Output-output correspondence
• Phonology-morphology interface
– Level-related affixation
– Reduplication
– Hypochoristics
• Gradient attraction
• Syntax-morphology interface
– Case
– Passive morphology
– ECM constructions
– Coordinate structures
03.10.2015
1
Output-output correspondence
• Output-output correspondence was introduced by
McCarthy & Prince (1995) to account for
morphologically-based phonological effects.
• Instead of taking an input as a reference, a
morphological operation applies to a ready output, a
form which has already been through phonology.
• Faithfulness to input is ranked differently than
correspondence between two outputs.
03.10.2015
2
Output-output correspondence
Definition of Correspondence (McCarthy & Prince
1995:262):
Given two strings S1 and S2, correspondence is a
relation R between the elements of S1 and those of S2.
When a R b, the elements a of S1 and b of S2 are called
correspondents of each other.
03.10.2015
3
The notion of correspondence is vague. The correspondence relation takes its substance from a series of
constraints implementing the kind of relation needed in
each case:
MAX (no deletion), DEP (no epenthesis)
Additional constraints are:
IDENT(F), LINEARITY (saying something about the ordering
of the elements) CONTIGUITY (saying something about
the adjacency of elements), ANCHOR-Edge (about the
edges of the corresponding elements), HEAD-MATCH (if
one of the correspondent has a head, its correspondent
has the same head), etc.
03.10.2015
4
Input-Output faithfulness and Output-Output
correspondence
Input
(full model)
F
C
Output1
Output2
Prediction of the Correspondence Theory
IO-Faith >> Prosodic Constraints >> OO-Faith/Corr
or
OO-Faith/Corr >> Prosodic Constraints >> IO-Faith
Relations betw. Input and Output2 are assumed to be rare.
03.10.2015
5
Correspondence
/b 1 e 2 a 3 m 4 t 5 \ 6 /
 B 1 e 2 ÷a 3 m 4 t 5 e 6
B 1 e 2 a 3 m 4 t5 e 6
B 1 a 3 m 4 t5 e 6
A 1 u 3 t4 o 5
03.10.2015
NO
H IA T US M A X D E P I D E NT
*
*
*!
*!
*!
****
6
1. Reduplication
Reduplication is a morphological operation (often plural,
iterative, habituative, intensifier …) consisting in
copying (reduplicating part or whole of a stem).
According to McCarthy & Prince, only ‘authentic’
prosodic constituents (syllables, feet, prosodic words…)
can be reduplicants.
Ilokano Reduplicant Template: Heavy syllable (McCarthy
& Prince 1995)
s
/\
m m
03.10.2015
7
Reduplication in Ilokano
Reduplicant consists of a closed syllable
tra.ba.ho
trab - tra.ba.ho
‘work’
Red Stem
Reduplicant consists of a syllable with a long vowel
ró÷ot
ro: - ró÷ot
‘litter’
Red Stem
03.10.2015
8
Reduplication
Lardil Reduplicant Templates (McCarthy & Prince 1995):
Binary feet
F
/\
s s
03.10.2015
F
|
s
/\
m m
9
Reduplication in Lardil
Reduplicant consists of two syllables
[kele-th]
kele
kele-kele
‘to cut’
[pareli-th] pareli parel-pareli
‘to gather’
Reduplicant consists of a heavy syllable
[la-th]
latha
laa-la
‘to guide’
[˜aali-th]
˜aali
˜aal-˜aali
‘to be thirsty’
03.10.2015
10
Why does reduplication needs OOcorrespondence?
• In some languages, the segmental make-up (so-called
melody, a misnomer) of the reduplicant copies the
segmental make-up of the full form, rather than taking
its raw material from the input. Two cases:
- overapplication: a phonological process has seemed to
apply, though its context of application is not visible at
the surface (non-surface apparent)
- underapplication: a phonological process does not apply,
though its context of application is present at the
surface (non-surface true)
03.10.2015
11
Overapplication
In Javanese, there is a process of h-deletion taking place
intervocally:
Javanese h-deletion
Root
Root+my
an´h
an´h-ku
arah
arah-ku
03.10.2015
Root+Dem.
an´-e
‘strange’
ara-e
‘direction’
12
Overapplication
In reduplication this process takes also place in
environements other than intervocalic. The phonological
result of h-deletion is copied from the base to the
reduplicant.
Reduplication: Overapplication of h-deletion
bedah
bedah-bedah
beda-beda-e ‘broken’
dajøh
dajøh-dajøh
dajø-dajø-e ‘guest’
03.10.2015
13
Underapplication
In Akan, there is a process of palatalization. Coronals are
affricated before a front vowel, and /h/ is realized as a
palatal fricative.
Palatalization in Akan
t∫´
*k´
dΩe
*de
çi
*hi
03.10.2015
‘divide’
‘receive’
‘border’
14
Underapplication
In reduplication, though the vowel of the reduplicant is
always [i], no palatalization takes place. The consonant of
the base is faithfully copied.
Reduplication
ki-ka÷
*t∫i-ka÷
‘bite’
hi-haw÷
*çi-haw÷ ‘trouble’
03.10.2015
15
2. Different levels of affixation
It has been observed that affixes appear in a certain order,
and that they behave as classes of affixes w.r.t. this
property.
In English, besides other morphological operations like
compounding and inflection, two levels of derivational
affixation have been described.
- Level I affixes which influence the phonology of the
stem: -ic, -ation, -al
- Level II affixes which do not: -less, -ness, -y, -ing
03.10.2015
16
- Level II affixes are peripheral to Level I affixes.
(but see Fabb 1988 who showed that more restrictions are
at play than just ordering)
To account for this, Kiparsky, Mohanan and others
developed a model of Lexical Phonology, in which
morphology and phonology are interleaved:
Some morphology applies (level I affixation), then
phonology. Phonology consists of a set of ordered rules.
After completion of phonology, some more morphology
applies (level II affixation), then the whole phonology
applies again.
03.10.2015
17
Level II phonology has no access to morphological
information provided at earlier levels (and vice-versa):
we thus have a cyclic model of the morphologyphonology interactions (but see Mohanan who allows
loops in Malayalam).
When all levels have been completed (there may be more
than two), the so-called post-lexical phonology applies,
which is the sentence-level phonology. This phonology
is automatic, applies in all contexts, and doesn’t care
about levels. Final Devoicing in German is an example
of this type.
03.10.2015
18
Why does affixation needs OOcorrespondence?
OT has problems with the results of Lexical Phonology.
It can replace the set of ordered rules inside of each level,
but the levels themselves are more difficult to account
for.
Some examples:
03.10.2015
19
Why does affixation needs OOcorrespondence?
Level ordering of affixes (Benua 1995): New YorkPhiladelphia dialects (æ-tensing: E is tense)
Unaffixed
class [klEs]
mass [mEs]
pass [pEs]
03.10.2015
Class 1 Affix
classic [klæ.sik]
massive [mæ.sˆv]
passive [pæ.sˆv]
Class 2 Affix
classy [klE.si]
massable [mæ.s-]
passing [pæ.sˆ˜]
20
Why does affixation needs OOcorrespondence?
A standard kind of OT cannot account for the different
vowel in the stem of these words, due to the different
kind of affixation.
The alternation between the two kinds of vowels is due to
syllabification: Benua has the following constraint:
æ-tensing (*æC]s)
This constraint cannot be ranked as to deliver all forms
properly.
03.10.2015
21
Benua (1995) proposes to account for level II affixes with
correspondence to related outputs, in the examples
above class, pass, and so on.
Level I affixes take the input as input, and level II affixes
take the output of class and pass as inputs.
The faithfulness to the output, when relevant, is assumed
to be greater than the faithfulness to the input. This
explains why level II affixes do not trigger much
phonological changes in the stem.
03.10.2015
22
pE s / pæ s
OO -
* C]s *E
IO Fa it h
Fa it h
*
pE s
*!
p æs
p æss iv e/p Ess iv e
p æ .ss iv e
*!
pE . s s i ve
p æss in g / pE s s i n g
p æ .ss in g
pE . s s i n g
03.10.2015
*!
*
23
A second example of Benua:
condemn/ condemnation / condemning
-ation is a class 1 suffix and takes the input as base
-ing is a class 2 suffix and takes the output as base
03.10.2015
24
R ecursion (A)
/k a n dE m n/
a. k¦ n.d E m n
*m n] s IO- M AX OO 1 -D EP >>
*!
b.  k¦ n.d E m
*
c.
*
k¦ n.d E m
03.10.2015
25
R ecursion (B)
>> /k a n dE m n + e y•/ n
*m n] s IO- M AX OO 1 -D EP
a’. k¦ n.d E m.n e y •.n
b’.  k¦ n.d E m.n e y •.n
c’. k¦ n.d E .m e y .•n
03.10.2015
*
*!
26
R ecursion (C)
>> /k a n dE m n + IN /
*m n] s OO 2 -D EP IO- M AX
a'. k¦ n.d E m.nIN
b'. k¦ n.d E m.nIN
c'.  k¦ n.d E .mI N
03.10.2015
*!
*
27
3. Hypochoristics
A third kind of morphological process for which OOcorrespondence has been assumed is hypochoristic
formation.
A first example comes from the i-formation in German
which consist of a syllabic trochee, the unmarked (but
not the minimal foot) of German:
Prosodic Constraint on German i-formations
i-formations = F = [s's]
03.10.2015
28
Katharína –>
Káthi
Bénjamin –>
Bénni
Klínsmann –> Klínsi
Andréas –>
Ándi
Mánfred –>
Mánni
Wílhelm –>
Wílli
Wést/Ostdeutscher
–>
03.10.2015
Tóm –>Tómmi
Úlrich –> Úlli
Hirn –>
Hirni
Gabriéle –> Gábi
Wáltraud –> Wálli
Cornélia –> Cónni
Wéssi / Óssi
29
Many languages build hypochoristics in a similar way.
Prosodic Constraint in French
Hypocoristics = F = [s] or [ss']
True hypochoristics:
Véronique
Véro
Dominique
Domi, Dom, Do
Bénédicte
Béné
Elisabeth
Zabeth, Babé, Babette, Beth
03.10.2015
30
French also has:
Reduplications (Echo-words) = [s s']
/\
(C)V
père –> pépère, ours –> nounours, main –> main-main
The input to these reduplications is a monosyllabic word.
But the syllabification is not part of the input: it is an
added structure pointing to the fact that these
reduplications are faithful to an output rather than to an
input.
03.10.2015
31
IO-Faithfulness >> Prosodic Constraints >> BTFaithfulness
The emergence of the unmarked (TETU) is a landmark of
this pattern. The prosodic constraints in the middle are
responsible for the unmarked pattern of the language:
bisyllabic foot, trochaic pattern, open syllables …
03.10.2015
32
If the relation between input and output is active, the
unmarked form has no chance to emerge, since all kinds
of inputs are there, and faithfulness is high.
But the forms entering the relation OO have a chance to
emerge as unmarked, since the prosodic constraints are
higher.
Trochaic feet (iambic in the case of French, open syllables
and the like) emerge.
03.10.2015
33
Conclusion and open problems
1. Since correspondence is a vague notion, all kinds of
forms should be able to enter into a correspondence
relation. How can we delimit the desirable
correspondence relations from the undesirable ones?
2. OO-constraints lead to an explosion of the constraints.
3. OO-correspondence needs an existing output in order to
be workable. In some cases, surface forms seem to be
faithful to a form which is never realized as an output.
In those cases, we have opacity.
03.10.2015
34
Conclusion and open problems
4. Lexical Phonology, as well as all models using ordered
rules have no problems with opacity. The existence of
intermediate forms, neither inputs nor outputs, is a
natural consequence of rule ordering.
5. OT has big problems with those. Since no derivation
enters phonology, no intermediate step should ever be
needed.
6. We will see later on that alternative solutions have been
offered to the opacity problem.
03.10.2015
35
Gradient attraction
• If output output correspondence is needed anyway, why
not treat all kinds of morphological relationships as
output-output correpondences?
• This is the step taken by Burzio (to appear) in his
Gradient Attraction theory.
• Burzio claims that similar (output) representations
attract each other and that they do so gradiently. The
more similar they are, the greater the attraction.
03.10.2015
36
Modified OT (Burzio, to appear)
other representations
Input –>
Grammar
–>
Output
The other representations are forms which are related in
terms of morphemic parenthood or of analogy.
03.10.2015
37
Gradient attraction
• Allomorphs consist to a large extent of the same
segmental material and have (partly) the same semantic
representation.
• But they also contrast with each other in order to keep
their distinctness (Flemming’s dispersion theory)
• Gradient attraction is different from output-output
correspondence. One of the reasons os that allophonic
variations of complex words can be triggered not only
by the stem but also by the affix(es).
03.10.2015
38
Examples
• Stress position 1: titánic is attracted not only by títan
but also by barbáric and dynámic
• Stress position 2: módernist is influenced by módern
and not by the one of modérnity, because -ist adjoins
only to adjectival bases.
03.10.2015
39
• Segmental alternation: allophony of french gros, grosse
and gros ‘fat’ with liaison.
• According to Burzio, the third form is attracted by both
other forms, takes its vowel quality from the feminine
form and its consonant from the masculine (both facts
are unfortunately wrong! The vowel quality is always
the same, and the liaison consonant is voiced.)
• Steriade cites a much better example also from French:
an adjective like ancien ‘old’ has three allomorphs:
[ãsj´]~, [ãsj´n] and [ãsj´~n]. The liaison case takes ist
vowel quality from the masculine and ist vowel from
the feminine.
03.10.2015
40
OO-Correspondence in Syntax?
With syntax, there seems to be little
compelling evidence for the need for
output-output correspondence.
Possible evidence for OOcorrespondence in the syntax comes from
at least two domains:
the syntax-morphology interaction
coordinate structures
03.10.2015
41
Syntax-Morphology Interaction
Alternations that affect grammatical
functions tend to minimize differences
among the various construction types
In a representational model, this suggests an
influence of OO-correspondence.
One case in point is the rule for Case
marking in the German passive
03.10.2015
42
Case rules for the active clause
Nom:
Acc:
NPs bear nominative case
NPs that are not the
highest argument bear
accusative case
Dat:
NPs that are neither the
highest nor the lowest
argument bear dative
case
+Uniqueness, etc.
03.10.2015
43
Case rules for the active clause
Er kommt
he comes
nom
er sieht ihn
he sees him
nom acc
er gibt ihr es
he gives it to her
nom dat acc
03.10.2015
44
Case rules in the passive
What we find:
Es
wird ihr gegeben
it-nom
is
her-dat given
What we should get:
*sie wird es gegeben
she-nom is it given
03.10.2015
45
Case rules in the passive
A possible account:
Maximize faithfulness between the
active and the corresponding passive!
(00-correspondence)
The Alternative: Rule Ordering
1. Case potential is determined
e.g. by a lexical rule
2. „Absorption“ of the accusative
e.g. late in the syntax
03.10.2015
46
Case rules in other constructions
Similar ideas can be applied to
Complex predicates (retaining the Case
of the preposition)
jemanden
anwinken
someone.acc
at-wave
jemandem
zuwinken
someone.dat
to-wave
but ... is this OO?
03.10.2015
47
Case agreement 1
ECM-constructions and Case Agreement
interact in a fashion that may also be
understood in terms of OOcorrespondence
Case Agreement of some predicate
nominals
Ich bin ein Esel
I-nom am a-nom donkey
ich bleibe ein Esel
I-nom remain a-nom donkey
03.10.2015
48
Case agreement 2
Case Agreement of adverbials
er grüsst die Männer einen nach dem anderen
he greets the-acc men one-acc after the other
die Männer grüssen ihn einer nach dem
anderen
the men greet him one-nom after the other
03.10.2015
49
Case agreement 3
Predicates and some adverbs may take over
the Case of the noun phrase they are linked to
in terms of semantics ...
For ECM-constructions, we expect Case
agreeing expressions to always take over the
Case of the NP they are linked to.
03.10.2015
50
Case agreement in ECM-contexts
But there seem to be two dialects:
1. Ich lasse ihn einen Helden sein
I let him-acc an-acc hero be
2. Ich lasse ihn ein Held sein
I let him-acc an-nom hero-nom be
1. Agreement maintained
2. Nominative maintained
03.10.2015
51
Case agreement in ECM-contexts
1. Ich lasse die Männer einen nach dem
anderen ankommen
I have the men one-acc after the other
arrive
2. Ich lasse die Männer einer nach dem
anderen ankommen
I have the men one-nom after the other
arrive
03.10.2015
52
Case agreement in ECM-contexts
Solution 1:
OO-Correspondence between the finite
clause and the infinitive
Solution 2:
Case determination before nom > acc
change in the subject position of the
infinitive
03.10.2015
53
Summary
The Case effects described so far may
either be interpreted as being due to
OO-correspondence
lexical determination of Case, followed by
a syntax-triggered change
more complex Case rules
03.10.2015
54
Parallelism in coordinate structures
It may thus make more sense to look at a
construction type that bears some vague
resemblance to reduplication --conjunctions.
In principle, the two parts of a
coordination construction are fairly
independent of each other ...
.... this changes when they are affected
by a reduction operation.
03.10.2015
55
Parallelism in coordinate structures
Scope is a very interesting example for
this.
Independent
I introduced one of the boys to every
teacher
is scope-ambiguous:
ONE > EVERY
EVERY > ONE
03.10.2015
56
Parallelism in coordinate structures
I introduced one of the boys to every
teacher, and Bill did, too
involving a reducing coordination, is
two-ways ambiguous, NOT four ways, as
one might expect!
03.10.2015
57
Parallelism in coordinate structures
More examples
an American runner seems to have won a
gold medal, and a Russian athlete does,
too
the two indefinite NPs agree w.r.t.
specificity
one guard was seen in front of every
building, and a policeman was, too.
03.10.2015
58
Parallelism in coordinate structures
In an ellipsis/coordination reduction
construction, the scope relations among
the elements in clause A must be
identical to the ones in clause B.
In the Y-model of grammar, in which
phonology and semantics do not
communicate, this is difficult to account
for.
03.10.2015
59
Parallelism in coordinate structures
Across-the-board rule application was
invented in order to account for such
facts.
Who did you meet t and invite t
The parallelism facts fit neatly into OOcorrespondence, however.
03.10.2015
60
A special form of OO-correspondence
Perhaps, quite a different concept of OOcorrespondence is called for in syntax ...
Many syntactic approaches assume more
than one level of representation ...
Surface structure
Logical Form
Argument Structure
03.10.2015
61
A special form of OO-correspondence
It has been observed that UG tries to
minimize differences between these
levels.
This „economy of derivation“ may
reflect OO-correspondence between
different levels.
03.10.2015
62
A special form of OO-correspondence
From a single input, two, three or more
of such representations are generated.
Minimal Link (=superiority) effects may
reflect the attempt to minimize structural
differences between lor.s (Müller,
Williams)
03.10.2015
63
MLC 1
Who do you expect to say what
*what do you expect who to say
More relations of the premovement/declarative structure are
preserved in the former example
koj kogo mišliš
03.10.2015
who what saw
64
MLC 2
In the clitic (Wackernagel) position
weil er es ihr gibt
because he it her gives
pronoun order has been claimed to be
identical to base order ...
03.10.2015
65
Descargar

Output-output constraints and Emergence of the Unmarked