国立国語研究所学術情報リポジトリ
A Note on Unstressability
著者(英)
journal or
publication title
number
page range
year
URL
Armin Mester, Junko Ito
NINJAL Project Review
4
27‑44
2011‑03
http://doi.org/10.15084/00000561
大学共同利用機関法人人間文化研究機構
国語研プロジェクトレビュー NINJAL Project Review
No.4 pp.27-44(March 2011)
国立国語研究所
National Institute for Japanese Language and Linguistics
A NOTE ON UNSTRESSABILITY
強勢不可能性に関する一考察
Armin Mester*†&Junko Ito*‡(アーミン・メスター, 伊藤順子)
*University of California, Santa Cruz(カリフォルニア大学サンタクルーズ校)
†
National Institute for Japanese Language and Linguistics
(国立国語研究所)[2010.9–2010.12]
‡
International Christian University(国際基督教大学)
Abstract: A familiar case of unstressability is the fact that crosslinguistically schwa is often
excluded as a stress-bearing element. Here we show that in some languages, such as German,
there is evidence for a different kind of requirement: Schwa must occupy the weak position
of a trochaic foot, attracting stress to the preceding syllable.
《要旨》シュワー母音が多数の言語において強勢不可能な要素であることはよく知られているが,
本稿では,ドイツ語等のシュワーが強勢を担えないのは他の理由から説明されることを指摘する。
シュワーは無強勢であると同時に,韻律構造の中で強弱格フットの弱音節に位置付けされなけれ
ばならないため,その先行音節は強音節に位置し,必ず強勢が付与される。つまり,これらの言
語におけるシュワーは,先行音節に強勢を引きつける特徴があると言える。
1. Introduction*
Many languages exclude schwa as the nucleus of a stressed syllable. The restriction comes in
different flavors. In some languages, such as Dutch or English,1 the ban is absolute, as shown
in (1) .
(1)
cut
( x )
[ˈkʌt]
market
( x
. )
[ˈmaɹkət]
*(x)
[ˈkət]
This is evidence for the existence of a constraint as in (2) , where "x" indicates the head of a
foot, its prominent position.
*
Many thanks to the colleagues and the staff at NINJAL for all their hospitality and support, and to our students
at UC Santa Cruz. Special thanks go to Ryan Bennett and Manami Hirayama for their helpful comments and suggestions.
1
I.e., in varieties of English that maintain a distinction between (unstressed) [ə] and (stressed) [ʌ].
27
Armin Mester & Junko Ito
アーミン・メスター,伊藤順子
A Note on Unstressability
強勢不可能性に関する一考察
(2) NOSTRESSED-ə
*x
ə
While (2) is undominated in the grammar of a language like English, in another type of language it is a dominated constraint. As a result, stressed schwa is permitted, but only under
special conditions. Thus in French, schwa-syllables that constitute clitics can be bearers of
phrase-final stress, but final schwa-syllables in content words cannot, as shown by the seg2
mentally similar pair of forms in (3) .
(3)
prens-ˈlə
*ˈprens-lə
*prenˈdrə
take it
ˈprendrə
take
To take another example, in the Austronesian language Tondano (see Sneddon 1975, Ball
2003, and Piggott 2010) , canonical stress is penultimate or final. It falls on the penult when
the final is light (has a short vowel) (4) .
(4)
ˈwale
ˈwuter
ˈγorəm
waˈnua
kariˈmaŋkaʔ
ˈwεːnu
tiˈkɔːku
kiˈnaːŋku
manuˈaːsa
kapaˈjaːsəm
house
heavy
inside
village
spider
will be given by you
my throat
has been eaten by me
one village
sour papaw
When the final is heavy (has a long vowel) , it attracts the stress (5) .
(5)
ˈkɔː
maˈwεː
məŋaːˈŋaːn
waˈmɔːʔ
tiˈkɔː
arm
is giving
is continually eating
will drink
throat
In other words, a standard analysis builds a quantity-sensitive trochee at the right word edge,
3
as shown in (6) .
(6)
(x .)
... L L
(x .)
... H L
(x)
... L H
(x)
... H H
In an OT analysis, the ranking WSP, FT=TROCHEE >> RIGHTMOST:ALIGN-R (ˈσ,ω) , where
"ω" stands for "prosodic word", produces such a system.
2
It is perhaps not irrelevant in this context that French schwa phonetically possesses a significant degree of lip
rounding, and, as Ryan Bennett points out, that French stress is really phrasal, and arguably not foot-based.
3
Here shown with a classical quantity-sensitive trochee, with HL feet permitted. Moraic-trochaic footing would
be another option.
28
Armin Mester & Junko Ito
アーミン・メスター,伊藤順子
(7)
A Note on Unstressability
強勢不可能性に関する一考察
a. LLL
▶ L (ˈLL)
WSP
Ft=Trochee
Al-R (ˈσ,ω)
*
(ˈLL) L
**!
L (LˈL)
*!
b. LHL
▶ L (ˈHL)
(ˈLH) L
L (HˈL)
*
*!
**
*!
*
c. LLH
▶ LL (ˈH)
(ˈLL) (H)
L (ˈLH)
*!*
*!
*
*!
**
d. LHH
▶ L (H) (ˈH)
(ˈLH) (H)
L (ˈH) (H)
*!
Tondano shows an elaborate system of stress-avoidance for schwa, with the option of stressing schwa when no other vowels are available. In words ending with the sequence schwasyllable followed by a final L, the expected stress should fall on the penultimate schwa. This
is not what happens: When the word has three or more syllables, stress falls on the antepenult, provided its vowel is not schwa (8) .
(8)
wiŋˈkɔtəna
ˈkaːnəna
ikəˈtɔrəna
kəˈtɔrəna
will be asked by him
will be eaten by him
will be used by him to cut with
will be cut by him
This is evidence for the ranking NOSTRESSED-ə above RIGHTMOST.
(9)
wiŋkɔtəna
▶ wiŋ (ˈkɔtə) na
wiŋkɔ (ˈtəna)
wiŋkɔ (təˈna)
NoStressed-ə
Ft=Trochee
Al-R (ˈσ,ω)
**
*!
*
*!
When the vowel of the antepenultimate syllable is also schwa, or there is no antepenult because the word has only two syllables, stress falls on the final syllable, provided its vowel is
not schwa (10) .
(10)
rəməˈdej
is standing
səˈraʔ
fish
mərərəˈdej
əˈsa
intends to stand
one
29
Armin Mester & Junko Ito
アーミン・メスター,伊藤順子
A Note on Unstressability
強勢不可能性に関する一考察
This means that NOSTRESSED-ə outranks FT=TROCHEE, resulting in iambic footing and satisfying RIGHTMOST.
(11)
rəməˈdej
▶ rə (məˈdej)
rə (ˈmədej)
(ˈrəmə) dej
NoStressed-ə
Al-R (ˈσ,ω)
Ft=Trochee
*
*!
*
*!
**
Finally, in all-schwa-words like those in (12) , stress falls on the penult (attested examples of
this kind are apparently all disyllabic) .
(12)
ˈrəpət
ˈwəʔŋəl
ˈpərəŋ
ˈələw
fast
stupid
blind
lid
The OT analysis developed so far (WSP, NOSTRESSED-ə >> FT=TROCHEE>> RIGHTMOST,
with high ranking IDENT-VPLACE and HEADEDNESS (ω) ) correctly predicts penult schwa
stress.
(13) rəpət
Headedness (ω)
▶ (ˈrəpət)
*
(rəˈpət)
(ˈrepət)
rəpət
Ident-VPlace NoStressed-ə Ft=Trochee Al-R (ˈσ,ω)
*
*!
*
*!
*
*!
2. German schwa
German is a language where stressed schwa is categorically ruled out. Interestingly, recent
work has shown that, over and above the unstressability of schwa, a different restriction is at
work: Schwa should be the weak member of a foot-i.e., besides being ruled out as the
prominent member of a foot, it is also ruled out as an unfooted syllable. Observationally
speaking, schwa-syllables need to occur in post-stress positions. As will be seen, this constraint is a dominated one and hence violated under certain conditions, but there are many
situations where it is operative. This kind of pattern is of interest because it provides a window on the hidden underlying rhythmic structure whose overt manifestation is stress, in the
conception of metrical form inaugurated in the work of Liberman and Prince (1977) and others. Beyond the dichotomy of the "stressed" and the "unstressed", there is a dichotomy between two kinds of unstressed syllables: the "footed" and the "unfooted" (McCarthy 2008,
Bennett 2009) .
As background, we begin with a short summary of the stress system of German. With
very few exceptions, word stress is limited to the last three syllables of the word and, as in
Dutch and English, works along the lines of the familiar Latin stress rule, with some variations and modifications. First, superheavy finals (long vowel followed by one or more consonants, short vowel followed by two or more consonants, diphthong with or without following
30
Armin Mester & Junko Ito
アーミン・メスター,伊藤順子
A Note on Unstressability
強勢不可能性に関する一考察
consonants) are stressed in the majority of cases.4
(14)
VːC1
Banˈdiːt
Athˈleːt
VCC1
Aˈlarm
Atˈtest
VVC0
Glosses
Abˈtei
aˈhoi
robber
alarm
abbey
athlete
certificate ahoy
Attenˈtaːt
Efˈfekt
alˈlein
assassination
effect
alone
Disˈpuːt
Konˈzert
Papaˈgei
despot
horizon
applause
dispute
concert
parrot
Paˈpiːr
Proˈdukt
Poliˈzei
comet
Olymp
party
paper
product
police
Samuˈrai
paradise
recipe
row
pirate
talent
samurai
tariff
riot
therapist
Desˈpoːt
Koˈmeːt
Horiˈzont
Oˈlymp
Apˈplaus
Parˈtei
Paraˈdiːs
Reˈzept
Raˈdau
Taˈriːf
Tuˈmult
Theraˈpeut
Piˈraːt
Taˈlent
In final position, superheaviness reduces to heaviness (bimoraic status) if word-final consonants do not carry a mora. With this assumption, the rest follows along familiar lines: When
the penult is heavy, it is stressed; when it is light, stress falls on the antepenult (in words of
three or more syllables) .Writing "σ" for "syllable" (of any weight) , "H" for "heavy syllable", and "L" for "light syllable", we have the following.
(15) ... σσ H]ω
ˈ
Papagei
... σˈH L]ω
Hibiskus
... ˈσ L L]ω
Tremolo
parrot
rose mallow quaver
The pattern here amounts to a quantity-sensitive trochee at the right word edge, with final
consonants not carrying moras.
(16)
x
L L (H)
Papagei
x .
L (H L)
Hibiskus
x . .
(LL) L
Tremolo
As in Dutch, the most controversial issue in German is the behavior of open syllables with
long vowels. Besides the question of whether length or vowel quality (tense vs. lax) is the
operative feature, there has been a long-standing debate whether long vowels make syllables
heavy. For our purposes, we can proceed under either assumption. There are good reasons to
assume, however, that such syllables behave as light for purposes of the stress system, different from both closed syllables and from open syllables with diphthongs. This is no doubt related to the fact that there is a process of vowel lengthening in open syllables. For concreteness, we follow Zonneveld, Trommelen, Jessen, Rice, Bruce and Árnason (1999)(see also the
literature cited there) in assuming that long vowels per se do not make syllables heavy. This
is why we generally get antepenultimate stress when the penult is open and has a long vowel
(underlined) , provided the final is not superheavy (see (14) above) .
4
Examples are given in German orthography, with phonetic details only indicated when essential (e.g., schwa and
vowel length) .
31
Armin Mester & Junko Ito
アーミン・メスター,伊藤順子
(17)
A Note on Unstressability
強勢不可能性に関する一考察
Eˈlisabeth
ˈDomino
ˈTremolo
ˈKolibri
ˈGigolo
ˈPaprika
ˈAlibi
ˈExodus
ˈMarabu ˈAlbatros
ˈAnorak
ˈLexikon
Indiˈviduum Curˈriculum
Aˈnaphora
inˈkognito
(name)
alibi
exodus
domino
(name)
albatross
quaver
anorak
lexicon
hummingbird individual
gigolo
anaphor
paprika
incognito
curriculum
The issue is not entirely straightforward because this pattern in (17) is not exceptionless, but
only a statistical tendency. Penult stress is also found under these circumstances, as the examples in (18) show. The choice is lexically determined, and the antepenultimate pattern is
considerably more frequent.
(18)
Aˈroma
Mosˈkito
Arˈthritis
Kaˈsino
Boˈtanik
Abrakaˈdabra
Biˈkini
Muˈseum
Hiˈatus
aroma
mosquito
arthritis
bikini
museum
hiatus
casino
botany
abracadabra
On the basis of the data presented in Féry (1998), which are extracted from the large lexical
database CELEX developed at the Max-Planck-Institute in Nijmegen, and focusing on trisyllabic monomorphemic words (and disregarding all cases of final stress) , we find the following distribution of antepenult and penult stress in words with full vowels in their last syllable.
(19)
final full vowel
antepenult stress
217
(61%)
penult stress
136
(39%)
Total
353
(100%)
(20)
antepenult stress
61%
penult stress
39%
final full vowel
Against this background, it is significant that in words ending with a schwa syllable the antepenultimate pattern is very rare. Word stress in such examples is overwhelmingly penultimate, as in (21) .
32
Armin Mester & Junko Ito
アーミン・メスター,伊藤順子
(21)
A Note on Unstressability
強勢不可能性に関する一考察
Apotheˈosə Lokomoˈtivə Karaˈwanə
Meˈthodə
Ziˈtronə
Taˈpetə
Westˈfalən
Aˈka[zi̯ə]
Okˈtobər
Antiˈlopə
apotheosis
locomotive caravan
method
lemon
Troˈphäə
wallpaper
Westphalia oboe
acacia
October
Oˈboə
antelope
trophee
The post-stress generalization for schwa-syllables has exceptions, but as Zonneveld, Trommelen, Jessen, Rice, Bruce and Árnason (1999) show, most of these are obscure compounds
or pseudo-compounds as in (22) (there are often independent indicators for this, such as the
onset [ʔ] in Einöde 'wasteland' [ˈʔaIn.ʔøː.də, *ˈʔaI.nøː.də], indicating a syllabification which
can only be induced by compound structure and is impossible in monomorphemes.
(22)
ˈAlmosən
ˈAmeisə
ˈBrosamən
ˈHerbergə
ˈAlkovən
ˈEidechsə
ˈEinödə
ˈentwedər
ˈUrkundə
ˈDarlehən
alms
lizard
ant
wasteland
crumbs
either
hostel
document
alcove
loan
Returning to the survey in Féry (1998) and again focusing on trisyllabic monomorphemic
words with non-final stress, we find the following distribution of antepenult and penult stress
in words with full vowels and schwa in their last syllable.
(23)
final schwa
final full vowel
antepenult stress
38
(7%)
217
(61%)
penult stress
528
(93%)
136
(39%)
Total
566
(100%)
353
(100%)
(24)
93%
antepenult stress
61%
penult stress
39%
7%
final schwa
final full vowel
33
Armin Mester & Junko Ito
アーミン・メスター,伊藤順子
A Note on Unstressability
強勢不可能性に関する一考察
The remaining 7% of antepenult stress in words with final schwa in (23)/(24) include not
only compound-like exceptions as in (22) , but also, as Féry notes, words like Akazie, which
have a pronunciation with three syllables [a.ˈka.z̯iə], with a glided [̯i] and thus actually conform to the penult stress generalization. As Zonneveld, Trommelen, Jessen, Rice, Bruce and
Árnason (1999) point out, this interpretation is independently supported by the observation
that words whose final schwa-syllable is unambiguously onsetless, such as Oˈboə and
Troˈphäə in (21) , are indeed stressed on the penult and not on the antepenult. Onsetless final
schwa therefore does not result in antepenult stress. Overall, the post-stress generalization for
schwa-syllables is very firmly anchored in the prosodic system of German.
Independent support for the generalization that schwa attracts stress to the syllable that
precedes it can be found in cases, as in (25) , whose last vowel has historically been reduced
to schwa. Here, we find stress shifting from antepenult to penult.
(25)
ˈGenesis
ˈViola
Diˈastol[ε]
>
>
>
Geˈnesə
Viˈolə
Diasˈtol[ə]
Koepcke (1995) notes borrowings with a similar stress shift to the pre-schwa syllable ( (26a)
from English, (26b) from Spanish; there is also a variant caniˈbal with stress on the last syllable) .
(26)
a.
b.
ˈmormon
ˈshaman
ˈIroquoise
caˈnibal
>
>
>
>
Morˈmon[ə]
Schaˈman[ə]
Iroˈkes[ə]
Kanniˈbal[ə]
The pre-stressing character of schwa amounts to the generalization that the schwa syllable
must appear in immediate post-stress position, as the weak member of a trochaic foot. This is
a subset of all unstressed positions, which also include unfooted occurrences.
Referring to this position as "foottail", we can state the constraint as in (27) , which
should be understood as being violated by any schwa not occurring in the position indicated.5
(27) FOOTTAIL-ə
Obligatory position for ə: (x
. )
ə
Constraint (27) is violated by two kinds of schwa, stressed schwa and unfooted or loose
schwa. A schwa syllable needs another syllable to depend on. The unstressability of schwa in
a language like German, known as a separate constraint (2) operative in languages like Eng5
The formulation doesnʼt cover iambic feet, and it is at present unclear to us whether it should be generalized or
not. Another possibility to consider is a parsing constraint: PARSE-SCHWASYLL-INTO-FEET, separate from PARSESYLL-INTO-FEET, which would collaborate with NOSTRESSED-ə in restricting schwa to weak positions of feet. As
Ryan Bennett points out, PARSE-SCHWASYLL-INTO-FEET has an intrinsic problem: When dominating NOSTRESSEDSCHWA and ranked sufficiently highly (e.g., PARSE-SCHWASYLL-INTO-FEET >> ALLFEET-RIGHT >> PARSE-SYLL-INTO-FEET)
, it predicts systems where only schwa is stressed in certain positions, to the exclusion of all other vowels.
Another potential problem is the existence of languages like Indonesian where it has been argued that they show evidence for a constraint that schwa-syllables must not be footed at all (see Cohn and McCarthy 1994) .
34
Armin Mester & Junko Ito
アーミン・メスター,伊藤順子
A Note on Unstressability
強勢不可能性に関する一考察
lish, thus follows already from (27) , which is a stricter limitation.
FOOTTAIL-ə also provides an explanation for the systematic stress shift in nouns ending
in -or when a schwa-syllable is added (28) .More examples are given in (29) .
(28)
(
x
. )
. (
ˈDoktor
Proˈfessor
ˈTraktor
(29)
Adˈduktor6
Agˈgressor
ˈAktor
ˈCursor
Deˈskriptor
Deˈtektor
x
.)
(
x
.).
Dokˈtorən
*ˈDoktorən
Trakˈtorən
*ˈTraktorən
Profesˈsorən
ˈDoktor
Diˈrektor
Efˈfektor
ˈFaktor
Inˈduktor
Inˈspektor
*Proˈfessorən
Inˈvestor
ˈLektor
ˈKantor
ˈMentor
Komˈpaktor Proˈfessor
Korˈrektor
Proˈjektor
Korrepeˈtitor ˈProktor
Kolˈlektor
Proˈtektor
Proˈzessor
ˈQuantor
Reˈaktor
ˈReflektor
ˈSektor
ˈSenior
ˈSensor
ˈTensor
ˈTraktor
Tranˈsistor
ˈVektor
ˈZensor
Cf. also the shift by two syllables in (30) (besides the pronunciation variants Meteˈor,
Semaˈphor, with no shift) .
(30) ˈMe.te.or~Me.te.ˈo.rən, ˈSe.ma.phor~Se.ma.ˈpho.rən
The analysis proceeds as in (31) , with an unviolated FTFORM=TROCHEE constraint.
(31)
a. Tremolo
▶ (ˈTremo) lo
WSP
Foottail-ə
NonFin
Al-R (ˈσ,ω)
** (σσ)
Tre (ˈmolo)
*!
*
b. Hibiskus
(ˈHibis) kus *!
**
▶Hi (ˈbiskus)
*
*
c. Papagei
(ˈPapa) gei
Pa (ˈpagei)
▶Papa (ˈgei)
*!
**
*!
*
*
*
d. Zitronə
(ˈZitro) nə
▶Zi (ˈtronə)
*!
**
*
*
Another consequence of FOOTTAIL-ə is that schwa should not be able to follow another
schwa-syllable:
(32)
6
*əC0ə
The English equivalents here are virtually identical to the German forms.
35
Armin Mester & Junko Ito
アーミン・メスター,伊藤順子
A Note on Unstressability
強勢不可能性に関する一考察
In the inflectional system, we find violations of this generalization, and sequences of schwa
syllables can arise, leading to a violation of the FOOTTAIL-ə constraint. There is still evidence
that the constraint is at work here, as well. First, it is easy to find examples of inflected forms
with schwa syllables which are clearly not in post-stress position, as in (33) .
(33)
ˈtroc.k[ə]n
ˈtroc.k[ə].n[ə]r
dry
COMPARATIVE
ˈtroc.k[ə].n[ə]
ˈtroc.k[ə].n[ə].r[ə]
ADJ+INFLECTION
ADJ+COMPARATIVE+INFLECTION
This outcome is due to the fact that other constraints dominating FOOTTAIL-ə, such as IDENT
(VPLACE) , NOSTRESSED-ə, and MAX, preclude any other outcome.
(34)
trockənər
Base: (ˈtrockən)
Ident
(VPlace)
▶ (ˈtrockə) nər
troc (ˈkənər)
troc (ˈkenər)
(ˈtrocknər)
NoStressed-ə Max
Foottail-ə
*
*!
*
*!
Al-R
(ˈσ,ω)
**
*
*
7
*!
*
The stressed schwa in *[troc. (ˈkə.nər)] violates both NOSTRESSED-ə and FOOTTAIL-ə and
shows that both constraints are operative. There is probably also a dominant Output-Output
constraint such as (35) (following Benua 1997:176) at work here, requiring the derived form
to preserve the stress of the base.
(35)
OO-IDENT (FOOTHEAD)
If αxβ is derived from x, segment s∈x corresponds to s' ∈αxβ, and s is part of a foothead, then s' is part of a foothead.
The prosodic parse of a form with three schwas such as [(ˈtroc.kə) .nə.rə] leads to two violations of FOOTTAIL-ə. An alternative hypothesis about the prosodic parse of such forms posits
an additional foot whose head is occupied by the second schwa-syllable, but realized without
phonetic stress: [(ˈtroc.kə) (nə.rə)]. This is undeniably another possibility, and relevant evidence is hard to come by. One piece of evidence arguing against this kind of prosodic interpretation is found in parallel examples of words with four syllables with a non-schwa syllable in the third position. Thus the plural forms of (ˈLehrə) rin 'teacher fem' and (ˈKündi) gung
'dismissal' are (ˈLehrə) (ˌrinnən) and (ˈKündi) (ˌgungən) , respectively, with perceptible secondary stresses on their third syllables. Comparing these forms with ˈtroc.kə.nə.rə, the intuitive feel of their prosodic profile is quite different. Needless to say, this issue cannot be regarded as settled, and further investigation is needed.
On the other hand, some inflectional suffixes have allomorphs with and without schwa,
and here we see the restriction against successive schwas (36a) and more generally against ə
following an unstressed syllable (36b) , again clearly in operation. When this factor is not at
stake (36c) , both variants are possible.
7
This variant, which fulfills the FOOTTAIL-ə constraint by deleting the medial schwa, is actually possible as a fastspeech form, showing that MAX is variably ranked w.r.t. FOOTTAIL-ə.
36
Armin Mester & Junko Ito
アーミン・メスター,伊藤順子
(36)
A Note on Unstressability
強勢不可能性に関する一考察
a.
des ˈRegəns
*des ˈRegənəs
rain
des ˈLehrərs
*des ˈLehrərəs
teacher
des ˈSessəls
b.
*des ˈSessələs
des ˈMonats
des ˈSchicksals
c.
*des ˈMonatəs
month
*des ˈAutoəs
car
*des ˈSchicksaləs
des ˈAutos
des ˈManns
des ˈMannəs
des ˈTags
des Erˈfolgəs
des Erˈtrags
des Erˈtragəs
fate
man
des ˈTagəs
des Erˈfolgs
armchair
day
success
8
yield
What about initial schwa, i.e., schwa in first syllables? Given (27) , it should be impossible,
unless (27) is outranked by another constraint demanding realization of such schwas and
hence to a violation of (27) .It is slightly unclear what the facts are. On the one hand, it is a
fact about German that schwa does not occur in strictly initial position, there are no words
like about [əˈbaUt].
This leaves cases where the schwa is preceded by a word-initial consonant, as in
Beamter 'civil servant', or in the prefixes be- and ge-. It is not unreasonable to assume that
FOOTTAIL-ə is violated in such cases. It turns out, however, that at least for certain speakers
something else happens. Such speakers do not have [ə] there, but instead a reduced version
of [ε] which is clearly distinct from [ə].
das Haar b[ε]nutzen
das hab[ə] Nutzen
to use the hair
that is said to be of use
̯
(37)
A rich area of the lexicon where the FOOTTAIL-ə constraint has a pervasive influence is names
for inhabitants of countries (henceforth "country demonyms") .As Fuhrhop (1998:152 sqq.)
has shown, the schwa-suffixes deriving such forms are strictly required to follow a stressed
syllable. The case is especially interesting because superficially parallel names for inhabitants of cities are not governed by this restriction.
Two main allomorphs are used in German to derive country demonyms, -ə and -ər. Each
occurs either alone, or with an additional interfix (-es- and -an-) , depending on the base
word.
(38)
without interfix
-ə
-ər
with interfix
-es-ə
-an-ər
The interfix occurs mainly (but not exclusively) to attract the word stress to the pre-ə syllable, as schematically shown in (39) .Overall, the number of country demonyms where
stress falls on the pre-ə syllable is overwhelming.
8
Segmental factors can also force the appearance of [ə], as in des ˈKlotzəs (*des ˈKlotzs).
37
Armin Mester & Junko Ito
アーミン・メスター,伊藤順子
(39)
A Note on Unstressability
強勢不可能性に関する一考察
Afˈghanistan
Af (ˈghan-ə)
ˈSenegal
Äˈgypten
Senega (ˈl-es-ə) ,
* (ˈSene) gal-ə,
*Sene (ˈgal-ə)
ˈMexiko
Ä (ˈgypt-ər)
Mexi (ˈk-an-ər) ,
* (ˈMe.xi) ko-ər,
* (ˈMexi) k-ər,
*Me (ˈxik-ər)
Besides a small number of other formations, exemplified in (40) , the overwhelming number
of country demonyms are formed as in (39) .
(40)
Others
ˈIsrael
ˈJemen
Israˈel-i
ˈUngarn
ˈZypern
Jemeˈn-it
ˈUngar
Zypriˈot
We begin with a list of country demonyms in -ər (41) .As can be seen, various adjustments
take place in order for stress to lodge on the pre-schwa syllable.
(41)
Country demonyms in -ər
Äˈgypten
Alˈbanien
Alˈgerien
Alˈger[i̯]-ər
Äquatorialguiˈne-ər
Arˈmenien
Arˈmen[i̯]-ər
Aserbaiˈdschan
Äthiˈopien
Argenˈtin[i̯]-ər
Aserbaiˈdschan-ər
Äthiˈop[i̯]-ər
Ausˈtralien
Ausˈtral[i̯]-ər
Banglaˈdesch
Banglaˈdesch-ər
Belaˈrus
Belaˈruss-ə
Bahˈrain
Barˈbados
ˈBelgien
Beˈlize
Beˈnin
Bhuˈtan
ˈBosnien
Botsuˈana
Bruˈnei
Burˈkina Faso
Buˈrundi
Côte d'Iˈvoire
10
Alˈban-ər
Äquatorialguiˈnea
Argenˈtinien
9
Äˈgypt-ər
Bahˈrain-ər
Barˈbad[i̯]-ər
ˈBelg[i̯]-ər
Beˈliz-ər
Beˈnin-ər
Bhuˈtan-ər
ˈBosn[i̯]-ər
Boˈts[u̯ ]an-ər
Bruˈnei-ər
Burˈkin-ər
Buˈrund[i̯]-ər
Iˈvor-ər
Besides Iˈrak-i.
Besides ˈKuˌwait, ˈKuˌwait-ər.
38
ˈIndien
Indoˈnesien
Iˈrak
Iˈran
Iˈtalien
ˈJaˌpan
Jorˈdanien
ˈKanada
Kap ˈVerde
Kaˈtar
Kiriˈbati
Koˈmoren
ˌKuˈwait
Leˈsotho
ˈLib[i̯]en
Maˈlawi
Maˈlaysia
Maleˈdiven
ˈMali
Maureˈtanien
Mauˈritius
Mazeˈdonien
Mikroˈnesien
ˈInd-ər
Indoˈnes[i̯]-ər
Iˈrak-ər
Iˈran-ər
Itaˈl[i̯]en-ər
Jaˈpan-ər
Jorˈdan[i̯]ər
Kaˈnad[i̯]ər
Kap ˈVerd[i̯]-ər
Kaˈtar-ər
Kiriˈbat[i̯]-ər
Koˈmor-ər
ˌKuˈwait-ər
Leˈsoth-ər
ˈLib[i̯]-ər
Maˈlaw[i̯]-ər
Maˈlays[i̯]-ər
Maleˈdiv-ər
ˈMal[i̯]-ər
Maureˈtan[i̯]ər
Mauˈrit[i̯]-ər
Mazeˈdon[i̯]-ər
Mikroˈnes[i̯]-ər
9
10
Armin Mester & Junko Ito
アーミン・メスター,伊藤順子
A Note on Unstressability
強勢不可能性に関する一考察
Dschiˈbuti
Dschiˈbut[i̯]-ər
Eriˈtrea
Eriˈtre-ər
ˈGambia
Geˈorgien
Greˈnada
Grenaˈdinen
Guiˈnea
Guˈyana
Herzeˈgowina
ˈSambia
São Toˈmé
ˈSchweiz
Seyˈchellen
Sierra Leˈone
Simˈbabwe
Soˈmalia
ˈSpanien
Sri-ˈLanka
ˈGamb[i̯]-ər
Geˈorg[i̯]-ər
Greˈnad-ər
Grenaˈdin-ər
Guiˈne-ər
Guˈyan-ər
Herzegoˈwin-ər
ˈSamb[i̯]-ər
São Toˈmé-ər
ˈSchweiz-ər
Seyˈchell-ər
Sierra Leˈon-ər
Simˈbabw-ər
Soˈmal[i̯]-ər
ˈSpan[i̯]ər
Sri-ˈLank-ər
Gaˈbun
Monteˈnegro
Monteneˈgrin-ər
ˈNiger
Oˈman
ˈPakistan
ˈNigr-ər
Oˈman-ər
Pakisˈtan-ər
ˌPhilipˈpinen
Ruˈanda
Saloˈmonen
St.ˈKitts
ˈSyrien
Tansaˈnia
ˈTschad
Tuˈnesien
Tuˈvalu
Uˈganda
Ukraˈine
ˈZypern
ˌPhilipˈpin-ər
Ruˈand-ər
Saloˈmon-ər
St. ˈKitts-ər
ˈSyr-ər
Tansaˈni-ər
ˈTschad-ər
Tuˈnes[i̯]-ər
Tuvaˈlu-ər
Uˈgand-ər
Ukraˈin-ər
ˈZypr-ər
Naˈmibia
Gaˈbun-ər
Papua-Neuguiˈnea
Naˈmib[i̯]-ər
11
Papua-Neuguiˈne-ər
12
In a special set of cases primary stress does not manage to fall on the pre-schwa syllable, but
there is still a secondary stress in this position, securing a foottail-status for the schwa syllable.
(42)
a.
b.
11
12
13
14
15
Secondary stress before -ər (ˈNiederˌländ-ər, etc.)
Compounds and pseudo-compounds
ˈEngˌland
ˈIsˌland
ˈLiechtenˌstein
ˈLuxemˌburg
ˈMolˌdau
ˈNiederˌlande
Others
ˈKameˌrun
ˈParaˌguay
ˈEngˌländ-ər
ˈIsˌländ-ər
ˈLiechtenˌstein-ər
ˈLuxemˌburg-ər
ˈMolˌdau-ər
ˈNiederˌländ-ər
ˈNorˌwegen
ˈÖsterˌreich
ˈSwasiˌland
ˈThaiˌland
ˌNeuˈseeˌland
ˈNorˌweg-ər
ˈÖsterˌreich-ər
ˈSwasiˌländ-ər
ˈThaiˌländ-ər
ˌNeuˈseeˌländ-ər
ˈKameˌrun-ər
ˈParaˌguay-ər
ˈSingaˌpur
ˈSuriˌname
ˈSingaˌpur-ər
ˈSuriˌnam-ər
14
Besides Pakisˈtan-i.
Besides ˌFilipˈpin-o.
Besides ˈSwas-i.
Besides ˌKameˈrun, ˌKameˈrun-ər.
Besides ˌSuriˈnam-ər, and ˌSurinaˈm-es-e.
39
13
15
Armin Mester & Junko Ito
アーミン・メスター,伊藤順子
A Note on Unstressability
強勢不可能性に関する一考察
Next, we give examples with the interfix -an-, which attracts the stress and so makes sure
that the FOOTTAIL-ə constraint is fulfilled.
Country demonyms in ˈ-an-ər
(43)
Anˈdorra
Andorˈr-an-ər16
Anˈguilla
Anguilˈl-an-ər
Anˈgola
Anˈtigua
Aˈruba
Angoˈl-an-ər
Antiˈg[u̯ ]-an-ər
Aruˈb-an-ər
Barˈbuda
Barbuˈd-an-ər
Boˈlivien
Boliˈv[i̯]-an-ər
Berˈmudas
Braˈsilien
Bermuˈdan-ər
Brasiˈl[i̯]-an-ər
CostaˈRica
Costa Riˈc-an-ər
Ecuaˈdor
Ecuadori-ˈan-ər
ˈFidschi
Fuˈtuna
ˈGuadeloupe
Guˈam
Haˈiti
Honˈduras
Fidschi-ˈan-ər
Doˈminica
El ˈSalvador
Koˈrea
Dominiˈc-an-ər
Salvadori-ˈan-ər
Futuˈn[i̯]-an-ər
Guadeloupi-ˈan-ər
Guaˈm-an-ər
Haiti-ˈan-ər
Honduˈr-an-ər
Jaˈmaika
Jamaiˈk-an-ər
Kamˈbodscha
Kamboˈdsch-an-ər
ˈKenia
Keni-ˈan-ər
Koˈlumbien
Kolumbi-ˈan-ər
ˈMalta
Malˈt-es-ər
Also:
ˈKuba
Liˈberia
Maˈca.u
Maˈrokko
ˈMexiko
Mosamˈbik
Nicaˈragua
Niˈgeria
ˈNiue
Peˈru
Puerto ˈRico
Saˈmoa
St. ˈHelena
St. Luˈcia
Südˈafrika
ˈUruˌguay
Vanuˈatu
Venezuˈela
Vereinigte Staaten
von Aˈmerika
ˈWallis
Kore-ˈan-ər
Kuˈb-an-ər
Liberiˈan-ər
Maca.u-ˈan-ər
Marokˈk-an-ər
Mexiˈk-an-ər
Mosambiˈk-an-ər
Nicaraˈg[u̯ ]-an-ər
Nigeriˈan-ər
Niu-ˈan-ər
Peru-ˈan-ər
Puerto Riˈc-an-ər
Samo-ˈan-ər
St. Heleˈn[i̯]-an-ər
St. Luci-ˈan-ər
Südafriˈk-an-ər
ˌUrug[u̯ ]ay-ˈan-ər
Vanuatu-ˈan-ər
Venezoˈlan-ər
17
18
Ameriˈk-an-ər
Walliˈs[i̯]-an-ər
Zentralafriˈkanische Zentralafri k-an-ər
ˈ
Repuˈblik
Finally, we turn to country demonyms in -ə. A list of forms derived by this suffix alone appears in (44) .In all of these cases, stress falls on the pre-ə syllable.
16
17
18
Another morphological interpretation takes the a as part of the base and analyses the interfix as -n-.
Besides ˈUruˌg[u̯ ]ay-ər and ˌUruˈg[u̯ ]ay-ər.
Besides Venezueˈl-an-ər.
40
Armin Mester & Junko Ito
アーミン・メスター,伊藤順子
(44)
A Note on Unstressability
強勢不可能性に関する一考察
Country demonyms in -ə
Afˈghanistan
Bulˈgarien
ˈDänemark
ˈDeutschland
ˈEstland
ˈFinnland
ˈFrankreich
ˈGriechenland
Grossbriˈtannien
Guateˈmala
ˈIrˌland
Jugoˈslawien
ˈKasachstan
Kirˈgisistan
Kroˈatien
ˈLaos
ˈLettland
Madaˈgaskar
Afˈghan-ə
Bulˈgar-ə
Mazeˈdonien
Mazeˈdon-ə
Mongoˈlei
Monˈgol-ə
Moˈnaco
ˈDän-ə
ˈDeutsch-ə
ˈEst-ə
ˈFinn-ə
Franˈzos-ə
ˈGriech-ə
ˈBrit-ə
Guatemalˈtek-ə
ˈIr-ə
Jugosˈlaw-ə
Kaˈsach-ə
Kirˈgis-ə
Kroˈat-ə
Laˈot-ə
ˈLett-ə
Madaˈgass-ə
Myanˈmar
ˈPolen
Ruˈmänien
ˈRussland
ˈSchottland
ˈSchweden
ˈSerbien
Slowaˈkei
Sloˈwenien
Taˈdschikistan
ˈTschechien
Türˈkei
Turkˈmenistan
Usˈbekistan
Moneˈgass-ə
Myanˈmar-ə
ˈPol-ə
Ruˈmän-ə
ˈRuss-ə
ˈSchott-ə
ˈSchwed-ə
ˈSerb-ə
Sloˈwak-ə
Sloˈwen-ə
Taˈdschik-ə
ˈTschech-ə
ˈTürk-ə
Turkˈmen-ə
Usˈbek-ə
Next, we give examples with interfixes (-es-, and some others) , which attract the stress and
so make sure that the FOOTTAIL-ə constraint is fulfilled.
(45)
Country demonyms in -ˈes-ə
ˈChina
Kaˈlabrien
ˈKongo
ˈLibanon
Neˈpal
Also:
ˈChile
ˈPortugal
Chiˈn-es-ə
San Maˈrino
Kalaˈbr-es-ə
San Mariˈn-əs-ə
Nepaˈl-es-ə
ˈSenegal
Suˈdan
Taiˈwan
Vietˈnam
Vietnam-ˈes-ə
Chiˈl-en-ə
Saˈvoyen
Savoˈy-ard-ə
Kongoˈl-es-ə
Libaˈn-es-ə
Portuˈgies-ə
Senegaˈl-es-ə
Sudaˈn-es-ə
Taiwaˈn-es-ə
There are some exceptions to the pre-ə stress generalization among country demonyms, but
they are few in number.
41
Armin Mester & Junko Ito
アーミン・メスター,伊藤順子
(46)
A Note on Unstressability
強勢不可能性に関する一考察
Exceptions to the pre-ə stress generalization:
Baˈhamas
ˈGhana
Giˈbraltar
Guiˌnea-Bisˈsa.u
ˈLitauen
ˈMarshallinseln
Na.ˈuru
ˈNevis
Paˈla.u
Baˈhama-ər
ˈGhana-ər
Giˈbraltar-ər
Guiˌnea-Bisˈsa.u-ər
ˈLitau-ər
ˈMarshall-ər
Na.ˈuru-ər
ˈNevis-ər
Paˈla.u-ər
ˈPanama
Saudi-Aˈrabien
St. ˈVincent
ˈTimor
Toˈbago
ˈTogo
ˈTonga
ˈTrinidad
ˈPanama-ər
Saudi-ˈArab-ər
St. ˈVincent-ər
ˈTimor-ər
Toˈbago-ər
ˈTogo-ər
ˈTonga-ər
ˈTrinidad-ər
It is significant that most of the demonyms in (46) have alternate variants which conform to
the generalization (47) . Even a form like Saudi-Aˈrab[i̯] -ər, which sounds outlandish because of the well-established word Arabər, is attested in non-facetious uses.
(47)
Alternate variants of the demonyms in (46)
Bahaˈm-an-ər
Panaˈm-es-ə~Panaˈm-en-ə
Ghaˈn-es-ə
Saudi-Aˈrab[i̯]-ər
Guinea-Bissau-ˈan-ər
Timoˈr-an-ər~Timoˈr-es-ə
Gibralˈt-an-ər
St. Vinˈcent-ər
Tobagoˈl-es-ə
Marchalˈl-an-ər, Marshall-Insuˈlanər
Togoˈl-es-ə
Na.uru-ˈan-ər
Tonˈg-an-ər~Tongaˈl-es-ə~ Tongaˈn-es-ə
Neviˈs-an-ər
Trinidaˈn-es-ə
Pala.u-ˈan-ər
The fact that this is not simply some kind of internet noise can be deduced from the fact that
speakers are explicitly aware of the prosodically problematic status of the forms in (46) , as
shown by blog entries such as the following: "[W]ie nennt man auf Deutsch wohl einen
Menschen aus Guinea-Bissau? Guinea-Bissauer, Guineer (Bissau) , Guinea-Bissauaner oder
gar Guinea-Bissaui? Das ist keine Scherzfrage. Ich muss einen Bericht schreiben..." ("What
does one call a person from Guinea-Bissau in German?... This question is not a joke. I have
to write a report...", and "was bin ich den[n]nun...? ein litauaner.., litaunese..,
litaunippizaner...?"19 ("So what am I?...") .
Matters are different with demonyms derived from city names. In this case, -ər is added
without any stress-related restrictions, as can be seen in (48) .We can attribute the behavior
of this suffix to the OO-IDENT (FOOTHEAD) constraint formulated in (35) above.
19
http://www.wer-weiss-was.de/theme143/article4660166.html and http://board1.2min.eu/showthread.php?t
=20&page=12, both retrieved on 11/19/2010.
42
Armin Mester & Junko Ito
アーミン・メスター,伊藤順子
(48)
A Note on Unstressability
強勢不可能性に関する一考察
Demonyms derived from city names
ˈAltona
Bad ˈOeynhausən
Berˈlin
ˈBielefeld
ˈBochum
ˈDeggendorf
ˈDüsseldorf
ˈEinbeck
ˈEnschede
ˈFrankfurt
ˈGöttingen
ˈHamburg
ˈHerford
ˈJena
21
ˈAltona-ər
Bad ˈOeynhausən-ər
Berˈlin-ər
ˈBielefeld-ər
ˈBochum-ər
ˈDeggendorf-ər
ˈDüsseldorf-ər
ˈEinbeck-ər
ˈEnschəd-ər
ˈFrankfurt-ər
ˈGötting-ər
ˈHamburg-ər
ˈHerford-ər
ˈJena-ər
20
ˈKöln
ˈKrefeld
ˈLeipzig
ˈLemgo
ˈLokkum
ˈMoskau
ˈMünchən
ˈPassau
Paˈris
ˈPotsdam
ˈWien
ˈWilsəde
ˈWuppertal
ˈKöln-ər
ˈKrefeld-ər
ˈLeipzig-ər
ˈLemgo-ər
ˈLokkum-ər
ˈMoskau-ər
ˈMünchən-ər
ˈPassau-ər
Paˈris-ər
ˈPotsdam-ər
ˈWien-ər
ˈWilsəd-ər
ˈWuppertal-ər
Some of these cases probably have a secondary stress on the pre-suffixal syllable (ˈBieleˌfeldər, etc.) , the issue is difficult to decide and requires a separate investigation. Except for cases
that have compound status (such as ˈKreˌfeld-ər) , a secondary stress can generally be excluded immediately after a stressed syllable (ˈLeipzig-ər, * ˈLeipˌzig-ər) and is out of the
question when the syllable in question contains schwa (ˈMünchən-ər, Bad ˈOeynhausən-ər) .
Some special cases appear in (49)–unsurprisingly, all of these fulfill the poststress-ə restriction.
(49)
Special cases
Hanˈnover
ˈHalle
Hannover-ˈan-ər
Halˈl-ens-ər
3. Conclusion
The theory of prosodic constituent structure is built on a three-way distinction between foothead positions, foot-nonhead positions, and unfooted positions. One kind of empirical evidence for distinguishing two kinds of weak positions, the footed and the unfooted, going beyond the simple dichotomy of the stressed vs. the unstressed, comes from the observation
that certain items must not just be unstressed, but are required to occupy the weak position of
a foot. German schwa, with its characteristic post-stress prosodic signature, is an example of
this kind.
20
21
Besides Bad ˈOeynhaus-ər.
Besides Jeˈn-ens-ər.
43
Armin Mester & Junko Ito
アーミン・メスター,伊藤順子
A Note on Unstressability
強勢不可能性に関する一考察
References
Ball, Douglas (2003) Insights into the patterns of sounds: Stress in Tondano. Journal of Undergraduate Research,
University of Rochester 2:15–19.
Bennett, Ryan (2009) Irish plural allomorphy and output optimization. Ms. UC Santa Cruz.
Benua, Laura (1997) Transderivational identity: Phonological relations between words. Doctoral dissertation, University of Massachusetts, Amherst. [ROA-259-0498].
Cohn, Abigail and John J. McCarthy (1994) Alignment and parallelism in Indonesian phonology. Ms. Cornell University and University of Massachusetts, Amherst, Ithaca, NY, and Amherst, MA.
Féry, Caroline (1998) German word stress in Optimality Theory. Journal of Comparative Germanic Linguistics
2:101–142.
Fuhrhop, Nanna (1998) Grenzfälle morphologischer Einheiten. Tübingen: Stauffenburg.
Koepcke, Michael (1995) Die Klassifikation der schwachen Makulina in der deutschen Gegenwartssprache.
Zeitschrift für Sprachwissenschaft 14:159–180.
Liberman, Mark and Alan Prince (1977) On stress and linguistic rhythm. Linguistic Inquiry 8:249–336.
McCarthy, John J. (2008) The serial interaction of stress and syncope. Natural Language and Linguistic Theory
26:499–546.
Piggott, Glyne (2010) Cyclic spell-out and the typology of word minimality. Ms. McGill University, Montreal.
Sneddon, J. N. (1975) Tondano phonology and grammar. Pacific linguistics, Series B. Canberra: Dept. of Linguistics, Research School of Pacific Studies, Australian National University.
Zonneveld, Wim, Mieke Trommelen, Michael Jessen, Curtis Rice, Gösta Bruce and Kristjan Árnason (1999) Wordstress in West-Germanic and North-Germanic languages. In: Harry van der Hulst (ed.) Word prosodic systems
in the languages of Europe, 477–603. Berlin & New York: Mouton de Gruyter.
Armin Mester
Invited Professor, NINJAL (September 2010−December 2010)
Present Position:
Professor of Linguistics, University of California, Santa Cruz (since 1997)
University Education:
1978 Erstes Staatsexamen, Georg-August-Universität Göttingen, Germany
1986 Ph.D. in Linguistics, University of Massachusetts, Amherst
Major Publications and Papers:
1986 The phonology of voicing in Japanese: Theoretical consequences for morphological accessibility [with Junko Ito]. Linguistic Inquiry 17: 49–73.
1990 Patterns of truncation. Linguistic Inquiry 21: 478–485.
1994 The quantitative trochee in Latin. Natural Language and Linguistic Theory 2: 1–61.
1999 The structure of the phonological lexicon [with Junko Ito]. In: Natsuko Tsujimura (ed.) The
handbook of Japanese linguistics, 62–100. Malden, MA, and Oxford, U.K: Blackwell.
2010 Recursive prosodic phrasing in Japanese [with Junko Ito]. In the Proceedings of the 18th
Japanese/Korean Conference. Stanford, CA: CSLI.
Junko Ito
Present Positions:
-Professor of Linguistics, University of California, Santa Cruz (since 1996)
-Director, University of California, Tokyo Study Center (since 2009)
-Invited Professor, International Christian University, Tokyo
University Education:
1979 M.A., B.A. International Christian University, Tokyo
1986 Ph.D. in Linguistics, University of Massachusetts, Amherst
Major Publications and Papers:
1985 Melodic dissimilation in Ainu. Linguistic Inquiry 15: 505–513.
1989 A prosodic theory of epenthesis. Natural Language and Linguistic Theory 7: 217–259.
1990 Prosodic minimality in Japanese. In: K. Deaton, M. Noske and M. Ziolkowski (eds.) CLS
26: Parasession on the Syllable in Phonetics and Phonology, 213–239. Chicago, Chicago
Linguistic Society.
2003 Japanese morphophonemics: Markedness and word structure [with Armin Mester]. Linguistic Inquiry Monograph Series 43. Cambridge, Massachusetts: MIT Press.
2009 Lexical classes in phonology [with Armin Mester]. In: Shigeru Miyagawa and Mamoru Saito
(eds.) Handbook of Japanese linguistics, 84–106. Oxford: Oxford University Press.
44