers have proposed that this is not the caseâthat at least sometimes, rules are intrinsically ordered.
Koutsoudas, Sanders & Noll 1974^1 : simultaneous repeated application, plus âproper inclusion precedenceâ
4. Simultaneous repeated application
= all rules apply simultaneously to the UR, then again to the result, and again until no more application is possible. This results in maximal application (feeding rather than counterfeeding, counterbleeding rather than bleeding).
5. Feeding example
Recall our previous analysis of Guinaang Kalingaâsomething like...
d+in+opana
- o Ă / VC__CV d+in+pana
- [+nas] (^)
place lateral / __
âsyll âc.g. place lateral
d+im+pana
(and something to deal with w )
o How would this work under simultaneous, repeated application?
6. Counterbleeding example
Our Polish example again: /voz/
- o u / __
âsyll +voice ânas
vuz
- [âson] [âvoice] / __# vus
o How would this work under simultaneous, repeated application?
By the way, Kiparsky^2 argued that historical changes in which rules change their order tend to result in feeding and counterbleeding.
(^1) Koutsoudas, Andreas, Gerald Sanders, and Craig Noll (1974). The application of phonological rules. Language 50:
1-28. (^2) Kiparsky, Paul (1968). Linguistic universals and linguistic change. In E. Bach & R. Harms (eds.), Universals in
linguistic theory. New York: Holt, Rinehart, & Winston.
Aside: if we adopt the analysis above itâs a bit of a problem for OT. Why is the problematic // resolved by changing place in one instance, and manner in the other? Even if we split up the markedness constraint, it doesnât help.
/ake/ * *# *V IDENT(place) IDENT(manner) *j# *lV a ake *! * b akel *! c akej * *
/ake+os/ * *# *V IDENT(place) IDENT(manner) *j# *lV a akeos *! * b akelos *! * c akejos *
The constraints at the bottom canât be doing the work, because *lV is violated all other the place (e.g., cielo ), and although violations of *j# arenât frequent, they exist (e.g., ley ).
This is the problem of âconstraint-specific repairsâ, which are predicted in OT or in some versions of rules+constraints, but not in OT.
Iâm not saying OT canât capture the Spanish dataâjust that the analysis couldnât be a direct translation of l / __ # and j.
8. Bleeding: example originally from Kiparsky
Schaffhause dialect of Swiss German: /bo/ /bod/ /bo+PL/ /bod+PL/
- V [âback] / complicated âumlautâ context, ---- ---- b bd including plurals
- o / __
+cons +cor âlat
(^3) ---- bd ---- ----
o Why is this ordering crucial?
o What happens if we use the Koutsoudas & al. approach?
K & al. propose that in all apparent cases of bleeding (and counterfeeding?), the rules need to be revised. In this case, they propose a context-free rule (remember Myersâs persistent rules).
o Apply this solution to /bod+PL/.
o What additional fact needs to be true in Schaffhause for this to work?
(^3) Actually, in the original itâs not [+cor] but [âgrave]. Grave is an acoustically based feature (roughly, lower
frequencies are stronger for [+grave] segments), not much used these days. Labials and velars are [+grave]; dentals and alveolars are [âgrave] (a.k.a. acute ).
9. The Elsewhere Condition (Anderson 1969, Kiparsky 1973 and ff.)^4
(It will come up again in Kiparsky 1982.)
Remember how rule schemas work: a schema expands into a set of rules, and they are ordered disjunctively : you get to apply at most one (except for (X)* schemas).
V [+stress] / __ C 0 (VC 0 )#
means
V [+stress] / __ C 0 VC 0 # and only if thatâs not applicable, V [+stress] / __ C 0 #
So /manok/ [mĂĄnok], /tinalot/ [tinĂĄlot], /tup/ [tĂşp].
Kiparsky argues that disjunctive ordering doesnât really have anything to do with expansion conventions (though it happens to occur those cases). He proposes that what really drives disjunctive ordering is the Elsewhere Condition:
(p. 94) âTwo adjacent [in the ordering] rules of the form A B / P __ Q C D / R __ S are disjunctively ordered if and only if: (a) the set of strings that fit [are nondistinct from] PAQ is a subset of the set of strings that fit RCS , and (b) the structural changes of the two rules are either identical or incompatibleâ
o What does the Elsewhere Condition say about our pair of stress rules above?
o Letâs discuss: How does the Elsewhere Condition compare to proper inclusion precedence? Are there cases where the two conditions apply differently?
We also need to define âincompatibleââprobably it means that the results of applying the two rules are distinct , in our technical sense.
[I cut some stuff here that we didnât get to, and moved it to next timeâs handout; and added a little more explanation in §10]
10. Excursus: back to non-derived-environment effects (NDEB)
Recall Finnish: t s / __ i /halut+i/ [halusi] but /tila/ [tila] but /vete/ [vesi] (ei/__# feeds ts)
(^4) Kiparsky, Paul (1973). âElsewhereâ in phonology. In Stephen Anderson & Paul Kiparsky (eds.) A Festschrift for
Morris Halle. New York: Holt, Rinehart & Winston.
