Mora Augmentation Processes in Japanese

Mora Augmentation Processes in Japanese
Stuart Davis & Isao Ueda
Indiana University & Osaka University of Foreign Studies
Abstract: Japanese displays a variety of mora augmentation processes. These processes typically
involve lengthening a word by a single mora, usually for emphasis. In this paper we focus on two
such processes: the formation of emphatic adjectives in the Shizuoka dialect and the formation of
intensified adverbs in standard Japanese. We first examine adjectival emphasis in the Shizuoka
dialect. We show that adjectives can be augmented by gemination, nasal insertion, or vowel
lengthening and that there is a priority to the means of augmentation. We offer an optimalitytheoretic analysis of the mora augmentation process. We then contrast the adjectival pattern with
the augmentation process found with intensified adverbials in standard Japanese which has the
same priority of means of augmentation, but with some interesting differences in the details. We
offer an analysis of the intensified adverbials showing how a unified analysis of both mora
augmentation processes can be offered under optimality theory. We conclude by briefly
discussing some other instances of mora augmentation in Japanese.
[Area of interest: Phonology]
1 Introduction
Japanese displays a variety of mora augmentation processes. These processes typically involve
lengthening a word by a single mora, usually for emphasis. In this paper we focus on two such
processes: the formation of emphatic adjectives in the Shizuoka dialect and the formation of
intensified adverbs in standard Japanese.
In Japanese adjectives can be emphasized by the addition of a mora. In standard Japanese
this additional mora is typically realized by the lengthening of an accented vowel, as in kataai
the emphatic form of katai ‘hard’ where the second vowel is accented. The realization of the
additional mora, however, may vary among dialects. In this paper, we will first consider the
rather complex pattern of mora augmentation found in adjectival emphasis in the Shizuoka
dialect. We will present and describe the data which are not generally known to phonologists or
to others working on Japanese linguistics. We will then show how the various instantiations of
mora augmentation can be accounted for under an optimality-theoretic approach. Afterwards we
will compare the Shizuoka pattern of adjectival mora augmentation with that found in standard
Japanese for intensified adverbs. These forms have been discussed by Kuroda (1965), McCawley
(1968) and Hamano (1998) as well as by Lombardi (1998) from an optimality-theoretic
perspective. The two patterns are similar but not identical. The Shizuoka pattern of adjectival
emphasis is more revealing in that adjectives are not phonotactically restricted to a (C)VCV
shape as are the adverbs. We will compare the two patterns from an optimality-theoretic
perspective and show how they can be accounted for in a unified way. We conclude by briefly
discussing some other instances of mora augmentation in Japanese. Our data on the Shizuoka
dialect come from Hino (1977), Yamaguchi (1987), as well as work with native consultants. The
pattern we describe is found among older speakers in a fairly large area of central Shizuoka
prefecture.
2 Data and Description of Emphatic Adjectives in the Shizuoka Dialect
Descriptively, in the Shizuoka dialect, an adjective can be emphasized by augmenting the first
syllable by one mora. The precise way in which the mora is realized, though, varies depending
on the phonemic makeup of the input adjective. Consider the representative data in (1)-(3)
illustrating the different ways that the mora can be realized.
(1)
Adjective Emphatic Form
a. hade
hande
b. ozoi
onzoi
c. yowai
yonwai
d. hayai
hanyai
e. karai
kanrai
f. nagai
nagai
g. kanašii
kannašii
h. amai
ammai
Gloss
showy
terrible
weak
fast
spicy
long
sad
sweet
(2)
Adjective
a. katai
b. osoi
c. takai
d. atsui
e. kitanai
f. kusai
g. ikai
Gloss
hard
slow
high
hot
dirty
stinky
big
(3)
Adjective Emphatic Form
a. zonzai
zoonzai
b. kandarui
kaandarui
c. onzokutai oonzokutai
d. suppai
suuppai
e. okkanai
ookkanai
Emphatic Form
kattai
ossoi
takkai
attsui
kittanai
kussai
ikkai
Gloss
impolite
languid
ugly
sour
scary
In (1), augmentation occurs by inserting a nasal consonant after the first vowel (or after
the first mora). Notice that the inserted nasal surfaces before a voiced consonant. In (2),
augmentation ensues by geminating the consonant after the first vowel (or mora). Notice that the
geminated consonant in (2) is always a voiceless consonant.1 In (3), when the first syllable of the
adjective is closed, augmentation results from vowel lengthening rather than by nasal insertion or
gemination. In none of the data does augmentation occur beyond the first syllable. Thus, in (3c)
for example, vowel lengthening occurs in the first syllable, rather than gemination of the
voiceless consonant later in the word. The generalizations on how mora augmentation is realized
can be summed up as follows: If the initial syllable of the adjective is (C)VC then mora
augmentation occurs by vowel lengthening as in (3); otherwise, augmentation occurs by means
of nasal insertion before a voiced consonant as in (1) or by gemination of a voiceless one as in
(2).2 Consequently, vowel lengthening is the least preferred means of mora augmentation. It only
occurs in CVC syllables; gemination or nasal insertion in such syllables would result in a highly
marked CVCC syllable. In the following section we will develop an optimality-theoretic analysis
of the mora augmentation data.
3 Analysis of Emphatic Adjectives
One issue that immediately arises for an analysis of emphatic adjectives is the question of the
representation of the morphological indicator of the emphasis. We posit an analysis whereby the
emphasis is morphologically indicated by a single floating mora. This is shown in (4). The mora
is realized toward the beginning of the word because of the alignment constraint given in (5).
(4)
(5)
Representation of the emphatic morpheme as a floating mora
µe (µe = emphatic mora)
Alignment constraint for the morpheme µe
Align-L(µe,Wd) -- Align the emphatic mora with the beginning (left edge) of the word.3
For ease of reference we will indicate the emphatic mora as being in prefixal position with the
understanding that it need not be anchored in that position.
A second issue that arises is the question of what the emphatic adjective is based on. We
posit that the emphatic adjective is in an output-output relation with its nonemphatic counterpart.
That is, the fully prosodified nonemphatic adjective serves as the base for the emphatic form. In
(6) we show the input of the emphatic adjective. It consists of a base which would be the
moraified (nonemphatic) adjective plus the floating mora (indicated in prefixal position).
(6)
Input of the Emphatic Adjective (example from 2a)
µe µ
µ µ
|
| |
k a t a i
output = [kattai] ‘hard (emphatic)’
In the representations in (6) and elsewhere in this paper we show only the mora structure.
However, we do assume that the words are syllabified and that following such works as Ito
(1986) the only permitted coda consonants in Japanese are either a nasal or the first part of a
geminate. With respect to the syllabification of vowel sequences, we consider a sequence of two
identical vowels (i.e. a long vowel) to be tautosyllabic. A sequence of two non-identical vowels,
such as [oi], is considered a tautosyllabic diphthong (McCawley 1968, 1978). For example, a
word like /oišii/ ‘delicious’ has the syllabification [oi.šii]. (The issue of the syllabification of a
diphthong or long vowel in a potentially trimoriac syllable, CVVC, is discussed later, following
(25).)
Given this as background, we will present a formal optimality-theoretic analysis of the
data in (1)-(3), introducing the constraints and constraint rankings as we consider the relevant
data. First, though, we reconsider the data in (1)-(3) which we repeat below for convenience.
(1)
Adjective
a. hade
b. ozoi
c. yowai
d. hayai
Emphatic Form
hande
onzoi
yonwai
hanyai
Gloss
showy
terrible
weak
fast
e.
f.
g.
h.
karai
nagai
kanašii
amai
kanrai
nagai
kannašii
ammai
spicy
long
sad
sweet
Emphatic Form
kattai
ossoi
takkai
attsui
kittanai
kussai
ikkai
Gloss
hard
slow
high
hot
dirty
stinky
big
(2)
Adjective
a. katai
b. osoi
c. takai
d. atsui
e. kitanai
f. kusai
g. ikai
(3)
Adjective Emphatic Form
a. zonzai
zoonzai
b. kandarui
kaandarui
c. onzokutai oonzokutai
d. suppai
suuppai
e. okkanai
ookkanai
Gloss
impolite
languid
ugly
sour
scary
In considering the data in (1) we note that, hypothetically, there are many different ways that the
additional mora can be realized. Some of these ways are shown by the potential output
candidates in (7) for the word in (1a). (The failed candidates are indicated by the asterisk.)
(7)
Possible candidates for the augmentation of [hade] ‘showy’ (1a)
a. *hhade
b. *haade
c. *hadde
d. hande
e. *hadee
The question that arises is why the form in (7d) with nasal insertion is preferred to the other
candidates with gemination or vowel lengthening First, candidate (7a) is phonotactically
impossible given that Japanese does not allow for initial geminate consonants. Moreover, (7c)
with the medial gemination of /d/ is ruled out because Yamato (native Japanese) words do not
allow voiced geminates (Ito & Mester 1999). The preference for the candidate in (7d) with an
inserted nasal over one like (7b) or (7e) with a long vowel reflects a ranking relationship
between the constraints given in (8) and (9).4
(8)
(9)
*LongVowel -- Long vowels are not permitted.
Dep-Nasal -- Nasal segments should not be inserted into the output.
The preference for (7d) must mean that the constraint militating against inserted nasals (DepNasal) is lower-ranked than the constraint militating against long vowels (*LongVowel). This,
thus, establishes the ranking shown in (10) where *LongVowel outranks Dep-Nasal.
(10)
*LongVowel >> Dep-Nasal
An analysis of the data in (2) requires the active constraint militating against geminate
consonants given in (11). This constraint has been independently posited by Rose (2000).
(11)
*Gem -- Geminate consonants are not permitted.
To show how this constraint is active, let us consider the three potential output candidates in (12)
for the emphatic form of the adjective in (2a).
(12)
Possible candidates for the augmentation of [katai] ‘hard’ (2a)
a. *kantai
b. *kaatai
c. kattai
The candidate in (12a) with an inserted nasal is interesting since it is the one we might expect to
be the actual output given the ranking in (10). However, (12a) violates the constraint in (13),
discussed by Ito & Mester (1999), that prohibits voiceless consonants after nasals. This is an
active constraint in the Yamato vocabulary.
(13)
No-NT --- Postnasal obstruents must be voiced.
The choice then is between (12b) with a long vowel or (12c) with a geminate consonant. The
preference for (12c) establishes the ranking of *LongVowel over *Gem shown in (14). If the
ranking were reverse, (12b) would wrongly be preferred.
(14)
*LongVowel >> *Gem
In (15) we put together the rankings from (14) and (10).
(15)
*LongVowel >> Dep-Nasal, *Gem
The data provide no evidence for a critical ranking between Dep-Nasal and *Gem.
Let us now consider the data in (3) focusing on (3c). Some relevant potential output
candidates for this data item are shown in (16).
(16)
Possible candidates for the augmentation of [onzokutai] ‘ugly’ (3c)
a. oonzokutai
b. *onnzokutai
c. *onzokkutai
The comparison of the three forms in (16) is interesting. The actual form that surfaces in (16a)
has the additional mora realized through the lengthening of the first vowel. The alternative in
(16b) witnesses nasal insertion after the first vowel; the alternative form in (16c) realizes the
additional mora by the gemination of a voiceless consonant, but here, the gemination is later in
the word, after the second vowel. First, in considering why (16b) fails to surface, we note that if
mora augmentation is realized through nasal insertion, the output would have impermissible
syllabification. The first two syllables of (16b) would either include a complex coda ([onn.zo])
or a complex onset ([on.nzo]), thus constituting a fatal violation of the *Complex constraint
given in (17) which is normally undominated in Japanese.
(17)
*Complex -- Neither complex onsets or complex codas are allowed.5
On the other hand, the output in (16c) with gemination of the /k/ after the second vowel is
phonotactically fine. It should be the actual output given the ranking established in (14) where
*LongVowel outranks *Gem. Consequently, for (16a) to be the winning candidate there must be
some other constraint that outranks *LongVowel which (16c) fares worse on. The constraint that
(16c) fares worse on in comparison with (16a) is the Align-L(µe,Wd) constraint given in (5).
That is, the candidate in (16a) with a long vowel in the initial syllable better respects Align-L
(µe,Wd) as compared to (16c).6 This establishes the ranking in (18) where Align-L(µe,Wd)
outranks *LongVowel. The fuller ranking in (19) combines (18) with (15).
(18)
(19)
Align-L(µe,Wd) >> *LongVowel
Align-L(µe,Wd) >> *LongVowel >> *Dep-Nasal, *Gem
For convenience to the reader, in (20a-f), we compile the constraints presented so far.
(20)
Constraints
a. Align-L(µe,Wd) -- Align the emphatic mora with the left edge of the word.
(Align-L)
b. *LongVowel -- Long vowels are not permitted. (*LV)
c. *Gem -- Geminate consonants are not permitted. (*Gem)
d. Dep-Nasal -- Nasal segments should not be inserted into the output. (Dep-N)
e. *Complex -- Neither complex onsets nor complex codas are allowed. (*Comp)
f. No-NT -- Postnasal obstruents must be voiced. (No-NT)
g. No Voiced Geminates -- Voiced geminate consonants are prohibited. (No-VG)
h. *Gem-Initial -- Word-initial geminate consonants are not permitted. (*Gem-In)
We add the constraint in (20g) militating against voiced geminate consonants which has been
discussed by Ito & Mester (1999). We also add a constraint militating against word-initial
geminate consonants in (20h). This constraint can be motivated by the rarity of word-initial
geminate consonants in the world’s languages. (See Davis 1999 for discussion.) The constraints
*Complex, No-NT, No Voiced Geminates, *Gem-Initial, and Align-L(µe,Wd) are all considered
undominated for purposes of our analysis.7 (After each constraint in (20) we provide in
parentheses the abbreviation to be used in the tableaux. In the tableaux, a period indicates a
syllable boundary.)
In (21), we present the tableau for the emphatic form of [hade] from (1a) to make clear
some of the constraint rankings discussed above. In order to make clear how some of the
constraint violations are being determined in (21) it is useful to go over each of the candidates.
First, (21a) is eliminated because it violates the undominated constraint against initial geminates.
With respect to *Complex, we have indicated a violation in parentheses. This signifies an
uncertainty as to whether an initial geminate would comprise a complex moraic onset or have an
extrasyllabic mora. Since this issue is not crucial in determining the status of (21a) we do not
take a position on it.
(21)
/µ + hade/ --- [hande] ‘showy (emphatic)’ (1a)8
No-NT
µe µ µ
No-VG
*Comp Align-L *LV Dep-N *Gem
| |
*Gem-In
hade
a. hhade
*! (Gem-In)
(*)
*
µeµ µ
\/ |
*!
b. h a d e
[haa.de]
µµe µ
\/ |
*!
c. h a d e
[haa.de]
µ µeµ
| | |
*! (No-VG)
*
d. h a d e
[had.de]
µ µe µ
| | |
*
) e. h a n d e
[han.de]
µ µeµ
| \/
*!
*
f. h a d e
[ha.dee]
Candidates (21b) and (21c) are phonetically identical. In candidate (21b) the emphatic mora is
realized on the first part of the long vowel which would be the first mora of the word while in
candidate (21c) the emphatic mora is realized as the second mora of the long vowel. Both (21b)
and (21c) can be compared with (21e) whereby (21b) and (21c) violate *LV while (21e) violates
Dep-Nasal. These two constraints are in direct conflict. The fact that (21e) is the winner
constitutes a ranking argument for *LV being higher ranked than Dep-Nasal, as discussed
earlier. Similarly, candidate (21f) violates *LV in addition to violating the Align-L constraint
and so would be eliminated. Finally, (21d) is eliminated because of its violation of the
undominated constraint militating against voiced geminates. The only ranking argument
involving dominated constraints that emerges from (21) is that *LV outranks Dep-Nasal as
discussed earlier.
In (22), we provide the tableau for the emphatic form in (2a) with gemination. In
examining the candidates in tableau (22) we first note that candidate (22a) is eliminated because
it violates the undominated constraint against initial geminates. Since such forms never surface
in Japanese, we will no longer consider such candidates in subsequent tableaux.
(22)
/µ + katai/ --- [kattai] ‘hard (emphatic)’ (2a)
µe µ µ µ
No-NT
| | |
*Comp Align-L *LV Dep-N *Gem
k a t a i No-VG *Gem-In
a. kkatai
*! (Gem-In)
(*)
*
µeµ µ µ
\/ | |
*!
b. k a t a i
[kaa.tai]
µµe µ µ
\/ | |
*!
c. k a t a i
[kaa.tai]
µµeµ µ
| | | |
*
d.
k
a
tai
)
[kat.tai]
µµe µ µ
| | | |
*! (No-NT)
*
e. k a n t a i
[kan.tai]
µ µeµ µ
| \/ |
*!
*
f. k a t a i
[ka.taai]
Candidates (22b) and (22c) both violate the constraint militating against long vowels, *LV.
These candidates can be compared with (22d) which violates *Gem. These two constraints are in
direct conflict. The fact that (22d) is the winner constitutes a ranking argument for *LV being
higher ranked than *Gem, as discussed earlier. Similarly, candidate (22f) violates *LV in
addition to violating the Align-L constraint and so would be eliminated. Finally, (22e) is
eliminated because of its violation of the undominated constraint No-NT which militates against
a voiceless obstruent after a nasal. The only ranking argument involving dominated constraints
that emerges from (22) is that *LV outranks *Gem as mentioned.
A comparison of the tableaux in (21) and (22) brings up two additional related points.
First, there is no clear evidence for the ranking between Dep-Nasal and *Gem. This is because
there are no Yamato forms where there would be a constraint conflict between the two. Consider
the comparison of (21d) with (21e). The former violates *Gem while the latter violates DepNasal. This is a potential constraint conflict. However, (21d) also violates the undominated
constraint militating against geminate voiced consonants, so we cannot see how the potential
conflict would be resolved. Similar is the comparison between (22d) and (22e). Here again there
is a potential constraint conflict between the same two constraints, but this time candidate (22e)
violates undominated No-NT, so again the conflict remains unresolved. The second related point
is that the constraint *Gem almost certainly cannot outrank Dep-Nasal, though it can be tied with
it. The evidence for this comes from a possible candidate [kandai] which is not considered in
tableau (22). This candidate is interesting in that it avoids a violation of No-NT by voicing the
obstruent. This would mean that it has a violation of the constraint Ident-voice (i.e. input and
output segments agree in the feature voice) in addition to Dep-Nasal. However, Ident-voice is
most likely a low-ranked constraint, especially given the discussion in Ito & Mester (1999)
where its effect can be seen in a form like [sinda] ‘died’ from underlying /sin + ta/. The
consequence of this is that if *Gem were higher ranked than Dep-Nasal then [kandai] would
wrongly emerge as the winner in tableau (22). Consequently *Gem must be equally ranked or
lower ranked than Dep-Nasal. We will assume equal ranking of these two constraints.
In (24) we provide the tableau for the emphatic form of (3c) with vowel lengthening. In
this tableau we include candidate (24e) in which the augmented mora is not realized. This
violates a constraint on morphological realization that requires morphemes to be realized. (See
Prince & Smolensky 1993, Abu-Mansour (1995), Kurisu (2001), Kubozono (to appear) among
others.)
(23)
MorphemeRealization-µe -- The emphatic morpheme (µe) must be realized (MorphR-µe)
This constraint is undominated, at least with respect to the mora (µe) which marks the emphatic
form of the adjective.9
(24)
/µe + onzokutai/ --- [oonzokutai]10 ‘ugly (emphatic)’ (3c)
µe µ µ µ µ µ µ No-NT
*Comp
| | | | | | No-VG
Align-L *LV Dep-N *Gem
MorphR-µe
o n z o k u t a i *Gem-In
µµeµ µ µ µ µ
\/ | | | | |
*
a.
o nzoku t a i
)
[oon.zo.kutai]
µ µµeµ µ µ µ
| | | | | | |
*!
*
*
b. o n z o k u t a i (No-VG)
(*Comp)
[onz.zo.kutai]
µ µ µµe µ µ µ
| | \/ | | |
*!
*
c. o n z o k u t a i
[on.zoo.kutai]
µ µ µµeµ µ µ
| | | | | | |
*!
*
d. o n z o k u t a i
[on.zok.kutai]
µe µi µ µ µ µ µ
| | | | | |
*!
e. o n z o k u t a i
(MorphR-µe)
[on.zo.ku.tai]
The interesting comparison in (24) is between (24d) with gemination of the voiceless consonant
after the second vowel and (24a) with lengthening of the initial vowel. This comparison makes
clear the role of Align-L. Given the tableau in (21) and (22) we know that both gemination and
nasal insertion are the preferred means of mora augmentation. However, these are not preferred
if they would occur later in the word (i.e. beyond the second vowel). As shown by the
comparison of (24a) with (24d), candidate (24a) wins out over (24d) because it better satisfies
the Align-L constraint. Notice, too, the role of Align-L in eliminating candidate (24c) where the
second vowel is lengthened instead of the first. Candidate (24e) shows the role of the constraint
Morph-Rµe. In (24e) the emphatic mora is not realized so that candidate is eliminated. Thus,
given the input in (24), vowel lengthening becomes the preferred means of mora augmentation.
Given our analysis, we now consider two types of cases that have not been formally
discussed. First we consider the case where the first syllable of the (base) adjective has a
diphthong and then we consider the case where it has a long vowel. For each case we need to
refer to constraints not previously introduced in order to account for the emphatic forms. The
first case involving a diphthong in the initial syllable is illustrated by (25a). Examples involving
a long vowel are shown in (25b-c). (The period indicates a syllable break.)
(25)
Adjective
a. oi.šii
b. kii.roi
c. too.toi
Emphatic Form
oo.i.šii
kiin.roi
toot.toi
Gloss
delicious
yellow
respectable
First, consider the first type of case in (25a) involving an adjective whose initial vowel is a
diphthong. In (26), we provide some possible output candidates for the emphatic form of [oi.šii].
(26)
Candidates for the augmentation of [oi.šii] (Subscript ‘e’ indicates the location of the
augmented mora.)
a. ooe.i.šii
b. *oiše.šii
c. *o.iše.šii
d. *oi.šii
Candidate (26a), [oo.i.šii], is the winning candidate, but our analysis as developed so far wrongly
predicts (26b), [oiš.šii], as the winner. The winning candidate has mora augmentation through
vowel lengthening while the losing candidate in (26b) has augmentation by means of
gemination. We have emphasized the priority of consonantal augmentation over vocalic
augmentation. Thus, there must be some higher ranked constraint that [oo.i.šii] respects which
[oiš.šii] violates. The relevant constraint is the one in (27) militating against trimoraic syllables
(cf. Sherer 1994).
(27)
*σµµµ -- Trimoraic syllables are not permitted.
For [oo.i.šii] to be the winning candidate, this constraint has to be higher-ranked than *LV. The
two other candidates in (26c), [o.iš.šii], and (26d), [oi.šii], violate high ranked constraints
previously introduced: (26c) violates Align-L(µe,Wd) since the augmented mora is in the second
syllable and (26d) violates the constraint on morpheme realization, Morph-Rµe. The full
evaluation tableau for the candidates in (26) is given in (28), incorporating the constraint against
trimoraic syllables.11
(28)
/µe + oi.šii/ -- [oo.i.šii] ‘delicious (emphatic)’ (25a)
No-NT
µe µ µ µµ
*Comp
No-VG
| | \/
Align-L
MorphR-µe
*Gem-In
o i š i
µµeµ µµ
\/ | \/
) a. o i š i
[oo.i.šii]
µ µµeµµ
| | | \/
b. o i š i
[oiš.šii]
µ µµeµµ
| | | \/
*!
c. o i š i
[o.iš.šii]
µe µ µ µµ
*!
| | \/
(MorphR-µe)
d. o i š i
[oi.šii]
*σµµµ
*LV Dep-N *Gem
**
*!
*
*
*
*
*
Now let us consider the evaluation of forms like (25b-c) where the first syllable contains
a long vowel. We will illustrate this pattern with consideration for the emphatic form of [kii.roi].
Some possible candidates are provided in (29).
(29)
Candidates for the augmentation of [kii.roi]12
a. *kii.i.roi
b. kiin.roi
c. *kiir.roi
d. *kiii.roi
Candidate (29b) is the winning candidate, though our analysis based on (28) would wrongly
predict (29a). This is because (29b) violates the constraint against trimoraic syllables while (29a)
respects it. However, (29a) is odd in that an extra long vowel sequence is divided between two
syllables, thus violating a constraint on long vowel integrity given in (30), which requires long
vowels to be tautosyllabic.13
(30)
Long Vowel Integrity -- Long vowels must be tautosyllabic. (Int-LV)
This constraint must be higher ranked than the constraint against trimoraic syllables. As a
consequence of this constraint, the emphatic forms of adjectives with initial long vowels as in
(25b) and (25c) will have a trimoraic initial syllable. Given that (29a) violates (30), the choices
are between the other candidates in (29) all of which have a trimoraic first syllable. Consider the
tableau in (31) showing all the candidates from (29).
(31)
/µe + kii.roi/ -- [kiinroi] ‘yellow (emphatic)’ (25b)
µe µµ µ µ
No-NT
Align-L
*Comp
No-VG
\/ | |
Int-LV
MorphR-µe
*Gem-In
k i roi
µµeµ µ µ
*!
\|/ | |
(Int-LV)
a. k i r o i
[kii.i.roi]
µµµe µ µ
\/ | | |
) b. k i n r o i
[kiin.roi]
µµµeµµ
*!
\/ | | |
(No-VG)
c. k i r o i
[kiir.roi]
µµeµ µ µ
\|/ | |
d. k i r o i
[kiii.roi]
*σµµµ
*LV
Dep-N
*Gem
*
*
*
*
*
*
**!
*
*
Candidate (29c) is clearly eliminated because of its violation of the high ranked constraint
against voiced geminates. Candidate (29d) with an extra long vowel is ruled out because of its
extra violation of *LV. The second violation of *LV is because of the vowel’s extra length.
Thus, (29b), [kiin.roi], emerges as the winner. We note that a virtually identical analysis would
apply to the mora augmentation of [too.toi] in (25c) with the exception that the winning
candidate, [toot.toi] would violate *Gem rather than Dep-Nasal.
In summary, we have presented a detailed optimality-theoretic analysis of the formation
of emphatic adjectives through mora augmentation in the Shizuoka dialect of Japanese. A
priority of the means of mora augmentation emerges from the analysis. The preferred way of
mora augmentation is by consonantal augmentation, either nasal insertion or gemination; if that
is not possible, then vowel lengthening occurs. Mora augmentation always occurs toward the left
edge of the word so that augmentation by lengthening the first vowel is preferred to geminating a
consonant after the second vowel as was shown by the discussion of (16). With minor
modification, our analysis applies to the words in (25) with long vowels or diphthongs in the first
syllable. An interesting point that emerges from our analysis of these forms is that a diphthong
such as /oi/ in [oi.šii] may end up heterosyllabic in order to avoid a trimoraic syllable (thus the
preference for [oo.i.šii] over [oiš.šii]), but a long vowel is treated as tautosyllabic even if the
resulting syllable is trimoraic (thus the preference for [kiin.roi] over [kii.i.roi]). In the next
section we will briefly discuss mora augmentation associated with intensified adverbs in
standard Japanese. While the data are phonotactically limited in that the relevant adverbs have a
strict (C)VCV shape, they do confirm the analysis of the prioritization of the means of mora
augmentation.
4 Comparison with Intensified Adverbials
In the previous section we offered a detailed optimality theoretic analysis of the mora
augmentation that occurs when making an adjective emphatic in the Shizuoka dialect of
Japanese. The analysis reflects a prioritization of the means of mora augmentation such that
consonantal augmentation is preferred to vocalic augmentation. It is interesting to note that this
same prioritization of mora augmentation is found with intensified adverbs in standard Japanese.
The formation of intensified adverbs have been discussed in such works as Kuroda (1965),
McCawley (1968), Hamano (1998), as well as Lombardi (1998) from an optimality-theoretic
perspective. Intensified adverbs are part of the Japanese mimetic vocabulary and involve the
suffixation of -ri to a base along with mora augmentation of the first syllable. Phonotactically,
the intensified adverbials are somewhat limited in that the base of affixation is always of the
shape (C)VCV. This is different (and less interesting) than the Shizuoka adjectival data where
the initial syllable of the adjectival base can be heavy as in (3) and (25). In this section we
briefly discuss the similarities and differences between the two mora augmentation processes and
then offer an analysis of the intensified adverbs. The analysis is essentially the same as the one
developed for the Shizuoka emphatic adjectives and it will be shown that the two analyses can be
combined into a single unified analysis of mora augmentation.
Let us consider the representative data in (32) showing the Japanese intensified adverbs.
(See Hamano 1998 for a full listing of the intensified adverbs plus discussion on their
semantics.)
(32)
Intensified adverbs
Base
Intensifed Adverb
a. kote
kotte-ri
b. bata
batta-ri
c uka
ukka-ri
d. huku
hukku-ri
e. koso
kosso-ri
f. kiči
kičči-ri
g. zabu
zambu-ri
h. šobo
šombo-ri
i. mai
mani-ri
j. koga
koga-ri
k. boya
bonya-ri
l. yawa
yanwa-ri
m. gena
genna-ri
n. simi
simmi-ri
o. fura
fura-ri
English gloss
densely
with a bang
thoughtlessly
plump, puffy
steathily
tightly
with a splash
sadly
a wink of sleep
brown
absently
gently
fed up
calmly
swaying
The augmentation pattern we see in (32) is quite similar to what we saw for the Shizuoka
emphatic adjectivals. If the consonant after the first vowel is voiceless, gemination occurs as in
(32a-f); but if it is a voiced consonant then augmentation is by nasal insertion as in (32g-n).
However, there are some significant differences with the adjectival emphatics. First, nasal
insertion does not occur before /r/ as in (32o). Moreover, vowel lengthening usually does not
occur as a means of mora augmentation.14 An analysis of the intensified adverbials needs to
account for these differences.
Before showing how the constraint ranking developed for the emphatic adjectives applies
to the intensified adverbials, we consider the nature of the input representation of the intensified
adverb. We view the intensified adverb as consisting of the suffix -ri and a floating mora which
we represent as µin along with the (C)VCV base.15 This is shown in (33) for [kotteri], (32a).
(33)
µin µ
µ
µ
|
| +
|
k o t e
r i
Given the nature of the input, it should be clear that the ranking shown in (19) also accounts for
the prioritization of the means of mora augmentation for intensified adverbials. We show this in
(34) where we add the relevant high ranking constraints discussed earlier.
(34)
*Comp, No-VG, No-NT, Align-L, >>*LongVowel >> *Dep-Nasal, *Gem
Both the Shizuoka emphatic adjectives and the intensified adverbials show a preference for
consonantal augmentation. The ranking in (34) will account for the intensified adverbial cases
with gemination (32a-f) and nasal insertion (32g-n). This is shown in the tableaux in (35) for
[kotteri] illustrating gemination and in (36) for [bonyari] illustrating nasal insertion.16
(35)
/µin + kote-ri/ -- [kotte-ri] ‘densely’ (32a)
µin µ µ µ
| | |
*Comp No-VG No-NT
ko t e - r i
µµin µ µ
\/ | |
a. k o t e -r i
[koote-ri]
µµin µ µ
| | | |
*!
b. k o n t e -r i
[konte-ri]
µµinµ µ
| | | |
) c. k o t e -r i
[kotte-ri]
µ µµin µ
| \/ |
d. k o t e -r i
[kotee-ri]
Align-L
*LV
Dep-N
*Gem
*!
*
*
*!
*
(36)
/µin + boya-ri/ -- [bonya-ri] ‘absently’ (32k)
µin µ µ µ
| | |
*Comp No-VG No-NT
b o y a -r i
µµin µ µ
\/ | |
a. b o y a -r i
[booya-ri]
µ µin µ µ
| | | |
) b. b o n y a -r i
[bonya-ri]
µµinµ µ
*!
| | | |
c. b o y a -r i
[boyya-ri]
µ µµin µ
| \/ |
d. b o y a -r i
[boyaa-ri]
Align-L
*LV Dep-N *Gem
*!
*
*
*!
*
In these tableaux, we do not show the Morph-R constraint presented in the Shizuoka analysis
since this constraint referred to the specific emphatic morpheme (µe). (Also, for these two
tableaux, we do not consider a candidate in which the floating mora (µin) remains unrealized
since this will be discussed later with respect to (32o).) The tableaux in (35) and (36) show the
similarity between the augmentation with the intensified adverbial and the Shizuoka emphatic
adjectives.
Given this, we must now account for the differences, the failure of augmentation to occur
in intensified adverbials for data like (32o) where /r/ is the intervocalic consonant of the base and
the lack of vowel lengthening. In the previous section, we showed how the emphatic floating
mora (µe) must always be realized with the emphatic adjective, no matter how phonotactically
odd. On the other hand, as indicated by (32o), the intensified floating mora (µin) does not have to
be realized. This suggests that there is a difference in the morpheme realization constraint
regarding the surfacing of µe as opposed to µin. Thus, we posit that morpheme realization can
have different rankings depending on the morpheme involved. Recall that the morpheme
realization constraint regarding the morpheme for the emphatic adjective (µe) given in (23) is
undominated. However, the morpheme realization constraint for the intensified adverbial mora
(µin) given in (37) is not undominated.
(37)
MorphR-µin -- The floating mora marking the intensified adverb must be realized.
If we rank (37) between the constraints *LV and Dep-N as in (38) we can account for why mora
augmentation in intensified adverbials is not realized through vowel lengthening.
(38)
*LV >> MorphR-µin >> Dep-N, *Gem
This ranking means that it is better not to realize the floating mora for the intensified adverbial
than to realize it through vowel lengthening.
However, the ranking in (38), on its own, does not tell us why nasal insertion fails to
occur before /r/ as in (32o). To account for this, we posit the constraint in (39) that militates
against the surfacing of the sequence [nr].
(39)
*nr -- The sequence [nr] is not permitted.17
If we rank this constraint above MorphR-µin as in (40), we can account for the lack of nasal
insertion before /r/ in intensive adverbials like in (32o). Moreover, if we also rank it below *LV
we would be able to account for the occurrence of nasal insertion before /r/ in Shizuoka emphatic
adjectivals as in [kanrai] ‘spicy’ in (1e).
(40)
*LV >> *nr >> MorphR-µin >> Dep-N, *Gem
Given this ranking, consider the tableau in (41) for the intensified adjectival [fura-ri] from (32o).
(We list the undominated constraints in one column. These include *Comp, No-VG, No-NT, and
Align-L.)
(41)
/µin + fura-ri/ --- [furari] ‘swaying’ (32o)
µin µ µ µ
*Comp, No-VG,
| | |
No-NT,
*LV *nr MorphR-µin Dep-N *Gem
f u r a -r i
Align-L
µin µ µ µ
| | |
*
) a. f u r a -r i
[furari]
µµin µ µ
\/ | |
*!
b. f u r a -r i
[fuura-ri]
µ µin µ µ
| | | |
*!
*
c. f u n r a -r i
[funra-ri]
µµinµ µ
| | | |
*!
*
d. f u r a -r i
(No-VG)
[furra-ri]
µ µµin µ
| \/ |
*!
*
e. f u r a -r i
(Align-L)
[furaa-ri]
In (41d) we see that the /r/ cannot geminate because of undominated No-VG. We would expect
nasal insertion as in (1e), [kanrai]. However, given that the constraint MorphR-µin is ranked
lower than *nr, nasal insertion will not occur. Thus (41c) fails to surface. Also the candidate with
vowel lengthening (41b) does not surface given that MorphR-µin is ranked lower *LV as well.
(41e) is ruled out because it violates both *LV and Align-L. Consequently, the optimal form is
(41a) where the intensive morpheme goes unrealized.
We have been assuming that Align-L is high ranked for the analysis of the intensified
adverbials just as it is for the Shizuoka emphatic adjectives, but there is no clear evidence for
this. That is, unlike the case of the Shizuoka emphatic adjectives there is no way of telling
whether mora augmentation occurs closer to the left edge of the word or the right edge of the
word. This is because the base of the intensified adverbials are restricted to a (C)VCV sequence.
Given that vowel lengthening is not typically permitted with intensified adverbials, there is only
one possible place for the augmented mora to be realized without violating high ranking
constraints. However, there is evidence for the role of Align-L. An anonymous reviewer points
out to us that some speakers do get [fuura-ri] for (32o) with a long vowel. For these speakers, the
constraint militating against long vowels (*LV) would be ranked lower than MorphR-µin (but
higher than Dep-Nasal and *Gem), thus requiring the floating mora of the intensified adverbial
to be realized (toward the left edge of the word).
As a final point, the analysis of the intensified adverbials is readily compatible with that
of the emphatic adjectives. For example, the constraint ranking as reflected in tableau (41) above
can also account for the Shizuoka adjectival emphatics. In other words, we can offer a single
unified analysis that accounts for both patterns of mora augmentation. This is shown by the
tableau in (42) with the example [kanrai] ‘spicy’ in (1e) where the augmented mora is a nasal
inserted before /r/. (To save space we do not show all the high ranked constraints that we have
previously discussed, but we include high ranked MorphR-µe which requires the emphatic
morpheme to be realized.)
(42)
/µe + karai/ --- [kanrai] ‘spicy (emphatic)’ (1e)
µe µ µ µ
MorphR-µe Align| | |
*LV *nr MorphR- µin Dep-N *Gem
No-VG
L
k arai
µe µ µ µ
*!
| | |
(MorphRa. k a r a i
µe)
[karai]
µµe µ µ
\/ | |
*!
b. k a r a i
[kaarai]
µ µe µ µ
| | | |
*
*
) c. k a n r a i
[kanrai]
µµeµ µ
*!
| | | |
*
(No-VG)
d. k a r a i
[karrai]
µ µµeµ
| \/ |
*!
*
e. k a r a i
[karaai]
What the tableau above shows is that the constraints that play a special role for the
intensified adverbials (*nr and MorphR-µin) do not affect the analysis of the adjectival
augmentation. Here the winning candidate is (42c) despite its violation of *nr. The alternatives
shown in the tableau each violates a higher ranking constraint. In (42), the key ranking is
between *LV and *nr. If *nr were ranked higher than *LV then the form in (42b) with a long
vowel would be the winner. Thus the tableau provides a ranking argument for *LV outranking
*nr.18
5 Conclusion
In this paper we have offered a detailed analysis of the mora augmentation associated with
emphatic adjectives in the Shizuoka dialect of Japanese. We have shown that there is a priority
of means of augmentation, with consonantal augmentation preferred over vocalic augmentation
and that this can be captured in an optimality-theoretic framework. We further demonstrated how
essentially the same constraint ranking can be extended to account for the mora augmentation
associated with intensified mimetic adverbials in Japanese which display the same priority of
means of augmentation.
In addition to the emphatic adjectives and intensified adverbials there are other mora
augmentation processes in Japanese that we do not discuss and that are beyond the scope of this
paper. These processes are similar to the adjectival and adverbial augmentation phenomena
focused on in this paper, but differ in details. Two such processes included the mora
augmentation that is associated with ma- prefixation (e.g. makkuro ‘very black’ from kuro
‘black’, makka ‘very red’ from akai ‘red’, and massao ‘very blue’ from aoi ‘blue’) and emphatic
mimetic reduplication (e.g. pikkapika ‘shiny’, subessube ‘smooth’, and dabbudabu ‘loose’)
which has recently been discussed by Kubozono (to appear) and Nasu (1999). As can be seen
from the examples, both these processes differ in interesting ways from the mora augmentation
phenomena discussed in this paper. For example, the mora augmentation associated with maprefixation never seems to involve vowel lengthening or clear cases of nasal insertion. Thus, we
find maorenzi ‘very orange’ and mayonaka ‘midnight’ without augmentation. However, nasal
insertion can occur before a nasal as in mannaka ‘the very center’ from naka ‘center’, The
emphatic mimetic reduplication is interesting because augmentation need not take place in the
initial syllable (e.g. subessube ‘smooth’) and voiced obstruents are allowed to geminate (e.g.
dabbudabu ‘loose’). Nonetheless, these augmentation processes still display the preference for
consonantal augmentation, and, in particular, gemination as the priority means of mora
augmentation. We leave the description and analyses of these and other mora augmentation
phenomena for future research.
Acknowledgments
We wish to thank Karen Baertsch, Masa Deguchi, Daniel Dinnsen, Matthew Gordon, Jon
Hathorn, Hideyuki Hirano, Haruo Kubozono, Mayumi Miyake, Akio Nasu, Toshiyuki Tabata,
and Natsuko Tsujimura for helpful comments related to this paper. We also acknowledge our
Shizuoka dialect consultants Kaneko Aoshima, Sachiko Aoshima, and Tamotsu Koizumi.
Finally, we wish to thank the two anonymous reviewers whose comments helped to substantially
improve this paper. Aspects of this paper were presented by both authors at various invited talks
and meetings in Japan, Korea, Germany, and the United States. We thank the many people who
gave us feedback on our presentations. An earlier version of this paper appears as Davis & Ueda
(2001). The current paper is a substantial reworking and expansion of the earlier paper and
supercedes it.
Notes
1. Following the suggestions of two reviewers we consider emphatic forms such as [kannašii] and
[ammai] in (1g) and (1h) as involving nasal insertion rather than gemination. We leave open the question
as to whether the two adjacent identical consonants in (1g) and (1h) surface as a true geminate (i.e. with
one common root node) or as a fake geminate (i.e. with two separate root nodes). See Rose (2000) for
relevant discussion regarding the representation of identical consonant sequences.
2. An anonymous reviewer points out that the emphatic forms with gemination in (2) are also possible in
the Tokyo dialect, though (1) and (3) are not.
3. The alignment constraint posited here is very similar in effect to the alignment constraint posited by
Kubozono (to appear) in his account of mora augmentation in reduplicated mimetics.
4. The constraint in (8) against long vowels has been used by such researchers as Sherer (1994) and
Rosenthall (1997). The Dep-Nasal constraint in (9) can be viewed as a specific realization of the general
Dep constraint posited in McCarthy & Prince (1995) which militates against insertion generally.
5. An anonymous reviewer points out that complex onsets may occur in Japanese if one considers
sequences of a consonant followed by [y] as in the first syllable of Kyoto. We will sidestep this point
given the uncertainty of the analyis of the [y] in such forms.
6. The constraint Align-L (µe,Wd) is evaluated with respect to the syllable. If the emphatic mora is
realized in the first syllable of the word (as in 16a), then the constraint is satisfied. Candidate (16c) has
one violation of Align-L (µe,Wd) since the emphatic mora is realized in the second syllable.
7. Some of the constraints that we are considering undominated, such as No-NT and No Voiced
Geminates are not undominated with respect to the non-Yamato vocabulary. See Ito & Mester (1999) for
discussion and analysis of the various types of non-Yamato vocabulary.
8. In the tableaux, we normally show the mora structure of the candidates. Unless otherwise indicated, a
prevocalic consonant that is moraic is to be interpreted as a geminate while a preconsonantal consonant
that is moraic is to be interpreted as a nongeminate coda. We follow Hayes (1989) in viewing a geminate
as being moraic. However, with respect to word-initial geminates as in (21a), we do not indicate the mora
structure of such candidates given uncertainties regarding their representation. See Hume et al. (1997) and
Davis (1999) for discussion on this issue. Also, we will not consider candidates with inserted vowels or
deleted consonants because these would violate high ranking constraints.
9. We assume that there may be cases where MorphR is not undominated with respect to certain input
morphemes. Specifically, there may be phonological or prosodic considerations that prevent the
realization of a morpheme. For example, the English superlative suffix -est does not normally occur with
adjectives that are more than two syllables in length.
10. The winning candidate in (24a) is shown with the inserted mora realized as the second part of the long
vowel. There is also a candidate which we do not consider in which the inserted mora is realized as the
first part of the long vowel. This is shown in (i).
µeµµ µ µ µ µ
\/ | | | | |
i. o n z o k u t a i
However, such insertion would violate a constraint HeadDep-µ (Hermans 2000 based on Alderete 1995)
which militates against an inserted mora being realized in syllable head position. In a nucleus with a long
vowel, the first part is considered the syllable head position. Hereafter, we do not consider candidates
with the inserted mora in syllable head position.
11. There are two issues that need to be briefly discussed. One possible candidate that we do not consider
in tableau (28) is [on.ni.šii] with two violations of Dep-Nasal. This could be the winning candidate if it
were included in tableau (28) since Dep-Nasal is lower ranked than *LV which is what (28a) violates.
However, if we assume a high-ranking self-conjunction constraint (i.e. Dep-Nasal & Dep-Nasal, or
simply Dep-Nasal2) then [on.ni.šii] would be eliminated since it has two instances of nasal insertion. (See
Ito & Mester 1996 for self-conjunction constraints and Fukazawa & Miglio 1996 for restricting constraint
conjunction to single families of constraints, such as faithfulness constraints or markedness constraints.)
The second issue concerns the emphatic form of [a.oi], ‘blue’. Our analysis predicts [aa.oi], but
the actual augmented form is [an.oi] with a coda nasal. We do not really have a clear explanation for this,
other than to note that all realistic candidates for the emphatic form of [a.oi] are phonotactically odd. The
problem with [aa.oi] is that there are four consecutive vocalic moras without any consonants intervening.
Such a vowel sequence is highly unusual in Japanese and probably does not normally occur. The actual
form, [an.oi] has a violation of Dep-Nasal. However, it is somewhat unusual in that there is an
intervocalic consonant that is syllabified in coda position rather than in onset position. An alternative,
[a.noi], with the intervocalic nasal in onset position, would violate the high ranked constraint on
morpheme realization, since the inserted nasal would not be moraic because it is exclusively in the onset
position. Moreover, as discussed above, a candidate with two violations of Dep-Nasal, [an.noi], would
violate the high ranked self conjunction constraint. Thus, the output [an.oi] seems the least bad of all the
alternatives. As an anonymous reviewer points out to us, there are other cases in Japanese where an
intervocalic nasal syllabifies in coda position. These can be found in Sino-Japanese words such as [an.itu]
‘live leisurely’ and [šin.in] ‘psychological cause’.
12. One possible candidate that we will not consider here is [kin.roi] with shortening of the original long
vowel so as to avoid a violation of the constraint against trimoraic syllables. This form, however, violates
the constraint Max-µ (McCarthy 2000) which requires an input mora to be realized in the output. The
effect of this constraint is that it prevents vowel shortening. This constraint is high-ranking in Japanese.
13. While to our knowledge, nobody has actually proposed the long vowel integrity constraint in (30), it
seems a very plausible and natural constraint that would rule out a candidate like (29a). In positing (30) as
a constraint, we expect cases where it can be violated. Kubozono (1999) suggests based on accentual
evidence that long vowels in Japanese loanwords can be split up into two syllables as in ku.re.en ‘crane’.
We are assuming that the long vowel integrity constraint would be respected in the native vocabulary.
14. A reviewer points out to us that some speakers lengthen the first vowel in (32o). This is discussed in
more detail after the discussion of the tableau in (41).
15. Hamano (1998) considers the suffix -ri and the intensifier mora (µin) to be different morphemes.
Hamano analyzes the intensifier morpheme as an affixed C-slot. We think it is superior to view it as a
floating mora. This more readily accounts for the location of affixation and the lack of an inserted
consonant before [r] as in (32o). It also accounts for the fact that the heavy syllables created in the
formation of intensive adverbs do not attract accent. This is significant given Hamano’s observation that
heavy syllables are normally accented in mimetic words. Also, our floating mora analysis does not need
to assume a separate CV-tier.
16. The optimality-theoretic analysis of intensified adverbials posited in Lombardi (1998) is somewhat
similar to the analysis given in the tableaux in (35) and (36). However, Lombardi does not posit a floating
mora in the underlying representation, nor does she specifically analyze cases like (32o) where neither
gemination nor nasal insertion occurs. Also, since she did not deal with other augmentation data, her
analysis is not embedded into a larger context. This should not be taken as a criticism of Lombardi (1998)
since her purpose in analyzing Japanese intensified adverbials was not to look specifically at mora
augmentation, but to consider the nature of feature-related constraints.
17. This constraint against [nr] sequences is not undominated since such sequences do occur in Japanese.
Mester & Ito (1989) and Hamano (1998) suggest an intriguing connection between the lack of nasal
insertion before /r/ and the observation that /r/ fails to palatalize in the mimetic vocabulary of Japanese.
These phenomena argue for /r/ being a placeless consonant, at least with respect to the mimetic
vocabulary. As such, it resists nasal insertion since the inserted nasal in data like (32g-n) assimilates to
the place of articulation of the following consonant. If the mimetic /r/ lacks a specified place of
articulation then the preceding nasal cannot assimilate to it. In other words, the failure of /n/ to occur
before [r] in mimetic words is due to the constraint Coda Condition (Ito & Mester 1999) which requires a
(word-internal) coda nasal to have the same place features as the following onset consonant. While it is
quite possible to incorporate the underspecified representation of mimetic /r/ into our analysis, for reasons
of convenience we will use the constraint *nr.
18. As mentioned, according to an anonymous reviewer, some speakers are able to get the adverbial
mimetic [fuura-ri] with vowel lengthening before /r/ for (32o). This is not predicted by the ranking shown
in tableau (42). Nonetheless, we can account for the output [fuura-ri] by ranking Morph-Rµin above *LV
so that the intensive morpheme must be realized. Assuming a consistent ranking whereby *LV outranks
*nr as in the Shizuoka adjectival pattern illustrated by the tableau in (42), we might expect the possible
candidate [funra-ri] as the adverbial mimetic for (32o) rather than [fuura-ri]. In order to account for the
preference of [fuura-ri] over [funra-ri] while keeping the ranking shown in (42), we we would need to
assume that /r/ in mimetic words is placeless as mentioned in the previous note. Thus, [funra-ri] would be
eliminated because of its violation of the high ranked Coda Condition constraint.
References
Abu Mansour, Mahasen Hasan. 1995. Optimality and conspiracy in the syllable structure of Arabic. In:
Beckman, Jill; Dickey, Laura Walsh; and Urbanczyk, Suzanne (eds), University of Massachusetts
Occasional Papers: Papers in Optimality Theory 18:1-20.
Alderete, John. 1995. Faithfulness to prosodic heads. Rutgers Optimality Archives 94.
Davis, Stuart. 1999. On the representation of initial geminates. Phonology 16:93-104.
Davis, Stuart and Ueda, Isao. 2001. Mora augmentation in Shizuoka Japanese. Japanese/Korean
Linguistics 10.
Fukazawa, Haruka and Miglio, Viola. 1996. Restricting conjunction to constraint families. Paper
presented at the Western Conference on Linguistics, University of California, Santa Cruz, Oct. 25-27.
Hamano, Shoko. 1998. The sound symbolic system of Japanese. Stanford: CSLI.
Hayes, Bruce. 1989. Compensatory lengthening in moraic phonology. Linguistic Inquiry 20:253-306.
Hermans, Ben. 2000. Compensatory lengthening in Slovak. Paper presented at Formal Approaches to
Slavic Linguistics 9, Indiana University, Bloomington, Feb. 18-20.
Hino, Sukezumi. 1977. Shizuoka-ken no hogen [The dialect of Shizuoka]. In: Furusato Hyakuwa 20 [A
Hundred Stories of our Native Place 20], 11-80. Shizuoka Newspaper Company.
Hume, Elizabeth; Muller, Jennifer; and van Engelenhoven, Aone. 1997. Non-moraic geminates in Leti.
Phonology 14:371-402.
Ito, Junko. 1986. Syllabic theory in prosodic phonology. University of Massachusetts Ph.D. dissertation.
(Published by Garland Press, 1988)
Ito, Junko and Mester, Armin. 1996. Structural economy and OCP interactions in local domains. Paper
presented at the Western Conference on Linguistics, University of California, Santa Cruz, Oct. 25-27.
Ito, Junko and Mester, Armin. 1999. The phonological lexicon. In: Tsujimura, Natsuko (ed.), The
handbook of Japanese linguistics, 62-100. Cambridge: Blackwell.
Kubozono, Haruo. 1999. Mora and syllable. In: Tsujimura, Natsuko (ed.), The handbook of Japanese
linguistics, 31-61. Cambridge: Blackwell.
Kubozono, Haruo. to appear. The syllable as a unit of prosodic organization in Japanese. In: Féry,
Caroline and van de Vijver, Ruben (eds.), The syllable in Optimality Theory. Cambridge: Cambridge
University Press.
Kurisu, Kazutaka. 2001. The phonology of morpheme realization. University of California, Santa Cruz
Ph.D. dissertation.
Kuroda, S-Y. 1965. Generative grammatical studies in the Japanese language. MIT Ph.D. dissertation.
(Published by Garland Press, 1979.)
Lombardi, Linda. 1998. Evidence for max feature constraints from Japanese. University of Maryland
Working Papers in Linguistics 7:41-62.
McCarthy, John. 2000. The prosody of phase in Rotuman. Natural Language and Linguistic Theory
18:147-197.
McCarthy, John and Prince, Alan. 1995. Faithfulness and reduplicative identity. In: Beckman, Jill;
Dickey, Laura Walsh; and Urbanczyk, Suzanne, (eds). University of Massachusetts Occasional
Papers: Papers in Optimality Theory 18, 249-384.
McCawley, James. 1968. The phonological component of a grammar of Japanese. The Hague: Mouton.
McCawley, James. 1978. What is a tone language? In: Fromkin, Victoria (ed.) Tone: A Linguistic Survey,
113-131. New York: Academic Press.
Mester, Armin and Ito, Junko. 1989. Feature predictability and underspecification: palatal prosody in
Japanese mimetics. Language 65:258-293.
Nasu, A. 1999. Tyoohukukei onomatope no kyootyookei to sokuon no iti [Emphatic Forms of
Reduplicative Mimetics and the Location of Consonant Geminates]. Proceedings of the 19th Meeting
of the Kansai Linguistic Society, 308.
Prince, Alan and Smolensky, Paul. 1993. Optimality Theory: constraint interaction in generative
grammar. Manuscript, Rutgers University and University of Colorado.
Rose, Sharon. 2000. Rethinking geminates, long distance geminates, and the OCP. Linguistic Inquiry
31:85-122.
Rosenthall, Sam. 1997. The distribution of prevocalic glides. Natural Language and Linguistic Theory
15:139-180.
Sherer, Tim. 1994. Prosodic phonotactics. University of Massachusetts Ph.D. dissertation.
Yamaguchi, Koyo. 1987. Shizuoka-ken no hogen [The dialect of Shizuoka]. Shizuoka Newspaper Co.
Stuart Davis
Department of Linguistics
Memorial Hall 322
1021 East Third Street
Indiana University
Bloomington, IN 47405
USA
Isao Ueda
Department of Area Studies
Osaka University of Foreign Studies
8-1-1 Aomadani-higashi
Minoo-shi
Osaka-fu, 562-8558
Japan
e-mail: [email protected]
[email protected]