Phonology 23 (2006) 225–257.

f 2006 Cambridge University Press

doi:10.1017/S0952675706000893 Printed in the United Kingdom

Word-prosodic typology*
Larry M. Hyman
University of California, Berkeley

Numerous proposals have been advanced as to how prosodic systems should be

typologised. In this paper I distinguish two prototype systems, tone and stress
accent, which, unlike systems analysed as ‘ pitch accent’, have two inviolable,
definitional properties : (i) obligatoriness (every word has at least one stress ac-
cent) ; (ii) syllable-dependency (the stress-bearing unit is necessarily the syllable).
In contrast, the oft-cited criterion of culminativity (every word has at most one
tone/accent) not only includes tone/accent systems that are neither obligatory nor
syllable-dependent, but also culminative non-prosodic features that are clearly
not accentual. I argue that there is no one pitch-accent prototype. Instead, since
tone and stress accent may co-occur, and since languages may ‘ pick and choose ’
between the non-definitional properties that tend to cluster within the tone
vs. stress-accent prototypes, there is a range of intermediate (and possibly inde-
terminate) word-prosodic systems which may or may not be best seen as ‘ types ’.

If stress is monarchic, and length oligarchic, we

may say that tone is democratic.
(Greenberg & Kaschube 1976 : 9)

1 Introduction
Over the past several decades, many linguists have made pronouncements
on how prosodic systems should be typologised. While no other area of
phonological typology has attracted as much attention, it is not clear how
much consensus there is on where things currently stand. Linguists
sometimes speak or write on the subject with great conviction, but the
views that are expressed often conflict on both general and specific issues,
e.g. whether all human languages must have ‘ stress’, whether the prosodic
system of Tokyo Japanese is ‘tonal ’ vs. ‘accentual’, and so forth. In this
paper I attempt to sort out some of the complexities of word-prosodic
typology. I begin in §2 with a brief discussion of problems inherent in
doing typology, particularly when applied to phonology. In the next two
sections I give brief definitions of tone (§3) and stress accent (§4), followed

* This paper was first presented as an invited talk at the conference Between Stress and
Tone (BeST). I would like to thank Bert Remijsen and Vincent van Heuven for the
invitation and those in attendance for their helpful comments. I have particularly
profited from extended discussions with Carlos Gussenhoven and José Hualde
on the tone vs. accent question and good Bantu exchanges with David Odden and
Gérard Philippson. I also would like to thank the editors and anonymous reviewers
for their very thoughtful comments on the original manuscript.
225
226 Larry M. Hyman
by a longer section (§5) which addresses the more problematic notions of
‘ accent’ and ‘pitch accent’. I argue that the highest-level typological cut
consists of identifying two prototypes, one with stress accent, the other
with tone. While it is possible to define these prototypes and establish a set
of properties that typically co-occur in each, establishing a third pitch-
accent prototype is more elusive. This is, in part, because languages which
have been identified as pitch accent freely ‘ pick and choose ’ between the
prototypical properties of stress-accent systems vs. tone systems. In other
words, there are many intermediate word-prosodic systems which are not
best seen as discrete types. While most of the discussion centres around
typological issues resulting from structural or systemic difference, §6 ad-
dresses the role of phonetics in typologising prosodic systems. §7 provides
a brief conclusion.

2 Typology
Since the concern of this study is to address the typology of word-prosodic
systems, it seems appropriate to begin by asking the questions : What is
typology ? Phonological typology ? Word-prosodic typology ? Concerning
the first question, Hagège (1992: 7) provides a traditional definition of
typology as ‘ a principled way of classifying the languages of the world by
the most significant properties which distinguish one from another ’. Vajda
(2001) answers the second question in a similar fashion : ‘it is possible to
classify languages according to the phonemes they contain º typology is
the study of structural features across languages. Phonological typology
involves comparing languages according to the number or type of sounds
they contain ’. It is significant that the above definitions speak of classifying
languages rather than subsystems of languages. In any case, only certain
parameters impress linguists enough to establish language types: within
phonology, there is a class of ‘click languages ’, but not ‘implosive
languages ’, ‘ open syllable languages ’, but not ‘onset languages ’ (meaning
that every syllable has to have an onset). The question is whether there is
any reality in classifying languages rather than analysing the properties of
the relevant subsystems. This issue is particularly pronounced in the area
of prosody, where the practice of distinguishing ‘stress languages’ from
‘ tone languages ’ is well established and has encouraged some to propose
additional types: ‘pitch-accent language ’, ‘restrictive tone language ’ and
so forth. However, as Greenberg (1974 : 14) notes, ‘ the same data can
be utilized either for a typology of linguistic properties or a typology of
individual languages’. A major aim of this study is to show that there is
considerably more diversity in prosodic systems than such labels have thus
far distinguished.
Most phonologists would probably agree that there is little, if any,
difference between doing phonological typology and doing phonological
theory. At the very least, it is unlikely that one can do insightful typology
without addressing the same analytical issues that confront phonological
 Word-prosodic typology 227
theory. In order to appreciate the complexity involved in doing proper
typology, consider the question of what phonologists might seek to typo-
logise. There are at least three dimensions to this issue:
(i) What are the properties to be typologised ? Segments ? Systems ?
Rules?
(ii) At what level should the properties be typologised ? Underlying
(morphophonemic)? Lexical (phonemic) ? Surface (phonetic) ?
(iii) Within what domain should the properties be typologised? The
morpheme? Word? Phrase?
For many phonological properties, one will obtain quite different re-
sults depending on the answers to the above questions. In some languages,
word-level prosody is predictable by rule (e.g. demarcative stress, mor-
phological tone) and hence is not present in underlying representations
(URs). In other languages, the prosodic properties are introduced only at
the phrase level.
In this study I take the following approach to the above three questions :
(i) The properties to be typologised are linguistic systems. Since there is
considerable room for interpretation, care must be taken to distinguish the
descriptive ‘ facts ’ of linguistic systems from the analyses that attempt to
account for them. On the other hand, a simple inventory is not a linguistic
system.
(ii) Linguistic systems necessarily capture the relationship between
underlying, lexical and phonetic representations. However, this study is
primarily concerned with properties which are present at the lexical level,
whether from URs or introduced by rule.
(iii) As seen in the title, this study is limited to prosodic properties
which hold within the word domain vs. properties which make reference
to the phrase or utterance levels.
Among the frequent assumptions underlying past attempts at phono-
logical typology are: (i) EXHAUSTIVITY : all systems can be assigned to
a type ; (ii) UNIQUENESS : no system can be assigned to two types ; (iii)
DISCRETENESS : types do not overlap. By these criteria we arrive at the
classic bi- and trifurcation of prosodic types in (1a, b).

(1) Classic bi- and trifurcation of prosodic types

a. b.
stress tone stress tone pitch
accent accent accent
c.

stress * pitch tone

accent accent
228 Larry M. Hyman
In (1a), prosodic systems are identified as either stress accent (SA)
or tone (T), while (1b) allows a third category, pitch accent (PA). As
indicated by the asterisk, the above assumptions do not condone (1c),
where PA belongs simultaneously to two different types, the first with SA,
the second with T.
As attractive as they are, all three assumptions are problematic. First,
despite the discreteness of the putative types (SA, PA, T), prosodic sys-
tems may be ambiguous, i.e. analytically indeterminate. In languages such
as Kinga (Schadeberg 1973) and Nubi (Gussenhoven 2006), every word
must have one and only one high tone. Should such systems be categorised
as ‘restricted ’ T, as PA or as SA, where stress would be realised simply
as H pitch? All three interpretations seem reasonable, perhaps not only
to linguists but also to the native speakers. Given this indeterminacy, it
may be that such languages provide the transitional ‘pivot ’ for change
from one type of system to another (cf. Salmons 1992, Gussenhoven
2006).
Besides ambiguity, a second issue is that prosodic systems may be
mixed. In Saramaccan (Good 2004), in one class of words a single syllable
is marked /H/, the remaining syllables being underlyingly toneless. On
the other hand, in a second class of words, all syllables must be specified
for tone (all H, all L or combinations of H and L). While Saramaccan has
a stratified lexicon, which Good analyses as PA vs. T, the same words
may have both properties in Cherokee. As analysed by Johnson (2005), a
Cherokee word may or may not have a single L*H pitch accent. Whether
or not it does, a non-accented syllable may also have a contrastive H
or L tone. While Saramaccan and Cherokee combine PA and T, other
languages combine SA and T (see §5). Whatever its usefulness, the three-
way distinction between SA, PA and T cannot therefore be a discrete,
non-overlapping one.1
A final word before addressing SA, PA and T in more detail. The above
discussion assumes that word-prosodic typology is limited to this tri-
chotomy, or the properties it encodes. We have however not explicitly
addressed the question of what counts as word prosody. Words are not
only characterised by the classic three suprasegmentals of tone, length
and stress (Lehiste 1970, Greenberg & Kaschube 1976), but also by
various kinds of harmonies and quantitative/qualitative distributional
constraints concerning minimum/maximum mora/syllable count, initial
vs. final position restrictions, C/V combinatorics and so forth. If word-
prosodic typology were to take all of these issues into account as well, the
result would be an explosion of disparate prosodic word types, as indicated
in (2).

1 Another kind of mixed system has been occasionally proposed, based on different
levels of representation. Although subject to other interpretations, McCawley
(1978: 126) suggests that Luganda ‘ can be described as having a pitch-accent sys-
tem in its deep phonology and a tonal system in its surface phonology ’.
 Word-prosodic typology 229

(2) Di‰erent kinds of prosodic word?

a. The demarcative word
A property marks the beginning or end of the word.
b. The culminative word
A property occurs only once per word.
c. The harmonic word
A property is realised throughout the word.
d. The metrical word
A word consists of hierarchically arrayed moras or syllables.
e. The minimum word
A word must consist of a minimum number of moras or syllables.
f. The maximum word
A word can consist of a maximum number of moras or syllables.
g. The phonotactic word
A word permits only certain output segments/sequences.
h. The morphophonotactic word
A word permits only certain input segments/sequences.

Beyond practicality there are legitimate reasons to group tone, length and
stress together. First, there is the intuitive recognition of all three as
overlaid suprasegmentals (Lehiste 1970), whose phonetic variations (F0,
duration, intensity) can be realised not only in spoken language, but also
on a musical instrument. Second, these three phonetic properties fre-
quently interact, e.g. in the realisation of stress and intonation. Finally,
tone, length and stress also share functional similarities, e.g. culminativity
and rhythmic alternations.
For the above reasons, the present study on word-prosodic systems will
be limited to T, SA and PA, and their manipulation of F0, duration and
intensity. In the following sections, tone, stress and (pitch) accent are
taken up in turn.

3 Tone
In order to establish prosodic types in a reasonably objective way, it is
necessary to provide definitions for tone, stress and accent. I will assume
the following as a working definition of tone (Hyman 2001a : 1367, slightly
modified from Welmers 1959, 1973) :

(3) A language with tone is one in which an indication of pitch enters into
the lexical realisation of at least some morphemes.
As indicated, the domain of tone is the morpheme. Depending on the
language, morphemes with tone may be all, many or few. They can be
230 Larry M. Hyman
lexical or grammatical morphemes, including tonal morphemes which
lack a segmental content. According to (3), the presence of tone is estab-
lished on the basis of lexical realisation, i.e. the output of the lexical pho-
nology. Tone may thus either be present in URs or be introduced by rule.
If this latter occurs at the lexical level, rather than postlexically, this would
count as tone by the above definition.
Tone is featural and paradigmatic : in a prototypical tone system such as
Cantonese or Yoruba, one IDENTIFIES the tone of each tone-bearing unit
(TBU). In other words, tone has a ‘distinctive function ’ (Martinet 1960).
The tone system may consist of a simple binary opposition, e.g. /H, L/, or
involve more levels, contour tones and specific tones accompanied by
phonation types (glottalisation, breathiness, etc.).
The definition in (3) refers to ‘an indication of pitch’ rather than pitch
itself. This is intentional, and is designed to extract away from analytic
preferences. For example, it was mentioned in §2 that Kinga and Nubi
require exactly one H tone per lexical word. The question was also raised
as to whether these systems should be thought of as SA, PA or T. They
could thus be analysed with a /H/ tonal feature or an abstract accent,
which could be said to be realised as [H] by a postlexical pitch-assignment
rule. However, since only pitch is involved, it is clear that the abstract
accent is nothing but ‘an indication of pitch’. By (3), Kinga and Nubi are
tonal.
The question that is relevant in this context is whether we want to ty-
pologise according to the properties of prosodic systems or according to
the analyses given to them by diverse linguists. This issue has arisen in a
number of cases, the most prominent of which concerns Tokyo Japanese.
As seen in (4), Tokyo Japanese has been analysed both accentually and
tonally (Haraguchi 1977, McCawley 1978, Poser 1984, Pierrehumbert &
Beckman 1988, etc.) :

(4) accentual ma0kura ga koko0ro ga atama0 ga sakana ga

tonal makura ga kokoro ga atama ga sakana ga

H H H
output mákùrà gà kókórò gà átámá gà sákáná gá
output mákùrà gà kòkórò gà àtámá gà sàkáná gá
with ª
‘pillow+nom’ ‘heart+nom’ ‘head+nom’ ‘fish+nom’

In the accentual account in the first row, the mark 0 indicates where a drop
occurs between H and L. (Others have used a tick mark or asterisk.) The
tonal analysis in the second row starts with a /H/ prelinked to the mora
that precedes the pitch drop. The third and fourth rows show that moras
following the accent or /H/ are L, while the remaining moras are H, except
where an initial x boundary tone affects the first mora.
A final note concerning the notion of ‘word tone ’ (Donohue
1997, Mazaudon 2005). In the definition in (3) tone is identified with the
 Word-prosodic typology 231
morpheme, and nothing is said about the TBU, which can be the syllable
or the mora. In Tamang, Kukuya and a number of other languages, there
is a fixed number of word- (or stem-) tone patterns independent of the
number of syllables or moras (cf. the discussion of Skou in §5). For such
systems it is sometimes suggested that the word takes precedence over the
morpheme, or is even the TBU, rather than the domain determining the
tone (cf. Garde’s 1968 : 12–13 distinction between ‘unité accentuable’ and
‘unité accentuelle ’ applied to stress systems). Like Tokyo Japanese, there is
a sense in which such languages diverge from what I have called proto-
typical tone systems. As in the case of vowel-harmony systems, such tone
systems have a syntagmatic dimension which is lacking in Cantonese,
Yoruba, etc., where tone contrasts paradigmatically on syllables.

4 Stress
Whereas tone has to do with pitch features, the general approach to
stress has been that it has to do with metrical PROMINENCE. The definition
I would propose for stress accent is given in (5).
(5) A language with stress accent is one in which there is an indication of
word-level metrical structure meeting the following two central criteria:
a. obligatoriness: every lexical word has at least one syllable marked
for the highest degree of metrical prominence (primary stress);
b. culminativity: every lexical word has at most one syllable marked
for the highest degree of metrical prominence.
In many characterisations of SA, the phrase ‘one and only one’ primary
stress is used : ‘every prosodic word contains one and only one head foot,
which is the locus of main stress. The existence and uniqueness of the
head foot are usually taken to be axiomatic – universal properties of GEN
rather than violable constraints’ (McCarthy 2003: 110). As indicated
in (5), it is however fruitful to separate the phrase into two properties :
obligatoriness and culminativity. Obligatoriness states that every lexical
word has to have a primary stress. Although there is variation in the use
of the term in the literature, culminativity is intended here to mean that
a lexical word cannot have more than one primary stress.
In contrast to tone, which is featural and paradigmatic, stress is struc-
tural and syntagmatic. Prototypical tone has a distinctive function, while
stress accent has a CONTRASTIVE function (Martinet 1960), even where
minimal pairs are possible. Whereas one IDENTIFIES the T of each TBU,
one LOCATES stress within the lexical word domain.
Of the two criteria in (5), obligatoriness is the more important. It is an
absolute universal – DEFINITIONAL – of a SA system, which requires that
an obligatorily headed metrical constituent be built at the word level.
Rather than HEAD(PWd), ‘ which says that each phonological word
must have a unique head and therefore exactly one accent’ (McCarthy
232 Larry M. Hyman
2002 : 78), and which thereby conflates obligatoriness and culminativity,
I shall borrow from optimality syntax and refer to obligatoriness as
OBLIGATORYHEAD (OBLHEAD).
Because of obligatoriness, no language fails to assign metrical stress to a
word for lack of a specific phonological property. Thus OBLHEAD rules out
stress systems such as the following :2

(6) a. Stress the penultimate syllable; monosyllabic words are stressless.

b. Stress the last heavy syllable; words that lack a heavy syllable are
stressless.3
c. Stress the initial syllable if it has an onset; words beginning with a
vowel are stressless.
d. Stress the leftmost syllable with /a/, otherwise stress the leftmost
syllable containing /e/ or /o/; words with only /i/ or /u/ are stressless.
e. Stress the last H tone; words lacking a H tone are stressless

What’s wrong with the above systems is that they do not exercise default
stress assignment in case stress-attracting phonological properties are
lacking. Except for this property, (6a–e) would seem to be a priori
reasonable, if not also grounded : stress is often avoided on light syllables,
so why not on a word that consists solely of light syllables ?
In fact, the above systems might be argued to be less ‘crazy’ than what
we actually do find in the interaction between stress and phonological
properties. In Tukang Besi, under appropriate conditions, ‘ gemination
occurs º on the [onset] consonant of the stressed (penultimate) syllable ’
(Donohue 1999 : 34), as in (7a).
(7) Tukang Besi (Donohue 1999)
a. top:áΩa ‘cut branches’ motÇt:mrÇ ‘sleepy’
mEl:ái ‘far’ mo?om:mrÇ ‘hungry’
b. káp:i ‘wing’
pá:Ωa ‘branch’
c. Àk:a ‘climb’
Àl:a ‘tongue’
However, Donohue goes on to state : ‘if the stressed syllable is also
the initial one (i.e., the word is di-syllabic), the gemination can jump
one syllable forward ’. As seen in the examples in (7b), where initial
2 Cf. the Unstressable Word Syndrome of Hayes (1995: 110–113), who also treats
alleged exceptions.
3 Dobrovolsky (1999) considers a possible counterexample in Chuvash, where stress
is assigned to the last full vowel of a word. The question is whether stress is assigned
to the first syllable of words lacking a full vowel, or whether such words are
stressless. Based on a phonetic study of disyllabic words, Dobrovolsky concludes
that ‘ there is something different about stress on reduced vowel words ’ (1999 : 542)
and hypothesises in favour of the interpretation that they are stressless. Thanks to
Susannah Levi for reminding me of this paper.
 Word-prosodic typology 233
gemination is prohibited, the onset of the post-stress syllable is geminated
instead. The same is true in (7c), where the first syllable has no onset. It
would appear that stress-based gemination is important to Tukang Besi
speakers. Rather than suppressing gemination in (7b, c), they shift the
syllable to guarantee gemination will occur. An otherwise reasonable
option would be to restrict gemination to the onset of the stressed syllable,
and block stress assignment to words such as *[kapi], *[pa‰a], *[Eka] and
*[Ela]. This, however, would violate OBLHEAD.
I suggest that word stress has this ‘ obligatory ’ property because stress
accent is DEFINITIONAL of wordhood. A violation of OBLHEAD would im-
ply that stressless words are not words. More generally ; we observe that
local phonology does not override higher levels of grammatical and pro-
sodic organisation. Thus, in a language where all words must begin with
a consonant, a word which is underlyingly vowel-initial will not fail to
become a word. Instead, a default onset may be inserted or the vowel may
be deleted. As will be seen in §5, prosodic systems which are exceptions to
OBLHEAD invariably involve tone or length.
Returning to the definition in (5), a second definitional property of SA
systems is that OBLHEAD obligatorily targets syllables. The same cannot
be said for T systems or PA systems, in which the TBU may be either the
syllable or mora. This property of SA systems makes two interesting
predictions, which are borne out. First, a language which lacks syllables
clearly cannot have SA, e.g. Gokana (Hyman 1985). Second, a language in
which some words lack syllables cannot have SA. The reason for this is
that it would violate OBLHEAD : words with syllables would have stress,
while those which lack this phonological property wouldn’t. According to
Bagemihl (1991), in Bella Coola a syllable must contain a sonorant (as in
most languages). A word consisting solely of obstruents contains no syl-
lables, e.g. [¯ktskW¯] ‘he arrived’. Bella Coola thus provides the test
case – the possibility that words with a sonorant (and hence a syllable) will
have stress, while words lacking a sonorant won’t. Newman (1947 : 132)
clearly states that in Bella Coola there is ‘no phonemically significant
phenomena of stress or pitch associated with syllables or words’. He
continues : ‘when two or more syllabics occur in a word or sentence, one
can clearly hear different degrees of articulatory force. But these relative
stresses in a sequence of acoustic syllables do not remain constant in re-
petitions of the utterance ’. That Gokana and Bella Coola lack SA is thus
predicted by the definition in (5).
It appears that OBLHEAD and syllables are definitional properties of SA
systems in two senses. First, all SA systems have these properties. Second,
systems other than SA are not required to have these properties. Thus,
there can be accentless words in many so-called PA languages (e.g. Tokyo
Japanese, Somali), which are also free to target moras instead of syllables.
Besides culminativity (to which we will return in §5), additional com-
mon properties of SA systems may or may not characterise non-SA sys-
tems. Here I mention four of these, since they have been occasionally
cited as evidence that systems which I would identify as restricted T are
234 Larry M. Hyman
really PA : (i) PRIVATIVITY : stress is either present vs. absent on a given
syllable ; (ii) SUBORDINATION : one stress may be subordinated to another,
as in compounding, focus marking, etc. (McCawley 1970, 1978) ; (iii)
DEMARCATION : stress fixed on initial, final, etc., syllables can tell you where
the word boundary is; (iv) RHYTHMICITY : echo-stresses frequently occur
on every other syllable. With these additional properties identified, we can
now move on to consider the nature of (pitch) accent.

5 Accent
There has been considerable terminological ambiguity as well as concep-
tual confusion concerning the notion of ‘accent’ (cf. Hyman 1977a,
Clements & Goldsmith 1984a, Beckman 1986, Pulleyblank 1986, van der
Hulst & Smith 1988a, Salmons 1992, Ladd 1996, van der Hulst 1999a,
Odden 1999, Fox 2000, Gussenhoven 2004, among others). Some of the
uses of the term ‘accent’ are indicated in (8).
(8) The term accent has been used to refer to the following:
a. Prominence tout court: ‘an accent is simply an ‘abstract’ mark where
a culmination of prosodic features occurs, thereby marking that
syllable (or accent-bearing unit) with greater salience than surrounding
syllables’ (Hyman 1977a: 4).
b. A place of F0 activity (cf. phrase-level ‘pitch accents’, e.g. English
intonational H*L).
c. A lexical diacritic (which is exceptional or unpredictable)
i. used to indicate a potential metrical head;
ii. used to indicate the alignment of a tone.
d. A tone that allegedly acts like a stress (cf. ‘stress accent’ vs. ‘pitch
accent’).
It is the last two uses that are most relevant to the definitions of tone and
stress accent given in §§3 and 4. These are discussed in the next two
subsections.

5.1 Diacritic accent

A representative case of diacritic accent appears in Donohue’s (2003)
treatment of tone in Skou (New Guinea). As seen in the summary of tonal
distributions in (9), Donohue analyses Skou as having five underlying tone
melodies, which show two interesting properties.
 Word-prosodic typology 235

(9) Tonal distributions in Skou (Donohue 2003)

tone melody monosyllabic disyllabic trisyllabic
/H/ H H-H H-H-H
/L/ L L-L L-L-L
/LH/ — L-H L-H-H
/LHL/ — L-HL L-H-L
/HL/ HL H-L, HL-L, H-H-L, H-HL-L,
H-HL H-H-HL
First, since Skou does not permit LH rising tones, /LH/ and /LHL/ do
not exist on monosyllables. Second, /HL/ shows three different mapping
possibilities on di- and trisyllabic words. Donohue’s solution is to indicate
the prelinking of /HL/ tones by means of a diacritic accent, as in (10).

(10) a. 's s b. s 's c. s s

H L H L H L
The /HL/ melody prelinks to the accented first syllable in (10a) vs. the
accented second syllable in (10b). Where there is no accent, as in (10c),
tones link to TBUs on a one-to-one basis.
Donohue (2003: 335) states that ‘an accent is an attractor for tonal
units ’ (cf. Goldsmith 1987). In Skou this accent is limited to either the
final or penultimate syllable of a word. As seen in (11), the diacritic is also
contrastive on /HL/ monosyllables.

(11) Monosyllabic /HL/ and /'HL/ words are indistinguishable in isolation,

but di‰er in compounding
a. HL + L [H-L]
kÉ ta £ kfl.tà ‘beard’ (cf. kû
É ‘jaw’)
‘jaw’ ‘hair’
b. HL + L [HL-L]
'hø na £ hø̂.nà ‘sago flesh’ (cf. hø̂ ‘sago’)
‘sago’ ‘flesh’

The diacritic accent cannot co-occur with /LH/ and /LHL/ melodies,
because of the prohibition against rising tones, or with /H/ and /L/ mel-
odies, because it would have no effect (Donohue 2003: 336–337).
If it could be shown that Donohue’s diacritic marks a potential metrical
head, we would have an argument that there is something accentual about
the Skou tone system. However, it seems that his ‘accent ’ is used only as a
diacritic ‘ indication of pitch ’, as per the definition of tone in (3). The
prelinked representations in (10a, b) could have been given without the
indicated accents. Alternatively, the same outputs would be obtained if
the L were prelinked to the first syllable in (10a), and the H to the second
236 Larry M. Hyman
syllable in (10b). The prosodic system of Skou is therefore exclusively
tonal, but with an interesting complication in the mapping of /HL/
melody.

5.2 Pitch accent

If Skou is strictly tonal, what, then, is a ‘pitch-accent ’ system ? Several
attempts at defining PA assume a syntagmatic accentual component and
focus on its different realisation from stress accent:
(12) Putative realisational di‰erences between PA and SA
a. PA has a constant tonal realisation.
‘In a tonal or pitch-accent system (like Japanese, for instance), in
addition to the abstract accent, there is a constant physical property
[i.e. a tone] associated with the accent’ (Hyman 1977a: 4).
‘Pitch accent languages must satisfy the criterion of having invariant
tonal contours on accented syllables, since tone is a lexical property.
This is not so for pure stress languages, where the tonal contours
of stressed syllables can vary freely, being determined postlexically
by the intonational system’ (Hayes 1995: 49–50).
b. SA has more than a tonal realisation.
‘Hypothesis: Stress accent differs phonetically from non-stress
[pitch-] accent in that it uses to a greater extent material other than
pitch’ (Beckman 1986: 1).
The above definitions attempt to distinguish PA from SA. It is also
necessary to distinguish PA from T, since both employ pitch features
at the word level. As stated by Bybee et al. (1998 : 227), ‘a pitch accent
system is one in which pitch is the primary correlate of prominence
and there are significant constraints on the pitch patterns for words’.
The classic example of Tokyo Japanese would be a case in point, since
the distribution of output H and L tones is highly constrained. How-
ever, languages in which tones are sparsely distributed, e.g. one /H/ per
word, have also been referred to as having a ‘restricted tone system’
(Schadeberg 1973, Voorhoeve 1973), a point argued specifically for Tokyo
Japanese by Meeussen (1972). Since tone systems themselves differ
in the degree to which their tones are distributionally restricted, Tokyo
Japanese-type PA may simply be a tone system at one end of the re-
strictedness scale.
The claim of this paper is that pitch accent is not a coherent notion,
rather a ‘pick and choose ’ among the properties that characterise proto-
typical tone vs. stress-accent systems (Hyman 2001a, b: App.). Despite
vague attempts such as in (12), because there is no single prototype of PA,
it is not possible to give an explicit definition of PA, as one can for T (3)
and SA (5). By PA linguists have generally meant one or more of the
following properties:
 Word-prosodic typology 237

(13) ‘PA-like’ properties

a. A system whose underlying prosody is abstractly di‰erent from
surface realisations.
b. A system which combines tone and stress.
c. A system which has restricted, sparse or privative tone (e.g. /H/ vs. 0).
A representative case of (13a) is Tonga, which Meeussen (1963) analyses
with underlying ‘determinant ’ vs. ‘neutral ’ moras and the following
general algorithm: (i) moras occurring between determinant moras are
realised H; (ii) all other moras are realised L. While Goldsmith (1984)
uses diacritic accents (asterisks) where Meeussen has determinants,
Pulleyblank (1986) argues for a strictly tonal analysis with /H/. It should
be clear that, as in Skou, these diacritics are ‘ indications of pitch’ and not
accentual in the SA sense in (5).
The property in (13b) refers to systems which combine tone and
stress. Table I categorises languages which have been claimed to have
lexical SA, lexical T, both or neither (cf. Hyman 1977a, Ladd 1996: 156,
Fox 2000 : 265) :

+stress accent —stress accent

Ma'ya, Usarufa, Fasu, Serbo- Yoruba, Igbo, Kuki-Thaadow,
Croatian, Swedish-Norwegian, Skou, …
+tone
Ayutla Mixtec, … (Tokyo Japanese, Somali,
Western Basque, …)
English, Russian, Turkish, Bella Coola, French, Tamazight,
—tone
Finnish, … Bengali, …

Table I
Classification of languages according to whether they have
stress accent, tone, both or neither.

A quick glance at the languages listed in the upper left quadrant will es-
tablish that there is no single prosodic type consisting of T+SA. As
summarised in (14), coexistent T and SA may have very different relations
to each other (Hyman 1977a, van der Hulst & Smith 1988a, van der Hulst
1999a):

(14) A PA type as T+SA is insucient

a. {T, SA}: T and SA as independent systems
e.g. Ma'ya (Remijsen 2001, 2002): /H/and /LH/ tones occur only
on the final syllable, contrasting with 0; penultimate and final stress
contrast on words with final /H/; stress is otherwise final.
238 Larry M. Hyman

b. {T; SA}: SA dependent on T

e.g. Usarufa (Bee & Glasgow 1962): in words with a pitch drop,
stress is assigned to a /H/ or /HL/ preceding a L, else to the first
H of an all H word, else to the last L of an all-L word.
c. {SA; T}: T dependent on SA
e.g. Fasu (May & Loeweke 1964): /H/ and /L/ contrast only on the
stressed syllable; non-stressed syllables are toneless.
d. {SA=T}: T and SA as co-dependent systems
e.g. the Neo-Wtokavian dialect of Serbo-Croatian (Zec 1999): stress
is assigned to a pre-/H/ syllable; words without /H/ receive initial
stress, which in turn requires a H tone.
Turning to the properties in (13c), systems in which tones are sparse,
ultimately limited to a single occurrence per word, have often been
identified as PA. However, as seen in the upper right quadrant in Table I,
I have interpreted Tokyo Japanese, Somali and Western Basque as
[+T, —SA], They clearly meet the definition in (3) for T. While some of
their properties may be reminiscent of SA, they do not meet the definition
for SA in (5). Thus, from the point of view of the T vs. SA dichotomy
proposed in §§3 and 4, they, like Skou and Tonga, are strictly tonal. In §4
we claimed that if a system is SA, it will satisfy OBLHEAD and culmi-
nativity, and it may exhibit additional properties of metrical systems (sub-
ordination, demarcation, rhythm). To classify Tokyo Japanese, Somali,
Basque, etc. as PA represents a reversal of the antecedent and consequence
of the condition: if a tone is obligatory, culminative, privative, subordi-
native, demarcative or rhythmic, then it must be PA (by analogy with SA).
Any one of these properties can and has been cited as evidence that some
language is ‘accentual’. While these are important and interesting sub-
properties of prosodic systems, the question is whether any property other
than OBLHEAD suffices to characterise a tone as an accent.
Consider, for example, the culminative property. Some alleged PA
languages have culminative, but not obligatory, underlying accent/tone.
As exemplified in (4), in Tokyo Japanese, there is one or no /H/ per word,
following which there is a drop to L. In Western Basque, there is similarly
one or no /H/ per word, with a drop to L afterwards (Hualde 1991, 1999).
In Somali, a word may have a single /H/, which is assigned to the final or
penultimate vowel according to morphological rules. Subject nouns and
finite verbs in main clauses are toneless (Hyman 1981, Banti 1988, Saeed
1999). In Iraqw, /H/ is assigned mostly by the morphology to the final
syllable of some words. However, ‘ the overall majority of nouns are low
throughout ’ (Mous 1993: 21). It is significant that these systems, which
do not require that every word carry an overt lexical mark, are as easily
analysed in strictly tonal terms – an option which is not available in the
case of such SA languages as English, Russian and so forth.
The condition ‘if culminative, then accent’ is also suspect for the con-
clusions it forces us to draw concerning non-tonal features. In Cuzco
 Word-prosodic typology 239
Quechua, a language with penultimate stress, glottalised and aspirated
consonants have the properties in (15).

(15) Culminative glottalised and aspirated consonants in Cuzco Quechua

(Parker 1997: 2)
a. Root-controlled: occur only in roots, never in suxes.
b. Non-obligatory: occur in some roots, not in all.
c. Culminative: may occur only once per root.
*T’VT’V, *THVTHV, *T’VTHV, *THVT’V
d. Directional: only occur on first stop of root.
T’VTV, THVTV vs. *TVT’V, *T’VTHV
(C’ and CH occur only in onset position of the syllable, never in codas.)
On the basis of its culminative nature and the restriction of glottality to
prominent positions (roots, onsets), should one conclude that Cuzco
Quechua is a ‘glottal accent’ (GA) language?
A similar example concerning vowel length is seen in Mam (Mayan).
As stated by Willard (2004 : 7), ‘only one long vowel is allowed per word.
In addition, the presence of a long vowel in a word disallows the presence
of a glottalised vowel in the same word’. Willard argues that besides being
culminative, Mam vowel length shows dominance effects, another prop-
erty which is frequently identified with SA. As seen in the examples in
(16), a suffix may be dominant or recessive, independent of whether it has
a long or short vowel itself :

(16) Dominant/recessive vowel length in Mam

a. recessive short sux
/ooq’-b’il/ £ ooq’-b’il ‘sth. which causes crying’
b. recessive long sux
/b’iitz-oo-/ £ b’iitz-a- ‘sing’
c. dominant short sux
/q’ooj-le’n/ £ q’oj-le’n ‘state of fighting’
d. dominant long sux
/liich’-ich’iin/ £ lich’-ich’iin ‘breakable’

When a suffix is recessive, as in (16a, b), length is preserved on the pre-

ceding root. If both the recessive suffix and the root have a long vowel, as
in (16b), the culminative condition requires that the former be shortened.
On the other hand, when a suffix is dominant, as in (16c, d), length is lost
on the root. While the shortening seen in (16d) can also be attributed to
culminativity, that in (16c) cannot be: the dominant short-vowel suffix
imposes a short vowel on the root as well. Should one then talk of a
‘quantity accent’ (QA) in Mam?
An argument militating against GA and QA interpretations of Cuzco
Quechua and Mam respectively is that each language has an independent
240 Larry M. Hyman
SA system. Sress is penultimate in Cuzco Quechua, and is assigned in
Mam according to the following hierarchy :

(17) Stress hierarchy in Mam (England 1983, Willard 2004)

VV >> V? >> VC >> V
a. stress a long vowel a.'q’uuntl ‘work’
b. stress the rightmost vowel closed by pu?.'la? ‘dipper’
a glottal stop
c. stress the rightmost vowel closed by xpi.'chaq ‘raccoon’
a consonant
d. stress the penultimate syllable '[Link]’a ‘clear’

While length is implicated in stress assignment in Mam, length and stress

must be considered two separate systems.
Unless we are to proliferate accents to include GA, QA, etc., and allow
for languages to have several competing accentual systems at the same
time, it should be clear that culminativity is not a sufficient criterion
for establishing PA. As we have seen, features other than tone can be
culminative, even demarcative. As seen in Cuzco Quechua and Mam,
such culminative systems can be independent of SA, and of each other.
Thus, culminative H tone and culminative vowel length occur in different
positions in phrasal outputs in Chimwiini (Kisseberth 2005). The features
of glottality, length and tone can have very stress-like qualities, but
crucially not obligatoriness – unless they are in fact implicated in the
realisation of SA. As far as I know, no language has been reported with
two independent obligatory-culminative systems, that is, where every
word obligatorily has two of the following properties : SA, PA, GA, QA
(others ?). The reason for this should be clear: given the definition of
SA in §4, only one system can meet the OBLHEAD requirement or there
would be conflicting definitions of the metrical word.
One can easily imagine hypothetical prosodic systems with indepen-
dent, obligatory SA and PA, e.g. one with penultimate stress and a single
obligatory /H/ which can occur on different syllables. In quest of such a
system, consider the class of Bantu languages which have been analysed
with privative /H/ vs. 0 tone. A subset of these which have lost the his-
torical *V/*V: opposition are reported to have penultimate SA, realised
typically as vowel lengthening (Hyman 1977a, 1989, Odden 1999,
Kisseberth & Odden 2003, Downing 2004). A different, smaller subset of
/H/ vs. 0 Bantu languages has imposed obligatoriness on the /H/ tone.
However, such ‘restricted tone’ systems as found in Kinga (Schadeberg
1973) and Bena (Odden 1988) have kept the vowel-length opposition, and
do not show evidence of penultimate stress. Other languages such as
Chimwiini (Kisseberth & Abasheikh 1974) have retained the historical
vowel-length opposition underlyingly, but restrict surface long vowels to
one per word (or phrase), typically either the penultimate or antepenulti-
mate syllable. In these languages, if there is no underlying long vowel,
 Word-prosodic typology 241
output vowels will all surface as short. I have to date not found a single
Bantu language which has two independent accentual properties in sense of
OBLHEAD, e.g. obligatory culminative /H/ and obligatory culminative
vowel length.4
In summary, where languages have been reported to have two accent-
like systems, one or both of these fail to meet the OBLHEAD requirement.
In Maya (§5.2) stress is obligatory and restricted to the final or penulti-
mate syllable. The /H/ and /LH/ tones which are restricted to the final
syllable are culminative, but not obligatory (words may lack tone alto-
gether). In Cherokee (§2), neither T nor the culminative L*H PA are
obligatory (words may lack both), and so forth.
There appear to be two ways to interpret the results thus far. The first,
which we have been assuming, would be to impose a Unique Obligatory
Head Constraint which would prohibit words from exhibiting more than
one obligatory ‘accentual ’ property. The second interpretation would be
to assert that only SA can meet the OBLHEAD requirement – and therefore,
any prosody properties which meets OBLHEAD must be SA by definition.
This would include obligatory vowel lengthening and obligatory /H/ tone.
Assuming that a word cannot have two SA systems (cf. §6), no language
would have two accent-like systems meeting the obligatoriness require-
ment. Note that this interpretation has the effect of labelling the obliga-
tory /H/ of languages such as Nubi (Gussenhoven 2006) as SA, although
it also seems to meet the definition of T in (3). Gussenhoven (2006)
appropriately suggests that the Nubi prosodic system is ‘pivotal’ in
that its obligatory one-H-syllable-per-word may be interpreted either
as SA or restricted T. While Nubi has one and only one H tone syllable
per word, both Safwa and Kinga exhibit an obligatory one-H-mora-
per-word system, as seen in the realisation of antepenultimate H tone
in (18).5

(18) Antepenultimate /H/ in two obligatory-/H/ Bantu languages

a. Safwa (Voorhoeve 1973)
Sílihu ‘which one (cl 7)’
amáfuko ‘bags’
amíina ‘holes’
tuúbale ‘let us go’
bahwiítiha ‘they believe’

4 Gérard Philippson (personal communication) points out that some Makonde dia-
lects which have penultimate lengthening come close to requiring a H on every
word. The Comorian language Ngazidja appears to be another case where obliga-
tory /H/ is in the process of being imposed – at least at the phrase level. Whether a
metrical account is adopted, as in Cassimjee & Kisseberth (1989), or a tonal domains
account, as in Cassimjee & Kisseberth (1998), words do not require both /H/ and
SA.
5 Voorhoeve (1973: 1) is non-committal as to whether the H pitch which co-occurs
with greater amplitude is a tone or stress.
242 Larry M. Hyman

b. Kinga (Schadeberg 1973)

Ykûheka ‘to laugh’
Ykûvala ‘to count’
YkYgéenda ‘to go’
YkYgeendélela ‘to walk around’
YkYhwaánana ‘to become similar’

In §4 it was required that SA target syllables. Since Safwa and Kinga

assign their obligatory /H/ by mora, their obligatory /H/ cannot be SA. As
Voorhoeve and Schadeberg propose, they can be appropriately viewed as
‘ restricted tone systems’. Their prosodic systems are thus less ‘pivotal’
than Nubi’s, showing again that languages ‘ pick and choose ’ between the
prototypical properties of T vs. SA systems. More evidence for this will be
seen in the next subsection.

5.3 Rhythmic accent

The definition of SA in (5) refers to obligatory metrical structure. A SA
system necessarily identifies an obligatory head syllable for each word.
The assumption is that SA languages hierarchise syllables into a metrical
word constituent whose most prominent syllable receives primary stress.
However, metrical properties not observing OBLHEAD should not be
confused with SA. Independent of its head-defining function, metrical
phonology has a well-known counting function used to locate or assign
phonological properties within a string. This is particularly clear in the
case of rhythmic alternations. To show that this counting or rhythmic
function need not be an expression of SA, consider the alternation of H
tones in Kirundi in (19).

(19) Alternation of H tones in Kirundi (Goldsmith & Sabimana 1986)

a. ku-sab-a -sab- ‘ask for (inf)’
ku-bi-sab-a -bi- ‘them’
ku-bí-mu-sáb-a -mu- ‘(to) him’
ku-bí-mu-kú-sab-ir-a -ku- ‘(for) you’
ku-há-bi-mú-ku-sáb-ir-a -ha- ‘there’
‘to ask him (for) them for you there’
b. ku-báz-a -báz- ‘ask (question) (inf)’
ku-bí-baz-a -bi- ‘them’
ku-bí-mu-báz-a -mu- ‘(to) him’
ku-bí-mu-kú-bar-iz-a -ku- ‘(for) you’
ku-há-bi-mú-ku-bár-iz-a -ha- ‘there’
‘to ask him (for) them for you there’
 Word-prosodic typology 243
As indicated, the verb root /-sab-/ is toneless, as are the infinitive prefix
/ku-/ and the final inflectional vowel /-a/. The form [ku-bi-sab-a] ‘to ask
for them ’ shows that when a single object marker (OM) is present, it too
is toneless. It would therefore appear that the /H/ of /-báz-/ has undergone
anticipatory shifting onto the OM in [ku-bı́-baz-a] ‘to ask them’. The full
pattern of alternating Hs is seen in forms with two or more OMs. From
comparative evidence we know that a second (third, fourth) OM carries
a /H/ tone. From the Kirundi point of view, however, all that can be said
is that combinations of OMs introduce a /H/. The generalisation is that
if there is at least one /H/, the span consisting of the OMs+verb root will
receive a H on every other mora. As seen, H tone is neither obligatory nor
culminative in Kirundi. Thus, while Goldsmith & Sabimana (1989) con-
struct binary s-w feet for the purpose of distributing the H tones, it should
be clear that this rhythmic pattern in Kirundi has nothing to do with the
definition of SA in (5).6
The above illustration raises the question of whether there is a necessary
relationship between metrical feet and SA. Two non-canonical prosodic
systems which have been covered in the stress literature (e.g. Prince 1983,
Halle & Vergnaud 1987, Hayes 1995) are Creek ‘ tonal accent’ and Seneca
‘accent ’, which I consider briefly below.
As described by Haas (1977), Creek /H/ obeys OBLHEAD but not cul-
minativity. Haas points out that ‘fixed’ accents may result in two H tones
in the same word, e.g. /náfka:kı́s/ ‘ they are hitting him’. Otherwise /H/
tones are mostly predictable, as Haas states in (20).

(20) Predictable tonal accent in Creek

a. ‘In any string of L[ight] syllables containing no fixed accents the
tonal accent will be placed on the last even-numbered syllable’:
/hicíta/ ‘one to see one’, /ahicitá/ ‘one to look after, watch one’ (Haas
1977: 202–203).
b. ‘If the penult is a H[eavy] syllable and the ultima a L[ight] one, the
tonal accent will be placed on the penult’: /cá:lo/ ‘trout, bass’, /sókca/
‘sack, bag’, /aLakkóycka/ ‘appreciation’ (Haas 1977: 203).
c. ‘However, if the ultima is a H[eavy] syllable, then the placement
of the tonal accent is no longer automatic. Instead morphological
rules come into play’: /hátki:/ ‘white’, /hoktí:/ ‘woman’ (Haas 1977:
204).

While the distribution is based on metrical principles, both the constant

H-tone realisation and the violation of culminativity make Creek ‘tonal
accent’ different from a prototypical SA system.

6 Re the possibility of ‘ tonal feet’, see Leben (2002).

244 Larry M. Hyman
The placement of alternating ‘accent’ in Seneca also follows metrical
principles. While Chafe (1977, 1996) presents an iambic analysis, the fol-
lowing trochaic account is based on Melinger’s (2002) reanalysis:

(21) Trochaic analysis of Seneca

a. Mark the first syllable extrametrical.
b. Build disyllabic trochees left-to-right.
c. Assign H tones (i) to the initial syllable of each trochee if closed;
(ii) to the initial syllable of each trochee if light, but followed by a
closed syllable.
As indicated, Melinger proposes to mark off the first syllable and build
trochees left to right. While she goes on to propose metrical structure
above the foot, there does not seem to be a need for a metrical word tree.
Instead, as I have indicated in (21c), H tones are assigned to the first
(head) syllable of each trochee that contains a closed syllable. As schema-
tised in (22), three out of the four possible combinations of open and
closed syllable trochees will result in an initial H tone, while one will result
is no H tone being assigned:

(22) Seneca violates OblHead (Prince 1983, Hayes 1995)

a. H on the first syllable b. no H
closed-closed [[Link]] open-open [[Link]]
closed-open [[Link]]
open-closed [[Link]]
Examples are given in (23), from Melinger (2002).

(23) a. <a> (ge ga)(jÊ? Oh) ‘I’m willing’

+OblHead, +culminativity
b. <de> (wá ge?)(njo da)(gÊ? Oh) ‘I’m busy’
+OblHead, —culminativity
c. <dE> (ga de)(nje o) dE? ‘I’ll put a necktie on’
—OblHead, —culminativity
Only the second foot in (23a) has a closed syllable (in fact, two), so the
single H assigned to this word satisfies both OBLHEAD and culminativity.
In (23b), both the first and third feet contain a closed syllable (the latter,
two). Two H tones are therefore assigned, in violation of culminativity.
The form in (23c) receives no H tone, because its footed syllables are all
open. (The last, left-over syllable does not get footed.) It therefore violates
both OBLHEAD and culminativity.
In his comprehensive and thoughtful survey of accentual phenomena,
van der Hulst (1999a: 94) states that ‘it remains mysterious º why
a weight-sensitive accent rule (which assigns accents to syllables rather
than moras) cannot produce accents that get a tonal interpretation ’. But
that appears to be what is going on in both Creek and Seneca. In Creek,
 Word-prosodic typology 245
the assignment of H is sensitive to syllable weight, and in Seneca it
is sensitive to open vs. closed syllables. Seneca is unusual not only in
targeting closed syllables, but especially in not caring whether the closed
syllable is initial or final in its trochee. It is as if Seneca distinguishes heavy
vs. light trochees (Wallace Chafe, personal communication).
Having considered Creek and Seneca, we are now in a position to
complete Table II.

+culminative —culminative
+obligatory Kinga Creek
—obligatory Somali Seneca

Table II
Restricted H-tone systems attest all four combinations of
obligatoriness and culminativity.

As indicated, prosodic systems which have been variously referred to

as ‘accent’, ‘pitch accent ’, ‘ tonal accent’ or ‘restricted tone ’ attest all
four combinations of obligatoriness and culminativity. Words have
exactly one H mora in Kinga. In Somali words may have at most one H
mora, but can also be toneless. Words must have one H syllable in Creek,
but may have more than one. Finally, in Seneca words can have one,
more than one or no syllable with H tone. While we could give a name to
each of these feature combinations, there is no reason to think they are
distinct ‘types ’. Rather, they simply share one vs. two vs. neither of the
properties required of SA but not T. In other words, the fact that we
obtain all four feature combinations shows that a hierarchical tree model
of prosodic types will not be possible, e.g. where the highest cut is accent
vs. tone, based, say, on culminative vs. non-culminative, and then only
culminative systems are further broken down into obligatory vs. non-
obligatory. We could call all four systems ‘accent’ or ‘pitch accent’, but
how would we capture the intuition that of the four, languages like Kinga,
Safwa and Nubi are the most SA-like ? My suggestion is that prosodic
typology address the properties of SA and T prototypes directly, rather
than ‘classifying’ systems by name.7

6 Phonetics
There has been a long-standing tension between typologising prosodic
systems by their linguistic function vs. their phonetic realisation (cf.
Hyman 1977b : 38–40). Phonologists have been mostly concerned with
7 Table II considers only languages which assign a single feature, here /H/. An ex-
pansion of the set would have to include systems where there are two (more ?) re-
stricted tones or tonal melodies, e.g. /H/ vs. /L/ in Fasu (May & Loeweke 1964), /H/
vs. /LH/ in Ma'ya (Remijsen 2001, 2002).
246 Larry M. Hyman
functional or structural properties of prosodic systems. For example,
Hayes identifies stress accent in terms of rhythm : ‘in stress languages,
every utterance has a rhythmic structure ’ (Hayes 1995 : 8). Non-stress
accent, on the other hand, is often identified in terms of culminativity:
‘ a word may contain several accented morphemes, but only one accent
may surface in a word domain’ (Hualde 1999: 962). The problems dis-
cussed in preceding sections have had to do with cases where such notions
run into difficulty : not all rhythmic phenomena indicate SA, not all
so-called PA languages respect culminativity, etc. We have claimed that
the unique defining feature of SA is OBLHEAD, while the unique defin-
ing feature of T is ‘ an indication of pitch’ at the output of the lexical
phonology.
Lost in much of this is the role of phonetics. In (12) I presented some of
the views on how SA and PA might be distinguished on the basis of the
phonetic realisations : like T, the realisation of PA is primarily pitch,
whereas the realisation of SA may involve different phonetic features.
Attempts to provide a phonetic characterisation of prosodic differences are
not contradictory, but rather should follow from functional definitions.
T is analysed in terms of phonological features for which one should ex-
pect relatively constant phonetic analogues. SA, on the other hand, is an
abstract or underlying mental construct (Weinreich 1954: 2, Lehiste
1970 : 150). It should therefore not be surprising if languages and speakers
vary in how they realise abstract SA in the output. As instrumental studies
have shown, the phonetic correlates of stress can be quite subtle – even
absent on syllables that one ‘feels’ are really stressed. Given this, the
functional basis of stress in actual speech may have less to do with aiding
the listener to segment words (as is often proposed) than with helping the
speaker organise phonetic material for his own purposes.
The above problem becomes particularly acute in cases where there is
a question about whether a language has stress at all. Researchers who
operate under the assumption that all languages have SA may read stress
into the phonetic variations they hear or observe instrumentally. Since
many tone systems are adequately described with no mention of SA, I am
less inclined to posit SA in non-tonal languages which make it so hard to
find evidence of stress, e.g. Bella Coola (§4), some variants of Berber and
a number of languages of Ethiopia and the Indian subcontinent (Hyman
1977b). In other words, we might say: if word-stress is so hard to find,
perhaps it is not there at all.8
On the other hand, some linguists seem to apply the flip side of this
motto : if it sounds like stress, it must be stress! The result has been to
posit stress features that are quite different from what we normally think

8 An interesting case pointed out by the guest editors concerns Indonesian. Whereas
Cohn (1989) reported weight-sensitive, word-level stress, subsequent acoustic stud-
ies and experiments by van Zanten & van Heuven (1998) and van Zanten et al.
(2003) have shown that the language only has intonational ‘ pitch accents ’, realised
near the right edge of the phrase.
 Word-prosodic typology 247
of as SA. Kähler-Meyer (1962 : 257–258) identifies a ‘Dehnungsakzent ’,
which lengthens a penultimate vowel in Chasu (Kipare), as in many
other Bantu languages (§5.2). The lengthened penult also receives the
‘Druckakzent ’ (stress), with one exception : as seen in (24a), a phrase-final
L that derives from the /H/ of a /L-H/ sequence instead receives the
stress :

(24) Alleged stressed vs. non-stressed final L in Chasu

a. /mì-tí/ £ [mì:'tì] ‘trees’ (cf. [mìtí 'mí:Ωgì] ‘many trees’)
b. /mì-sì/ £ ['mì:sì] ‘pestles’

As seen, the two ‘Akzente’ do not line up in (24a), whereas they do in

(24b), where the final L from /L/ is not stressed. Whatever the difference
is between final 'L and L (which Kähler-Meyer says is hard to hear), it is
likely to be a slight difference in the F0 contour, as occurs in Chaga, Rimi,
Bemba and elsewhere in Bantu (Hyman 1977a : 16–17). Thus, treating
final 'L in terms of stress rather than tone is suspect. It also brings us close
to the situation rejected in §5.2, wherein a language would have two dif-
ferent obligatory word-accent systems, here QA and SA.9
In other cases, claims of two independent SA systems are also dubious.
In the Mixtec dialects/languages stems are generally disyllabic (‘ coup-
lets ’), with stress on their first syllable. Such stem-initial stress often
conditions lengthening of the first vowel or second consonant. Certain
dialects, however, have been claimed to have potentially two stresses per
stem or word. According to Pike & Oram (1976), a second unpredictable
stress occurs on the second syllable of some, but not all, stems in Diuxi
Mixtec – specifically on those for which a final *? has been reconstructed.
Just as final 'L in Chasu is suspect, the alleged stem-final stresses in Diuxi
have been reanalysed in terms of tone (Daly 1978).
Another potential case of multiple stresses arises in Ayutla Mixtec.
Pankratz & Pike (1967) analyse Ayutla dialect as having tone-dependent
stress, as in (25) (wording from Mortensen 2005).

(25) Tone-dependent stress in Ayutla Mixtec

Within the stem+sux domain, stress the leftmost H-toned syllable
immediately followed by the L-toned syllable, else stress the leftmost
M-toned syllable immediately followed by a L-toned syllable, else
stress the leftmost H-toned syllable, else stress the leftmost syllable.
However, as Mortensen (2005 : 2) points out, ‘ the couplet-initial promi-
nence seems to be a much better candidate for stress than pitch prominence
(which isn’t present in some [monosyllabic] words and is confounded
by the ‘ allotonic’ variations in pitch described by [Pankratz & Pike]’.

9 The prohibition in question concerns two obligatory word-level accentual systems.

Kähler-Meyer’s accentual indications suggest that both penultimate lengthening
and final H £ L take place phrase-finally.
248 Larry M. Hyman
Assuming that the stem-initial syllable is stressed would account for the
preaspiration of voiceless stops and the lengthening of sonorants when
occurring in the second consonant position of CVCV stems. It also would
explain why /k/, /”/ and /v/ undergo optional weakening in post-stem, but
not stem-internal position. The tone-dependent stress algorithm in (25)
often results in stress coinciding with the first stem syllable. Where it
doesn’t (e.g. L-H stems, L-L stems+H suffix), it is not clear that stress
is involved at all. What would be convincing would be if the second-
consonant effects were absent in the case of L-'H stems, and if the suffix C
were instead preaspirated or lengthened if L-'H stems were followed by a
L suffix. If we were to posit both Pankratz & Pike’s tone-dependent stress
and stem-initial stress, we would still have to choose between two inter-
pretations. The first – and undesirable – one would be that a word could
have two stresses, i.e. 'L-'H, 'L-L+'H. The second is that stress is initially
assigned stem-initially, where it conditions the post-stress consonant
effects, but is subsequently attracted away when the most prominent
tone is found later in the stem+suffix domain.10 As Mortensen suggests, it
seems more likely that Pankratz & Pike’s tone-dependent stress is simply
phonetic pitch prominence.11
The logical endpoint of a system claimed to have multiple stress would
be one where stress fulfils the same distinctive function as segmental and
tonal features. Gilley (2004 : 100) has recently claimed that ‘ stress in
Nilotic is privative, a feature with binary opposition. º a word may be
stressed or not. The stress in itself may function as a morpheme in some
10 If the disyllabic stem (couplet)=a foot, preaspiration and lengthening could be
interpreted as strictly foot-based processes. Tone-dependent stress, on other hand,
is assigned within the stem+suffix domain. A general issue not addressed in the
present study is how to interpret positionally restricted distributions and realisa-
tions. For example, in a number of Northwest Bantu, Bantoid and Cross-River
languages, voiceless stops are found only stem-initially (Hyman 1990, in press), e.g.
Koyo /ð-t¸tÐ/, /è-kúkı́/ £ [ð-t¸rÐ] ‘ basket ’, [è-kúgı́] ‘ door ’. While such phenomena
are occasionally cited as evidence for stem-initial accent, it is significant that re-
duplications typically produce exceptions (e.g. [ð-t¸t¸] ‘ spark ’, [ı̀-kákànı́] ‘ legend ’).
Whether one treats these latter as having two stems with initial consonant corre-
spondence or by some other means, if voicelessness=accent, then these words
would have two (demarcative) accents. SA can have such effects, but I do not take
the presence of such effects to unambiguously indicate accent.
11 There have been a number of recent instrumental investigations into the realisation
of stress in tone languages. Remijsen & van Heuven (2005), for example, study both
the phonological properties and phonetic realisation of stress in Curaçao
Papiamentu, which also has an independent privative tone contrast. On the other
hand, earlier reports of non-restricted tone systems with tone-dependent stress are
often less satisfying, either describing the phonetic effects of stress in impression-
istic terms, or, as in the case of the alleged SA system in Usarufa summarised in
(14b), not indicating the effects or realisation of stress at all. What is typically
missing is phonological evidence. It would be impressive, for example, if the alleged
stressed syllable of Usarufa provided the reference point for building iterative
stress-feet, or if it constituted the locus for categorical segmental rules, e.g. a
stressed onset stop is aspirated, a post-stress consonant is geminated, a stressed mid
vowel is diphthongised, etc. While I have long been sceptical, if SA can be assigned
as claimed in Usarufa, Ayutla Mixtec and many more non-restricted tone systems,
such phonological evidence should be forthcoming.
 Word-prosodic typology 249
cases and stress may occur on any of the three level tones’. Examples from
Shilluk are seen in (26).

(26) Alleged distinctive stress in Shilluk (Gilley 2004: 113)

singular plural
a. 'dít diÕt ‘bird’
'¤òòk ¤óòk ‘mouth’
b. álठá'l£¤ ‘cloth’
pÀÃm 'pÃÀm ‘chest’

In (26a) singulars are said to be stressed, while in (26b) the plurals are
stressed. As seen in the first example of each set, the presence vs. absence
of stress may also be accompanied by segmental changes. There are,
however, at least four reasons why this is not stress in the usual sense.
First, a word can lack stress in violation of OBLHEAD. Second, a word can
have more than one stress, in violation of culminativity, e.g. /á-'cámmA-'á/
([Link]-eat AGGRESSIVE [Link] ACTION). Third, the
phonetic effect on vowel length is the opposite of what one expects :
‘phonetically, stress in these Nilotic languages is characterized by
a shortening of vowel length, and by an increased intensity or level of
force on the syllable ’ (Gilley 2004: 117). Finally, native speakers consider
the ‘unstressed’ syllables to be the marked member of the opposition.
This correlates with the apparent historical origin of the opposition :
‘suffixes were lost, but the timing unit of the suffix remained and was
incorporated into the root. º This analysis concludes that the normal
unmarked form is the shorter or “stressed ” word’ (Gilley 2004: 111). As
in the case of the alleged ‘multiple stresses ’ of Campa (Pike & Kindberg
1956, Pike 1974), the distinctive syllable properties in question should
not be identified with SA.
Most of the discussion in this section has concerned claims of stress
in languages which have unrestricted tone systems. As expressed in the
binary typological distinction between SA and T in (1a), it was once
thought that stress and tone were two mutually exclusive prosodic sys-
tems : since F0 is an important phonetic factor in the realisation of SA, it
seemed reasonable to assume that a language in which tonal oppositions
are realised solely in terms of F0 would not be able to support an inde-
pendent SA system. While we now know that this is false, the question
still remains as to whether SA has access to the same range of phonetic
exponents in languages with vs. without T. In fact, there is some reason to
think that SA may have different properties when co-occurring with T.
A case in point is Pirah± (Everett 1998), which, based on the hierarchy in
(27), assigns stress to the heaviest syllable within a final three-syllable
window (C=voiceless; G=voiced):
(27) Unusual syllable weight hierarchy in Pirahã
CVV >> GVV >> VV >> CV >> GV
250 Larry M. Hyman
As expected, long vowels are heavier than short vowels. Not expected,
however, is that onsets contribute to syllable weight – and that voiceless
onsets attract stress over voiced onsets. Reminiscent of the Shilluk data
in (26), Everett (1998: 114) finds that ‘the vowels of unstressed syllables
have a significantly greater duration than the vowels of stressed syllables ’.
However, this may be a compensatory adjustment to the fact that onsets
are significantly lengthened under stress :

The primary acoustic correlates of stress in Pirah± are amplitude and

duration. The vowels of stressed syllables have significantly greater am-
plitude. In addition, stressed syllables have a significantly greater
duration. The greater duration of stressed syllables is the result of
the greater duration of the onsets of stressed syllables. Finally, the
F0 and formant frequencies of F1 and F2 were not significantly
different in the vowels of the stressed and unstressed syllables (Everett
1998 : 121).

These findings are important because they suggest that the phonetic re-
alisation of SA may indeed be partly determined by the phonological in-
ventory of a language. Pirah± has T and contrasting vowel length, placing
a damper on both F0 and vowel duration as ideal phonetic cues of SA.
Instead, onsets not only attract stress, but also undergo significant modi-
fication, e.g. increase in length, when occurring under stress. It thus seems
that an alternative to resisting SA in a language with T is to incorporate
SA in a way that minimises interference with the relevant tone (and
length) contrasts.12

7 Conclusion
This paper has dealt with various aspects of word-prosodic typology.
I began with a quotation from Greenberg & Kaschube (1976 : 9). Stress,
they said, is ‘ monarchic’: a monarchy must have one and only one king
or queen. This corresponds to OBLHEAD, a definitional requirement
placed on SA. At the other end, tone is ‘democratic’: in a non-restricted
tone system, there can be more than one H tone per word. Although I have
not treated length separately in this study, Greenberg & Kaschube give
it the intermediate characterisation ‘oligarchic’, as (vowel) length can
be culminative, restricted to one occurrence per word (cf. Mam in §5.2).
Their – and my – characterisations have focused on structural and
functional aspects of word prosody. As implied in §6, it is also possible
to typologise on the basis of ‘substance ’, e.g. how is SA realised in dif-
ferent languages ? What role does F0 play in different languages ? The
variety and overlap of phonetic realisations no doubt contribute to the

12 For more discussion of languages with onset-sensitive stress placement, see Gordon
(2005).
 Word-prosodic typology 251
difficulties characterising the study of prosody. As Odden (1999: 189)
puts it :

Modern phonological research is strongly grounded in the idea that the

categories of phonological analysis are phonetically based, and therefore
if a language is analyzed as making active use of a given feature, it
is assumed that this feature has a certain element of phonetic reality.
Thus, for example, the opposition between [u] and [
] would be ac-
counted for in terms of rounding, and never nasalization, since linguists
have shared assumptions about the phonetic meaning of ‘rounding ’
versus ‘nasalization ’. The question is whether there is ever any hope of
being able to hear a difference between tone and stress : the answer
seems to be that there is not.

A major problem, therefore, is that one cannot go directly from the pho-
netic signal to the phonological analysis. By following well-established
methods, linguists will in most cases be ultimately convinced as to
whether a surface contrast should be characterised in terms of F0 or
something else. However, this is only the beginning : we must now
characterise the system underlying the surface contrast. Should Odden’s
[u] vs. [
] be analysed as /u/ vs. /
/, /u/ vs. /V/ or /
/W vs. /
/ ? In other words,
does the phonetic difference in rounding correspond to an opposition
which is binary ([+round]), privative (Round vs. 0), or prosodic (W vs. 0) ?
Similarly, should a [H-L] sequence be analysed as /H-L/, /H-0/ or
/0-0/HL ? Or should it be analysed as (s-w), (*-¡) etc. ? Maybe there are
more possibilities and more ambiguities in the area of prosodic systems,
but, to my knowledge, this has not been demonstrated. Instead, I contend
that the lack of clarity and the controversy that seem sometimes evident in
the identification of word-prosodic systems can be overcome by applying
the same methods of analysis and argumentation to prosody that routinely
apply to segmental phonology.
The history of word-prosodic typology has been characterised by
vagueness and disagreement, and yet great conviction on the part of dif-
ferent scholars. In this study I have not cited every past proposal nor
considered every logical possibility. I have proposed that the most sig-
nificant typological cut concerns prosodic systems which do vs. those
which do not satisfy the OBLHEAD property. Aside from a few ambiguous
systems such as Kinga and Nubi, this separates SA systems from the rest
(whether they co-occur with tone or not). A reviewer has, however,
pointed out that a different result would obtain if the culminative property
provided the first cut. Accent systems would be those which meet the
culminative requirement, further differentiated into those which satisfy
OBLHEAD (SA) vs. those which do not (PA). The result is the tripartite
typology (1c) which I rejected : SA and PA satisfy culminativity while PA
and T satisfy the tonal definition given in (3). Under this scheme, as the
reviewer further points out, Kinga and Safwa would have SA, despite the
fact that they count moras rather than syllables in assigning H tones.
252 Larry M. Hyman
Tokyo Japanese would have PA, while Creek and Seneca become non-
accentual tone systems even though their H tones are assigned by metrical
principles. There are of course many more ways to taxonomise prosodic
systems, some of which are not entirely counterintuitive (e.g. by whether
prosodic oppositions are unary, binary or involve multiple features,
whether they are fixed vs. free, and so forth). However, I would argue
that the ‘deepest ’ typological cuts should be based on properties which
make the strongest predictions about the systems in question. As was seen
in §4, the OBLHEAD property is a very restricted one, in that it does not
apply to features, rather to structure. It predicts that stressless major
category words will never occur in a SA language, which also will never
require that a second phonological property occur on all words (e.g.
H tone, vowel length, glottalisation, aspiration). It also turns out that all
SA systems assign prosody to syllables rather than moras. However, in
§5.2 we saw that the notion of culminativity has not been restricted to a
culmination of structure, but has in addition been applied to various
prosodic and non-prosodic features. Under the reviewer’s proposal,
Cuzco Quechua, which allows only one glottalised or aspirated consonant
per word, presumably has a ‘glottal accent’. Consider also the Bantu
language Punu, which has the five underlying vowels /i E u O a/, but re-
stricts /E O/ to the first root syllable (Fontaney 1980, Kwenzi Mikala 1980).
Since a word has at most one /E/ or /O/, mid vowel height has the culmi-
native property. However, would we want to say that Punu has a ‘mid
accent ’? We avoid all these problems if we start with the OBLHEAD
property : Punu does not require every word to have a mid vowel, any
more than Cuzco Quechua requires every word to have a glottalised or
aspirated consonant. Such systems, I would claim, do not exist. Concern-
ing word-prosodic systems, the primary typological questions are whether
a prosodic system has tone (‘an indication of pitch enters into the lexical
realisation of at least some morphemes’) and/or stress accent (‘every
lexical word has at least one syllable marked for the highest degree of
metrical prominence’).
In both segmental and prosodic phonology there is a paradox concern-
ing the role of theory. On the one hand theory is absolutely essential in
doing typology – which, recall from §2, does not simply aim to establish
inventories that vary across languages. On the other hand, we need to
extract away from the particulars of theories and ‘normalise ’ the data to be
sure that our typologies are based on the properties of prosodic systems
rather than the interpretations given to them by different analysts – or by
even the same analyst on different occasions. Finally, theory/typology
should be based on as complete a range of what is (potentially) attested in
languages as possible. As stated by Hayes (1995 : 403) : ‘Phonological
studies, which have access to perhaps the greatest breadth of language
data, have shown that individual languages may develop an amazing
variety of rich and idiosyncratic formal systems. Only broad typological
work can establish which traits are properties of Language, and which are
properties of individual languages’. Lehiste (1970: i) once observed that
 Word-prosodic typology 253
‘a certain degree of vagueness seems to characterize most discussions of
prosodic features ’. We have learned a lot since 1970, and although I hope
that the above proposals will help clarify some of the complexities facing
word-prosodic typology, there is much more to discover !

REFERENCES

Bagemihl, Bruce (1991). Syllable structure in Bella Coola. LI 22. 589–646.

Banti, Giorgio (1988). Two Cushitic systems: Somali and Oromo nouns. In van der
Hulst & Smith (1988b). 11–49.
Beckman, Mary E. (1986). Stress and non-stress accent. Dordrecht : Foris.
Bee, Darlene & Kathleen Barker Glasgow (1962). Usarufa tone and segmental pho-
nology. In James C. Dean (ed.) Studies in New Guinea linguistics. Sydney:
University of Sydney. 111–127.
Bybee, Joan, Paromita Chakraborti, Dagmar Jung & Joanne Scheibman (1998).
Prosody and segmental effect: some paths of evolution for word stress. Studies in
Language 22. 267–314.
Cassimjee, Farida & Charles W. Kisseberth (1989). Shingazidja nominal accent.
Studies in the Linguistic Sciences 19. 33–61.
Cassimjee, Farida & Charles W. Kisseberth (1998). Optimality Domains Theory and
Bantu tonology : a case study from Isixhosa and Shingazidja. In Larry M. Hyman &
Charles W. Kisseberth (eds.) Theoretical aspects of Bantu tone. Stanford : CSLI.
33–132.
Chafe, Wallace L. (1977). Accent and related phenomena in the Five Nations Iroquois
languages. In Hyman (1977c). 169–181.
Chafe, Wallace L. (1996). Sketch of Seneca, an Iroquoian language. In Ives Goddard
(ed.) Languages (Vol. 17 of Handbook of North American Indians). Washington,
D.C. : Smithsonian Institution. 551–579.
Clements, G. N. & John Goldsmith (1984a). Autosegmental studies in Bantu tone:
introduction. In Clements & Goldsmith (1984b). 1–17.
Clements, G. N. & John Goldsmith (eds.) (1984b). Autosegmental studies in Bantu
tone. Dordrecht : Foris.
Cohn, Abigail (1989). Stress in Indonesian and bracketing paradoxes. NLLT 7.
167–216.
Daly, John P. (1978). Notes on Diuxi Mixtec tone. Work Papers of the Summer
Institute of Linguistics, University of North Dakota 22. 98–113.
Dobrovolsky, Michael (1999). The phonetics of Chuvash stress: implications for
phonology. In John J. Ohala, Yoko Hasegawa, Manjari Ohala, Daniel Granville &
Ashlee C. Bailey (eds.) Proceedings of the 14th International Congress of Phonetic
Sciences. Vol. 1. Berkeley : University of California. 539–542.
Donohue, Mark (1997). Tone in New Guinea. Linguistic Typology 1. 347–386.
Donohue, Mark (1999). A grammar of Tukang Besi. Berlin & New York: Mouton de
Gruyter.
Donohue, Mark (2003). The tonal system of Skou, New Guinea. In Shigeki Kaji (ed.)
Cross-linguistic studies of tonal phenomena: historical development, phonetics of tone,
and descriptive studies. Tokyo : ILCAA. 329–364.
Downing, Laura J. (2004). What African languages tell us about accent typology. ZAS
Papers in Linguistics 37. 101–136.
England, Nora C. (1983). A grammar of Mam, a Mayan language. Austin : University
of Texas Press.
Everett, Keren Madora (1998). The acoustic correlates of stress in Pirah±. Journal of
Amazonian Languages 1:2. 104–162.
Fontaney, Louise (1980). Le verbe. In Nsuka-Nkutsi (1980). 51–114.
254 Larry M. Hyman
Fox, Anthony (2000). Prosodic features and prosodic structure. Oxford: Oxford
University Press.
Garde, Paul (1968). L’accent. Paris: Presses Universitaires de France.
Gilley, Leoma (2004). The feature of stress in Nilotic. In Rose-Juliet
Anyanwu (ed.) Stress and tone : the African experience. Cologne : Rüdiger Köppe.
99–120.
Goldsmith, John (1984). Tone and accent in Tonga. In Clements & Goldsmith
(1984b). 19–51.
Goldsmith, John (1987). Tone and accent, and getting the two together. BLS 13.
88–104.
Goldsmith, John & Firmard Sabimana (1989). The Kirundi verb. In Francis
Jouannet (ed.) Modèles en tonologie (kirundi et kinyarwanda). Paris : Éditions du
CNRS. 19–62.
Good, Jeff (2004). Tone and accent in Saramaccan : charting a deep split in the pho-
nology of a language. Lingua 114. 575–619.
Gordon, Matthew (2005). A perceptually-driven account of onset-sensitive stress.
NLLT 23. 595–653.
Greenberg, Joseph (1974). Language typology: a historical and analytic overview. The
Hague & Paris : Mouton.
Greenberg, Joseph & Dorothea Kaschube (1976). Word prosodic systems: a prelimi-
nary report. Working Papers on Language Universals 20. 1–18.
Gussenhoven, Carlos (2004). The phonology of tone and intonation. Cambridge :
Cambridge University Press.
Gussenhoven, Carlos (2006). The prosody of Nubi between stress and tone. Phonology
23. 193–223 (this issue).
Haas, Mary R. (1977). Tonal accent in Creek. In Hyman (1977c). 195–208.
Hagège, Claude (1992). Morphological typology. In William Bright (ed.) International
encyclopedia of linguistics. Vol. 3. New York & Oxford : Oxford University Press.
7–8.
Halle, Morris & Jean-Roger Vergnaud (1987). An essay on stress. Cambridge, Mass. :
MIT Press.
Haraguchi, Shosuke (1977). The tone pattern of Japanese : an autosegmental theory of
tonology. Tokyo : Kaitakusha.
Hayes, Bruce (1995). Metrical stress theory: principles and case studies. Chicago :
University of Chicago Press.
Hualde, José, Ignacio (1991). Basque phonology. London & New York : Routledge.
Hualde, José Ignacio (1999). Basque accentuation. In van der Hulst (1999b). 947–993.
Hulst, Harry van der (1999a). Word accent. In van der Hulst (1999b). 3–115.
Hulst, Harry van der (ed.) (1999b). Word prosodic systems in the languages of Europe.
Berlin & New York: Mouton de Gruyter.
Hulst, Harry van der & Norval Smith (1988a). The variety of pitch accent systems :
introduction. In van der Hulst & Smith (1988b). ix–xxiv.
Hulst, Harry van der & Norval Smith (eds.) (1988b). Autosegmental studies on pitch
accent. Dordrecht : Foris.
Hyman, Larry M. (1977a). Tone and/or accent. In D. J. Napoli (ed.) Elements of tone,
stress and intonation. Washington : Georgetown University Press. 1–20.
Hyman, Larry M. (1977b). On the nature of linguistic stress. In Hyman (1997c).
37–82.
Hyman, Larry M. (ed.) (1977c). Studies in stress and accent. Los Angeles: Department
of Linguistics, University of Southern California.
Hyman, Larry M. (1981). Tonal accent in Somali. Studies in African Linguistics 12.
169–203.
Hyman, Larry M. (1985). A theory of phonological weight. Dordrecht: Foris. Reprinted
2003 with an introduction, Stanford : CSLI.
 Word-prosodic typology 255
Hyman, Larry M. (1989). Accent in Bantu : an appraisal. Studies in the Linguistic
Sciences 19. 115–134.
Hyman, Larry M. (1990). Non-exhaustive syllabification : evidence from Nigeria and
Cameroon. CLS 26:2. 175–195.
Hyman, Larry M. (2001a). Tone systems. In Martin Haspelmath, Ekkehard König,
Wulf Oesterreicher & Wolfgang Raible (eds.) Language typology and language
universals : an international handbook. Vol. 2. Berlin & New York : de Gruyter.
1367–1380.
Hyman, Larry M. (2001b). Privative tone in Bantu. In Shigeki Kaji (ed.) Cross-
linguistic studies of tonal phenomena : tonogenesis, Japanese accentology, and other
topics. Tokyo : ILCAA. 237–257.
Hyman, Larry M. (in press). Directional asymmetries in the morphology and pho-
nology of words, with special reference to Bantu. Linguistics 45.
Johnson, Keith (2005). Tone and pitch accent in Cherokee nouns. Available
(April 2006) at [Link]
[Link].
Kaji, Shigeki (ed.) (2005). Cross-linguistic studies of tonal phenomena : historical devel-
opment, tone–syntax interface, and descriptive studies. Tokyo : ILCAA.
Kähler-Meyer, E. (1962). Studien zur tonalen Struktur der Bantusprachen. II : Chasu.
Afrika und Übersee 46. 250–295.
Kisseberth, Charles (2005). Accent and phrasing in Chimwiini. In Kaji (2005).
129–145.
Kisseberth, Charles & Mohammad Imam Abasheikh (1974). Vowel length in
Chi-Mwi:ni : a case study of the role of grammar in phonology. In Anthony Bruck,
Robert A. Fox & Michael W. La Galy (eds.) Papers from the parasession on natural
phonology. Chicago : Chicago Linguistic Society. 193–209.
Kisseberth, Charles & David Odden (2003). Tone. In Derek Nurse &
Gérard Philippson (eds.) The Bantu languages. London & New York : Routledge.
59–70.
Kwenzi Mikala, Jérôme (1980). Esquisse phonologique du punu. In Nsuka-Nkutsi
(1980). 7–18.
Ladd, D. Robert (1996). Intonational phonology. Cambridge : Cambridge University
Press.
Leben, William R. (2002). Tonal feet. In Ulrike Gut & Dafydd Gibbon (eds.)
Typology of African prosodic systems. Bielefeld : University of Bielefeld. 27–40.
Lehiste, Ilse (1970). Suprasegmentals. Cambridge, Mass. : MIT Press.
McCarthy, John J. (2002). A thematic guide to Optimality Theory. Cambridge :
Cambridge University Press.
McCarthy, John J. (2003). OT constraints are categorical. Phonology 20. 75–138.
McCawley, James D. (1970). Some tonal systems that come close to being pitch accent
systems but don’t quite make it. CLS 6. 526–531.
McCawley, James D. (1978). What is a tone language ? In Victoria Fromkin (ed.)
Tone : a linguistic survey. New York: Academic Press. 113–131.
Martinet, André (1960). Eléments de linguistique générale. Paris : Librairie Armand
Colin.
May, Jean & Eunice Loeweke (1964). The phonological hierarchy in Fasu.
Anthropological Linguistics 7:5. 89–97.
Mazaudon, Martine (2005). On tone in Tamang and neighbouring languages :
synchrony and diachrony. In Kaji (2005). 79–96.
Meeussen, A. E. (1963). Morphotonology of the Tonga verb. Journal of African
Languages 2. 72–92.
Meeussen, A. E. (1972). Japanese accentuation as a restricted tone system. Papers in
Japanese Linguistics 1. 267–270.
Melinger, Alissa (2002). Foot structure and accent in Seneca. IJAL 68. 287–315.
256 Larry M. Hyman
Mortensen, David (2005). Ayutla Mixtec tone. Handout from seminar presentation,
University of California, Berkeley.
Mous, Maarten (1993). A grammar of Iraqw. Hamburg : Buske.
Newman, Stanley (1947). Bella Coola I : phonology. IJAL 13. 129–134.
Nsuka-Nkutsi, François (ed.) (1980). Éléments de description du punu. Lyon : CRLS,
Université Lyon II.
Odden, David (1988). Predictable tone systems in Bantu. In van der Hulst & Smith
(1988). 225–251.
Odden, David (1999). Typological issues in tone and stress in Bantu. In Shigeki Kaji
(ed.) Cross-linguistic studies of tonal phenomena : tonogenesis, typology, and related
topics. Tokyo : ILCAA. 187–215.
Pankratz, Leo & Eunice V. Pike (1967). Phonology and morphotonemics of Ayutla
Mixtec. IJAL 33. 287–299.
Parker, Steve (1977). An OT account of laryngealization in Cuzco Quechua.
Work Papers of the Summer Institute of Linguistics, University of North Dakota 41.
1–11.
Pierrehumbert, Janet B. & Mary E. Beckman (1988). Japanese tone structure.
Cambridge, Mass. : MIT Press.
Pike, Eunice V. (1974). A multiple stress system versus a tone system. IJAL 40.
169–175.
Pike, Eunice V. & Joy Oram (1976). Stress and tone in the phonology of Diuxi Mixtec.
Phonetica 33. 321–333.
Pike, Kenneth L. & Willard Kindberg (1956). A problem in multiple stresses. Word
12. 415–428.
Poser, William J. (1984). The phonetics and phonology of tone and intonation in Japanese.
PhD dissertation, MIT.
Prince, Alan (1983). Relating to the grid. LI 14. 19–100.
Pulleyblank, Douglas (1986). Tone in Lexical Phonology. Dordrecht : Reidel.
Remijsen, Bert (2001). Word-prosodic systems of Raja Ampat languages. PhD disser-
tation, University of Leiden.
Remijsen, Bert (2002). Lexically contrastive stress accent and lexical tone in Ma’ya. In
Carlos Gussenhoven & Natasha Warner (eds.) Laboratory phonology 7. Berlin &
New York: Mouton de Gruyter. 585–614.
Remijsen, Bert & Vincent J. van Heuven (2005). Stress, tone and discourse promi-
nence in the Curaçao dialect of Papiamentu. Phonology 22. 205–235.
Saeed, John (1999). Somali. Amsterdam : Benjamins.
Salmons, Joe (1992). Accentual change and language contact : comparative study
and a case study of early northern Europe. Stanford : Stanford University Press.
Schadeberg, Thilo C. (1973). Kinga : a restricted tone system. Studies in African
Linguistics 4. 23–48.
Vajda, Edward (2001). Materials for Linguistics 201, Western Washington University.
Available (May 2006) at [Link]
[Link].
Voorhoeve, Jan (1973). Safwa as restricted tone system. Studies in African Linguistics
4. 1–21.
Weinreich, Uriel (1954). Stress and word structure in Yiddish. In Uriel Weinreich
(ed.) The field of Yiddish : studies in Yiddish language, folklore, and literature.
New York: Linguistic Circle of New York. 1–27.
Welmers, William E. (1959). Tonemics, morphotonemics, and tonal morphemes.
General Linguistics 4. 1–9.
Welmers, William E. (1973). African language structures. Berkeley : University of
California Press.
Willard, E. Rainbow (2004). Dominance effects in a dialect of Mam Maya. Ms,
University of California, Berkeley.
 Word-prosodic typology 257
Zanten, Ellen van, Rob Goedemans & Jos Pacilly (2003). The status of word stress in
Indonesian. In Jeroen van de Weijer, Vincent J. van Heuven & Harry van der Hulst
(eds.) The phonological spectrum. Vol. 2: Suprasegmental structure. Amsterdam &
Philadelphia : Benjamins. 151–175.
Zanten, Ellen van & Vincent J. van Heuven (1998). Word stress in Indonesian : its
communicative relevance. Bijdragen tot de Taal-, Land- en Volkenkunde 154.
129–149.
Zec, Draga (1999). Footed tones and tonal feet : rhythmic constituency in a pitch-
accent language. Phonology 16. 225–264.