A Theory of Consonantal Interaction*
Tobias Scheer
Abstract
Co-occurrence restrictions on word-initial consonant clusters are traditionally viewed as a
consequence of the relative sonority of both members of the CC. In the first part of this paper, I aim
to show that the reasoning underlying this approach is circular. The observation that sonority does increase
in word-initial clusters is relabelled explanation in saying that sonority must increase. Since
the crucial part of this circular argumentation is expressed by a constraint (“sonority must increase
within word-initial clusters”), I address the more general issue of constraints in linguistic theory.
In the second part of the paper, I propose a constraint-free theory where restrictions on word-initial
clusters follow from the interaction of more general principles. The main principles I draw on are
Government-Licensing (Charette, 1990), segmental complexity (Harris, 1990) and a strict CVCV
syllable-structure (Lowenstamm, 1996). None of these devices makes special reference to wordinitial
clusters. Since word-initial restrictions crucially depend on idiosyncratic properties of the
consonants involved, I also investigate the internal structure of consonants. In the representations I
introduce, the set of observations commonly subsumed under the label sonority is assigned no
phonological status. Rather, it is shown to be a function of known phonological primitives. Finally, a
theory of consonantal interaction built on the consonantal identities developed and the principles
mentioned is presented. This theory cannot possibly deal with the reverse phenomenology, predicting
that word-initial RT-clusters (where 'T' = obstruent, 'R' = sonorant) are impossible in languages of the
Indo-European (IE) type.
In §1, the traditional way of handling word-initial clusters is examined. §2 investigates the
theoretical status of constraints. The way to proceed in order to build a constraint-free theory is
outlined in §3 and §4. In §5, consonantal identities with in-built sonority are developed. §6 is
concerned with the paradigmatic/segmental aspect of word-initial restrictions on clusters. The
relevant technical devices used are discussed in §7, that is the phonological ECP and the CVCVmodel.
Finally, §8 addresses the syntagmatic aspect of the restrictions under focus.
1. Word-initial Consonant Clusters and Circularity
If a language exhibits word-initial consonant clusters, either there are no restrictions
on the possible combinations (like in certain Afro-Asiatic languages), or only #TRclusters
exist (like in IE languages). There is no language where only RT-clusters
occur word-initially. The usual way of accounting for the exclusion of #RT clusters
might be summarized as follows:1
1) Words cannot begin with a Coda. Thus, the
context “word-initial” corresponds to “Onset” on the
Text and pagination of this file are identical to the following reference:
Scheer, Tobias 1999. A Theory of Consonantal Interaction. Folia Linguistica
32, 201-237.
202
syllabic level. 2) In languages of the IE type, consonant clusters are not free wordinitially,
but both …TR… and …RT… occur word-internally. This distribution
matches that of syllabic constituents: “only Onsets in #__” vs. “both Onsets and
Codas word-internally”. Thus, syllabic structure is responsible for the observed
restrictions. 3) The sonority value for each segment can be established
independently. Word-initially, i.e. within a branching Onset, sonority must increase.
4) #RT clusters do not exist because their sonority is falling. Hence, they cannot
hold within a branching Onset. They cannot be interpreted as a Coda-Onset
sequence either because there are no word-initial Codas. Moreover, in typical
syllabification algorithms of the Kahnian kind that rely on the maximal clusterapproach,
the set of existing #CCs defines the very set of possible branching Onsets
for the whole language: “a possible branching Onset are all and only the CCs found
in the context # __” (cf. Kahn, 1976; Lowenstamm, 1981).
This approach is circular: It puts the word because between two observations.
(1) a. observation: “sonority always increases within #CCs”
b. syllabic interpretation: “TR = branching Onset”
c. explanation: “there are no #RT clusters because sonority must
increase within branching Onsets”
Circularity is introduced by the word “must”: the only thing the statement “sonority
must increase” follows from is precisely the observation “sonority does increase”.
Thus, the whole approach simply says “X is like it is because it is like it is.”
This kind of reasoning is not satisfactory. The problem of “why do RT-clusters
not occur word-initially in some languages” still begs the question. Before turning to
a different approach based on more general principles, I should like to discuss the
concept of constraints that is crucially involved in the circularity of the reasoning.
2. Constraints are observations, not explanations
As has been illustrated above, circularity is induced by the word “must”. Constraints
crucially rely on this word, which turns an observation into an obligation. Consider
the prototypical example in (2).
(2) a. observation trees grow straight up
b. the observed facts are not random trees always grow straight up
c. they must be as they are there is a constraint: GR.UP
“trees must grow straight up”
d. WHY do we observe these facts?
Because there is a constraint.
trees grow straight up because
GR.UP forces them to do so
Circularity is an intrinsic property of constraints. The observed facts are never
viewed as the consequence of something independent of the observa-
203
tion such as a more general principle. Uttering the “must” merely says “the observed
facts are as they are because I observe that they are as they are”. Actually, the "must"
suggesting an explanation is inherent in the word constraint. Instead of observing "X
is like that", using the word constraint in order to refer to an observation suggests
that "X is like that, and it cannot be in another way since it must be like that". If X is
constrained, something or somebody must be at the origin of its constrainedness.
However, the quest for this origin is never undertaken.
Still more oddly, constraints inhibit further investigation. Since the answer to the
question WHY? is the constraint, there is no more need to look for an explanation of
the facts. In the case of the tree-example, no one will have the idea of connecting the
observations to a conditioning factor such as sunlight because the constraint has
already explained why trees grow straight up. Or let us take the example of the
peach falling down. If physicists had considered that the peach does not go up or
zigzag around because there is a constraint saying that peaches must fall down, no
relation would ever have been established between falling peaches, their mass and
the mass of the earth. In the same way, the moon would be said to turn around the
earth because of a constraint that prevents it from drifting away. No relation between
the peach, the moon and something like gravitation would ever have been proposed.
Finally, one major goal of theories is to predict what cannot possibly exist. For
example, once physicists have observed the behaviour of falling peaches, the
phenomenon gravitation extends to anything that has a mass. Namely, the prediction
is made to the effect that there is no possible world where masses repel each other.
Nor is it possible for masses to remain without effect on each other. Thus, the
successive pieces of evidence adduced by Isaac Newton (masses attract each other),
Yuri Gagarin (but not in space) and Neil Armstrong (they do on the moon, but not so
much) showed different manifestations of the universally true principle “masses
attract each other”. There was no way for physicists to elude a unified account by
positing three different constraints (1. On earth, masses attract each other strongly, 2.
On the moon, attraction is poor, 3. In space, there is no attraction) or three different
rankings thereof. Anyone can imagine the state of our understanding of physics if
they had done so.
Returning to initial consonant clusters, no prediction of any kind is made by the
constraint “sonority must increase within a branching Onset”. If a planet were
discovered where the reverse phenomenology was found, nothing prohibits to
explain the new data by a constraint “sonority must decrease within branching
Onsets”. A theory that can cope with all possible data and their reverse is not a
theory at all, but a notational artefact enumerating observations. The interpretation
of the restrictions on word-initial clusters I develop in the remaining sections cannot
possibly do with the existence of such a planet.
204
3. Questions
If word-initial restrictions on consonant clusters are viewed as a consequence of the
constraint “sonority must increase within a branching Onset”, the problem has found
an answer. No further questions arise.
The questions I should like to ask are:
(3) a. What is sonority? Has some supreme being decided that [r] is a sonorant,
but not [t]?
b. What makes something a Liquid? For all other major classes, articulatory
criteria such as “obstructs the airflow more or less” can be appealed to.
These questions are worth asking because what is generally referred to as “sonority”
is a mere set of observations. The only available definition for sonorants is “this
segment is a sonorant because it behaves in a special way”. The special behaviour
opposing the subset of segments called sonorants to the subset called non-sonorants
is crosslinguistically stable. It can be observed when looking at the distribution of
segments (“only sonorants as the second member of word-initial clusters”, “only
(preferably) sonorants in Codas”, “the more distant a segment is from the Nucleus,
the less sonorant it is: C-Son-V-Son-C”) or their relative “vocalicity” (e.g. in typical
final devoicing systems such as German or Czech, Obstruents devoice domainfinally,
whereas sonorants and vowels do not). By these means, a so-called sonority
hierarchy can be established where segments are ordered according to their
behavioural kinship with vowels.
The sonority hierarchy is a crosslinguistically stable generalization that is based
on observations of different kinds. It encodes the obvious existence of a vocalic
continuum: segments are not either vocalic or non-vocalic, they are more or less
vocalic.
As such, the notion of sonority describes a fundamental observation. Hence,
phonological theory is called to account for this non-arbitrary, crosslinguistically
stable situation. One thing is self-evident and uncontroversial: sonority is an
idiosyncratic property of each segment. Thus, it must be encoded in what defines the
identity of each segment, i.e. its internal structure. And indeed, this is what models
of consonantal representation generally do: the sonority-value of a given segment is
commonly defined by a special primitive (features such as [son], [vocoid] or
[approximant] in Feature Geometry, e.g. McCarthy, 1988, the notion of Charm in
Government Phonology, cf. Kaye et al. (henceforth KLV), 1990).
If it is correct, however, to encode sonority in the internal structure of each
segment, using an extra-prime such as, say, [± son] is undesirable. Doing so is
nothing more than saying “according to my observations, this object is a sonorant. It
thus bears [+ son].” And, in turn, “this object is a sonorant because it bears [+ son].”
Again, the whole reasoning is circular.
205
Sonority should therefore be expressed as a function of the phonological
primitives that define each segment. Hence, a non-circular definition of sonority is
to be arrived at on the basis of consonantal structures that have been established on
sonority-independent grounds. Proposals such as Rice (1992) and Ritter (1997) take
their origin in this kind of consideration. I will follow this line when discussing the
phonological identity of sonority below. In short, sonority should fall out naturally
from an adequate model of consonantal representation. It should be a secondary
property of the internal structures proposed.
4. Two Different Kinds of Restrictions
Before turning to consonantal identities, I would like to discuss the correct
observation “word-initial restrictions on consonant clusters are related to sonority”
in some more detail. As a matter of fact, impossible word-initial clusters belong to
two different types:
(4) a. SYNTAGMATIC restrictions
#CCs that do or do not occur depending on the syntagmatic order of their
members: #TR is ok, but #RT out. In clusters of this type, the consonants
always contrast in sonority.
b. SEGMENTAL/PARADIGMATIC restrictions
There are also CCs of non-contrasting sonority that do not occur wordinitially:
e.g. *#lr, rl, nl, ln, tp. In these cases, the syntagmatic order of the
members is indifferent: they are unattested in any order.
The property “consonant cluster of contrasting sonority” seems to be related to syntagmatic
restrictions, whereas the non-occurrence of #CCs of non-contrasting
sonority has nothing to do with syntagmatic ordering. Rather, it is the consequence
of the cohabitation of two equally sonorant consonants. An explanation for the
former is to be sought in lateral relations holding among segments, while the latter
must be due to the genuine identity of the consonants involved. In the next section, I
will therefore address the question of consonantal representations.
5. Internal structure of consonants
The phonological identity of consonants is a classical issue under debate. Various
theories have put forward very different and partly incompatible models for the
representation of consonants. Representations diverge as to the phonological
primitives they use. Feature Geometry (e.g. Clements 1993; Sagey 1986) assumes
multi-valued features whereas Particle Phonology (Schane 1984), Dependency
Phonology (Anderson & Ewen 1987) as well as Government Phonology (KLV 1985)
rely on bigger, monovalent objects. They contrast with respect to the relations that
are supposed to hold between these primitives, i.e. arborescence (Feature Geometry)
vs. a dependency-type of relation (the other models
206
mentioned). But even within a given framework, the various proposals are far from
being consensual.
The model of consonantal representation that I shall discuss in more detail below
is couched within the framework of Government Phonology. Even though this
implies certain assumptions on the type and organization of the phonological
primitives used, I will try to present argumentation in a theory-neutral manner. The
present paper does not primarily aim at contributing to the debate on particular
consonantal representations that is going on within and among models. Rather, the
properties I evidence for different consonants should be general enough to be
transposable to other frameworks. This kind of approach is necessary because the
consonantal identities derived are only the first step of the demonstration intended.
In a second step, I shall capitalize on the sonority-oppositions then expressed by the
sole means of known phonological primitives in order to account for the restrictions
on word-initial clusters.
In the following section, I start by presenting the basic assumptions of the model
to be introduced.
5.1 Specific Assumptions on Consonantal Representations
Place limitations do not allow a detailed discussion of the model of consonantal
representation introduced below (see Scheer 1996, 1998c for a more developed
presentation). I only adduce evidence for properties of the model that differ from
other proposals. Special reference will be made to different proposals put forward
within Government Phonology including Harris (1990; 1994), Harris & Lindsey
(1995), Cyran (1994), Weijer (1994), Rennison (1990; in press).
1.The phonological primitives used, so-called Elements (see KLV 1985; de
Carvalho 1997 for discussion), are monovalent objects that are more complex than a
single distinctive feature. The vector of palatality for example, I, is defined as a
high, front, unrounded, non-ATR articulation and as such pronounceable in isolation
(cf. Harris 1996): I=[I].
2.As illustrated by the equation I=[I], there is a one-to-one correspondence
between phonetic manifestations and phonological representations: for a given
phonological structure, there is exactly one phonetic spell-out, and vice versa. Thus,
the acoustic signal of any language gives direct access to phonological structure, and
a given phonological representation has a crosslinguistically stable pronunciation. In
short, the observable objects and their postulated internal structure contract the same
relationship in phonology as, say, in chemistry: water is H2O on any continent. If it
was not, it would not be water anymore. And in turn, H2O cannot appear as
something else than water. The one-to-one relationship between phonetics and
phonology is necessary to constrain the grammar and to guarantee the possibility of
crosslinguistic argumentation. If [e] were X,Y in language A and Z,W in language
B, there would be no way to use evidence from the former in order to shed light on
the functioning of the latter.
207
Finally, a system that maximally restricts the phonetic interpretation of phonological
structure makes the most precise predictions: a given phonological identity is
predicted to sound alike in any language. Hence, a prediction of this kind can
potentially be falsified by any language. This situation is desirable because theories
are requested to be maximally easily falsifiable in the Popperian sense. The bijective
relation of phonetics and phonology is explicitly postulated by, among others, KLV
(1990:194), Clements (1993:101), Harris (1996) and Harris & Lindsey
(1990;1995:46ff). Under this assumption, the classically assumed phonetic
component of the grammar has no place: “purely phonetic effects” do not exist in
this view.
3.The contribution of primitives within a given segment is asymmetrical. Like in
any other linguistic domain, the head determines the phonetic output of the complex
expression to a greater degree than its dependents. In the representations developed
below, the head Element is underscored. Primitives that are not the head of the
expression are dependents/operators. In an I-headed expression, e.g. where A is
operator, I transmits more of its properties to the output than A, the result being the
front mid vowel [E]=I–A. In the reverse situation, the result is closer to A than to I,
that is A–I corresponds to the front low vowel [æ].
4.Vowels only use a subset of the places of articulation that can be found in
consonants: consonants and vowels can be palatal or velar, but only consonants may
be uvular, labio-dental or glottal. By contrast, consonants and vowels do not seem to
share common features as to the manner of articulation: only consonants are
obstruents or Liquids. As a consequence, exchanges of Place-defining primitives
between vocalic and non-vocalic positions are very frequent. In the case of
palatalizations for instance, there must be a single primitive responsible for
palatality in the triggering vowel and in the palatalized consonant: same causes,
same effects. If palatality were carried by a specifically vocalic primitive unknown
in consonants, it could never spread to non-vocalic positions. Place-assimilation
phenomena between vowels and consonants have led to the nowadays received
requirement that the set of primitives that define the Place of articulation be identical
for consonants and vowels (see e.g. Clements 1993; Smith 1988; Carvalho & Klein
1996; Weijer 1994:25,28; Cyran 1994:7; Harris & Lindsey 1995:65f). For
consonantal representations that have been proposed within Government Phonology
(e.g. Harris 1990; 1994), this means that the Element R representing coronality has
to be abandoned, as it does not occur in vowels (for more detailed evidence against
R, see Broadbent 1991; Backley 1993; Brockhaus 1994).
5.2 Inventory of Primitives
The inventory in (5) shows the primitives I assume to intervene in the definition of
vocalic as well as consonantal segments. The glosses indicate the articulatory
information carried by each prime.
208
(5) a. Place Elements phonetic result
I palatality [I]
U velarity [µ]
B labiality/roundness [¨]
A aperture, RTR2
[a]
v unmarkedness, relaxed tongue body position [ˆ]
b. Manner Elements
? constriction
h noise
B labiality/roundness
N nasality
–I ATR3
L lax vocal cords
H stiff vocal cords
T Trill
Most of the primitives shown are commonly used in the models quoted above. Only
B and T need to be introduced more carefully. The next section addresses issues
related to these two primes.
5.3 Velarity and Roundness Are Two Distinct Phonological Objects
Representational models like KLV (1985) or Anderson & Ewen (1987) define the
primitive U/|u| as the representative of high back tongue body position.
Nevertheless, its salient property is claimed to be roundness. This entails the
prediction that U/|u| is absent from any non-rounded articulation.
The proposal merging velarity and roundness is the consequence of a
crosslinguistic observation: in most languages, any back vowel is rounded, whereas
no language exists where any front vowel would be automatically rounded.
The prediction that U is absent from any unrounded articulation faces a number of
problems, however, among which the existence of back unrounded vowels [µ,F,√]
and the fact that all velar consonants [k,g,x,ƒ] are unrounded (see Roca 1994:120 on
this point). Claiming that the primitive that defines the back = velar Place of
articulation, i.e. U, does not contribute to the articulation of consonants (and some
vowels) that share the same Place of articulation seems odd in itself.4
Rather, data
from various languages given in (6) show that [u,w] that uncontroversially contain U
do interact with velar consonants.
(6) Interactions of [u,w] with velar consonants
a. In Fular,5
[w] regularly alternates with [g] (Klingenheben 1941:65;
1927:111ff explicitly on this point). Consider for example the different
forms of the stem wor ‘masculine’ when connected to the various adjectival
nominal class-suffixes.
209
class class class
1 N
gor-ba 9 gor-gal 18 N
gor-koj
2 wor-de 10 gor-gel 20 wor-/
be
3 N
gor-di 11 gor-gol 21 gor-/
de
4 wor-du 13 gor-ki 22 gor-/
di
5 N
gor-ga 15 N
gor-ko 23 gor-ko
8 N
gor-gu
b. Broken plural formation in Moroccan Arabic.
In the variety of Moroccan Arabic described by Ettajani (in prep.), only
velar and uvular consonants tolerate a labial secondary articulation:
[kw
,Xw
,qw
] exist, whereas *[sw
,Dw
] etc. do not occur. This distribution is
transparent in broken plural formation where a [w] tries to parachute onto
the first root-consonant (data and analysis by Ettajani):
1. Labial secondary articulation possible:
sing. broken plural
kbir kw
bar ‘tall’
Xubza Xw
bazi ‘bread’
Xurza Xw
razi ‘node’
kursi kw
rasi ‘chair’
qamiZa qw
amˆZ ‘shirt’
2. Labial secondary articulation impossible:
sing. broken plural
smin sman *sw
man ‘fat’
sˆlla slali *sw
lali ‘basket’
Drif Draf *Dw
raf ‘nice’
c. Short [u] in Ge'ez (Classical Ethiopian):
In Ge'ez (Ségéral 1995:155ff), short high peripheral vowels do not exist.
Only a short [u] can be observed in nominal morphology iff it is preceded
or followed by a velar or uvular consonant [k,g,q,x].
d. Czech vocative
In Czech, three vocative allomorphs occur with consonant-final masculine
nouns: -i iff the last consonant of the stem is palatal, -u iff it is velar, and -e
elsewhere.
nominative vocative
-i / Cpal__ kuu¯ kç¯-i ‘horse’
tçmaaS tçmaaS-i ‘Thomas’
ł˙aař ł˙aař-i ‘liar’
złçÔEj złçÔEj-i ‘thief’
słEc słEÔ-i ‘herring’
-u / Cvel__ ˙çx ˙çx-u ‘boy’
zdE¯Ek zdE¯k-u first name
ptaak ptaak-u ‘bird’
210
-e / elsewhere pEs ps-E ‘dog’
dçktçr dçktçr-E ‘doctor’
˙çłup ˙çłub-E ‘pigeon’
˙rat ˙rad-E ‘castle’
SEf Sv-E ‘seam’
The interaction of [u,w] with labials on the other hand is well established and
encoded in any of the representational models quoted. Thus, consonantal
representations should reflect the affinity of [u,w] with both labials and
velars/uvulars. In systems where velarity and roundness are undissociable, the
velarity-primitive cannot be linked to velar consonants because [k,g,x,ƒ] would then
be predicted to surface as rounded articulations. I therefore propose two different
primitives for velarity and roundness:
(7) a. U — velarity
b. B — roundness/labiality
This idea is not original, cf. Lass (1984:278f),6
Rennison (1990:187).7
Under these
provisos, labio-velar articulations such as [u,w] have the phonological identity U–B.
Their interaction with labial consonants is due to the labial agent B, their relation
with velars/uvulars involves the velar primitive U. B is present in all labial, U in all
velar/uvular consonants.
Proposing two independent vectors for velarity and roundness/labiality has
consequences for the representation of vowels in an Element-based system. Namely,
front rounded vowels are not a combination of I and U, but of I and B. Relevant
vocalic structures with U and B are given in (8).
(8) I I I I v v v v U U U U U I I/U-line
| | | | | | | |
A A A A A A A A A A-line
| | |
B B B B B B
i e y ø ˆ ´ ¨ P µ F u o A æ a
I E Y œ ъ √ U ç
j Á Â w
Phonological primitives are supposed to reside on autosegmental lines. The number
of lines is parametrically variable depending on the combination of Elements (KLV
1985). Elements sharing a given line cannot combine. Thus, in a language like
Arabic with only three vowels i,a,u where no elemental combinations occur, all three
Elements reside on the same line. In a five-vowel system where mid vowels
illustrate combinations of A and I/U, two lines are supposed, one carrying A, the
other shared by I and U. Finally, in systems with front rounded vowels, every
Element is said to reside on a line on its own because I and U combine. One
consequence of the dissociation of roundness and velarity
211
is that there is no combination of I and U in any circumstance: front rounded vowels
are expressed as a combination of I and B rather than of I and U. Hence, maximally
two melodic lines group A, I and U, as shown in the above diagram. The nonoccurrence
of combinations of I and U is a desirable effect. In models where both
primes are allowed to combine in vowels, they are expected to be able to do so in
consonants as well. However, in all models quoted, I and U do never co-define a
consonant. The issue of varying behaviour of I and U in consonants and vowels does
not arise when (8) is assumed.
Let us now turn to the status of the labial primitive. B was introduced as the
vector of roundness and labiality. This hybrid status is simply the translation of the
fact that it contributes roundness to vocalic articulations, whereas it is responsible
for the labiality of labial consonants. Two different terms are requested because both
objects stand in an intimate relation, without being identical: any rounded
articulation is labial, the reverse being wrong.
5.4 Only Place Definers Are Heads
Consonants can be grouped according to several Manners of articulation that do not
occur in the vocalic domain (stops, fricatives, voiced, unvoiced). Possibly, each of
the manners of articulation is to be represented by a special primitive. Postulating a
prime for each Manner of articulation leads to overgeneration, a classical problem of
systems of consonantal representation. I propose to approach this issue by limiting
the possible heads to Place-definers. Under these provisos, only the “melodic”
Elements v, I, U, A, B can dominate segmental expressions. This way of viewing
phonological structures has an articulatory foundation: if some segments were said
to be headed by a Manner-definer, say, ////, whereas others are not, the prediction
would be that these articulations have higher stopness than the ones where //// is an
operator. In Harris’s (1990) model, for example, bilabial stops are headed by ////,
whereas all other stops are dominated by an Element other than ////. According to the
dependency-principle that heads contribute more of their properties to the phonetic
output than operators, [p,b] should be more occlusive than, say, [t,d], a prediction
that is not supported by either articulatory, phonetic or phonological evidence.
The idea of restricting heads to Place-definers is not new. It implicitly controls the
vocalic representations proposed by KLV (1985): out of the six primitives of the
model (v,I,U,A,I-,N), only the Place definers v, I, U, and A head segmental
expressions. N- or I--headed structures complemented by another Element do not
occur.
The same idea is also present in the representations proposed within Feature
Geometry (e.g. Clements 1993) where Place and Manner definers are grouped under
categorically different and hierarchically distant nodes, and Dependency Phonology
(e.g. Anderson & Ewen 1987) where primitives depending on a Manner-gesture
never dominate segmental expressions.
212
As mentioned in the preceding section, B is by nature a hybrid primitive carrying
Place as well as Manner information. In the representations to be developed, B will
be able to head an expression iff it contributes to the Place-definition of the
articulation (labial consonants). On the other hand, B will never dominate
phonological expressions when contributing to the Manner of articulation (as in
rounded vowels and [S,Z], the only rounded consonants).
5.5 [t,d] Are Nothing
Another feature of the consonantal representations advocated here needs further
discussion. That is, [t,d] are articulated without contribution of any of the melodic
primitives I, U, A, B (see Szigetvári 1994 for an identical proposal). Two kinds of
arguments support this claim: markedness considerations and the epenthetic status of
[t,d].
Firstly, [t,d] seem to be present in almost all languages (cf. Nartey 1979;
Maddieson 1984). No other consonant has a similar status. Second, [t,d] seem to be
unmarked within the class of coronals that, as a whole, are unmarked with respect to
other Places of articulation (see the discussion in Paradis & Prunet 1991). Namely
the fricative coronals [s,z] sharply contrast with [t,d] as far as their phonological
behaviour is concerned (see e.g. Szigetvári 1994:198). This situation has led to a
view of [t,d] as placeless articulations in underspecification approaches (see e.g.
Archangeli 1988). In the vocalic domain, maximal unmarkedness is typically viewed
as a consequence of the absence of Place-definers. KLV (1985) for example attribute
the empty set represented by v to [ˆ], which is said to be the surface manifestation of
empty Nuclei. Following the same argumentation, [t,d] are melodically empty, i.e.
the phonetic manifestation of an empty Onset. Moreover, KLV (1985) argue for the
unmarkedness of [ˆ] on articulatory grounds: [ˆ] corresponds to the position of the
tongue body in relaxation. In the same way, [t,d] can be said to represent the
consonants that are produced with the position that the tongue body occupies when
no articulatory activity is carried out. The apex has just to be raised to the alveolus
(contribution of ////, h).
Accordingly, the internal structure of [t,d] is v–////–h.
Second, let us consider the status of [t,d] as epenthetic consonants. In French,
hiatuses that arise in certain morphological environments are regularly broken up by
[t]:
(9) /a il dit/ --> a-t-il dit ‘he has said’
/verra on/ --> verra-t-on ‘we will see’
The same phenomenon occurs between two morphemes belonging to the same
word:
213
(10) a. Epenthetic [t]
/esquimau + age/--> esquimautage
/glouglou + er/ --> glouglouter
/bijou + ier/ --> bijoutier
/indigo + ier/ --> indigotier
/tableau + in/ --> tableautin
/cacao + ière/ --> cacaotière
b. Epenthetic [d]
/Marivaux + er/ --> marivauder
From Middle High German to New High German,8
[t,d] develop at morpheme
boundaries after [n,r,s,x,g,f] (a phenomenon called Sproßkonsonant or Dentalwuchs,
cf. Paul et al. 1989:161).
(11) MGH NHG
a. After [n]
iergen irgend ‘any’
ieman jemand ‘somebody’
wîlen weiland ‘long ago’
vollen vollends ‘completely’
totzen Dutzend ‘dozen’
sinvluot Sintflut ‘Flood’
allenhalben allenthalben ‘everywhere’
wësenlîch wesentlich ‘important’
b. After [r]
anderhalp anderthalp ‘one and a half’
saf Saft ‘juice’
c. After [s]
ackes Axt ‘axe’
obeZ Obst ‘fruit’
sus sonst ‘otherwise’
bâbes Papst ‘pope’
d. After /x/
habech Habicht ‘hawk’
dornach Dornacht city
e. After [g] (rare)
bredige Predigt ‘sermon’
f. After [f]
werf Werft ‘shipyard’
The status of [t,d] as privileged epenthetic consonants is compatible with a minimal
phonological substance. If the expressive power of the grammar is constrained by an
Extended Projection Principle (see KLV 1990:221 for the
214
initial proposal) as in (12), then possible epentheses are limited to segments without
melodic content.
(12) Observable objects on the surface have a lexical origin or are the result of a
derivation based on lexically present material. “Nothing falls from heaven”.
Finally, it is intriguing to observe that [t,d] are very often and very easily affected by
phonological processes, but almost never constitute their output. Under the
assumption that [t,d] lack any melodic content, this observation receives a natural
explanation: if phonological processes move or replace primitives, the adjunction of
a primitive to a given structure cannot possibly yield an empty object. And of course
nothing can be replaced within an object with no content.
5.6 Distribution of A in Obstruents: spirantization
In this section, I use the evidence provided by spirantization to gain insight into the
internal structure of the segments involved in this process. I will first consider the
case of a classical spirantization, that is Grimm's Law (or First Consonant Shift),
where the stops and fricatives involved contrast with respect to their Place of
articulation. In a second step, I will examine Spanish and Tiberian Hebrew
spirantizations. In these languages, the Place of articulation of stops and fricatives
remains stable.
5.6.1 Spirantizations Accompanied by a Shift in the Place of Articulation
Consider the relevant data illustrating Grimm's Law given below (e.g. Paul et al.
1989:113f):
(13) Latin and Greek forms witness the Indo-European state of affairs (Gothic
spelling þ=[T]).
a. Spirantization9
IE >Germ.>Goth. Lat./Gr. Goth.
p,ph
f f pater fadar ‘father’
v b septem sibun ‘seven’
bh
v b fero bairan ‘carry’
t,th
T T tres *þreis ‘three’
ð d pater fadar ‘father’
dh
ð d Gr. δύρα daur ‘gate’
k,kh
X h cornu *haurn ‘horn’
“ g Gr. δάκρυ *tagr ‘tear’
gh
“ g hostis gasts ‘stranger’
b. Devoicing
b p p (s)lubricus *sliupan ‘sneak’
g k k ego ik ‘I’
d t t edo itan ‘eat’
215
The following three correspondences characterizing Grimm's Law can thus be
established for the oldest record of Germanic (see e.g. Collinge 1985:63ff):
(14) Indo-European Gothic
a. STOP, + voice, – asp STOP, – voice, – asp
b. STOP, + voice, + asp STOP, + voice, – asp
c. STOP, – voice, ± asp [FRIC, – voice, STOP, + voice] – asp
In the light of various secondary processes such as the Second Consonant Shift and
using arguments of comparative nature, the following correspondences are
commonly reconstructed for (unrecorded) Common Germanic (see Collinge 1985;
Paul et al. 1989 for further discussion):
(15) Indo-European Germanic
a. STOP, + voice, – asp STOP, – voice, – asp
b. STOP, + voice, + asp FRIC, ± voice
c. STOP, – voice, ± asp [FRIC, – voice]
According to classical interpretation (e.g. Paul et al. 1989:113), the chronology of
events is as follows: in a first step, IE non-aspirated unvoiced stops develop
aspiration: IE p,t,k > Germ ph
,th
,kh
. Then, all aspirated stops, voiced or not, become
fricatives: IE ph
,bh
, th
,dh
, kh
,gh
> Germ f/v, T/ð, X/“.10
The IE non-aspirated stops
that are left remain non-aspirated AND stops, but they devoice: IE b,d,g > Germ
p,t,k.
The table in (16) summarizes the evolution discussed.
(16) non-aspirated aspirated
voiced unvoiced unvoiced voiced
inventory of IE stops b, d, g p, t, k ph
, th
, kh
bh
, dh
, gh
Germanic ph
, th
, kh
bh
, dh
, gh
Grimm’s Law p, t, k f/v, T/ð, X/“
The crucial features of Grimm's Law to be emphasized are 1) only aspirated stops
spirantize, and 2) the Place of articulation of the stops is always different from the
one of the resulting fricatives (bilabial > labio-dental, alveolar > interdental, velar >
uvular).
As only aspirated stops spirantize in these languages, it seems adequate to regard
aspiration as a trigger of spirantization.
5.6.2 Spirantizations with no shift in the Place of articulation
In contrast with the example of spirantization discussed in the previous section, there
is another kind of spirantization where the Place of articulation of the stops and the
resulting fricatives remains stable. Consider the Spanish case below, where only
voiced stops undergo the spirantization triggered by a preceding vowel (data from
Hooper 1976:208ff; Hyman 1975:62):
216
(17) a. Fricatives occur after vowels:
la βaNka la banca ‘the bank’
la ðemora la demora ‘the delay’
la ƒƒƒƒana la gana ‘the desire’
b. Stops occur elsewhere:
1. word-initially
baNka banca ‘bank’
demora demora ‘delay’
gana gana ‘desire’
2. after consonants
ambos ambos ‘both’
onda onda ‘wave’
aldea aldea ‘village’
teNgo tengo ‘I have’
The alternations involved are [b ~ β], [d ~ ð] and [g ~ ƒ]. Apart from [d ~ ð] (on this
point, see Scheer, 1996:229ff), the place of articulation is invariable: bilabial–
bilabial, velar–velar. The phenomenon observed in Spanish also contrasts with the
spirantizations discussed earlier in that the triggering context has nothing to do with
aspiration but makes reference to sonority: fricatives surface in postvocalic contexts.
A similar situation obtains in Tiberian Hebrew where any stop, voiced or not,
undergoes spirantization in postvocalic contexts (data from Elmedlaoui 1993:124):
(18) perfective imperfective alternation(s)
√zkr zaaxar yi-zkor x ~ k ‘remember’
√kpr kaa∏∏∏∏ar yi-xpor k ~ x, ∏ ~ p ‘cover’
√bdl baaDDDDal yi-βdal b ~ β, D ~ d ‘separate’
√pth paaTTTTah yi-∏∏∏∏tah p ~ ∏, T ~ t ‘open’
√pgf paaƒƒƒƒaf yi-∏∏∏∏gof p ~ ∏, ƒ ~ g ‘meet’
Like in Spanish, the Place of articulation remains stable (except for the dentals), and
the triggering context is defined in terms of sonority (fricatives appear iff the
alternation-site is preceded by a vowel).
5.6.3 Summary
For the languages considered, the couples [triggering aspiration — shift in the Place
of articulation] and [triggering sonority — no shift in the Place of articulation] are in
complementary distribution. If segments of higher sonority assimilate nonsonorant
stops to more sonorant fricatives, the Place of articulation remains the same. If on
the other hand aspiration causes the spirantization of stops, a change in the Place of
articulation ensues.
217
Hence, we have to address the question regarding the causal relation between
aspiration and the instability of the Place of articulation. It is uncontroversial that
aspiration is a secondary articulation involving a fricative glottal activity. The
phonological primitive commonly related to glottal activity is the one responsible for
aperture, that is A (=RTR Retracted Tongue Root), [low], etc. (e.g. Harris 1994:119;
Clements 1993; Angoujard 1992). Accordingly, A is likely to be involved in the
secondary articulation at hand.11
(19) x
[Ch
]
C A
…
A will therefore be regarded as the melodic Element involved in the two changes
observed in Grimm-style spirantization (i.e. spirantization and Place-modification).
Since A is a Place-definer, it is natural to assume its implication in related processes.
If A is the melodic primitive responsible for the changes in the Place of articulation
of the stops, the following correspondences obtain:
(20) a. bilabial+ A = labio-dental
b. dental + A = interdental
c. velar + A = uvular
Hence, knowing about the phonological identity of bilabial (=B, see above), dental
(=v) and velar (v–U, according to the above discussion and the KLV-idea that velars
are v-headed, see e.g. Harris 1990; 1994) stops, the phonological identity of labiodental,
interdental and uvular fricatives may be calculated.
(21) bilabial labiodental dental interdental velar uvular
B B v v v v
| | | |
A A U U
|
A
If A is responsible for the Place-modification, the question as to why the stops at
hand spirantize remains. In order to approach this issue, let us consider what kind of
articulation would obtain if there were no spirantization. The incorporation of A into
the stops would yield labio-dental, interdental and uvular stops, respectively. Labiodental
and interdental stops do not exist (or are extremely marked, in any case nonphonemic).
This situation raises the more general question why some Places of
articulation lack stops (whereas there is no Place for which only stops exist to the
exclusion of fricatives): labio-dental (corresponding to [f,v]), interdental
(corresponding to [T,D]), alveo-palatal (corresponding to [S,Z]) or pharyngeal (corresponding
to [<,÷]) stops are unknown.
218
Why is there for example no stop with the same Place of articulation as [S,Z],
whereas [x,ƒ] have [k,g] as their stop-equivalents sharing the same Place of
articulation.
I propose a single answer to both questions:
(22) a. Stops incorporating A as in Grimm's law spirantize,
b. some Places of articulation lack stops because
A and //// are incompatible within a given phonological expression
This statement reflects the fact that both primitives involved, i.e. A = aperture and
//// = occlusion, are maximally antipodal both as far as their articulation is concerned,
and also with respect to the phonological behaviour of the segments they define. A
represents minimal, //// maximal obstruction of the airflow. A = [a] is maximally
sonorant, whereas stops are the least sonorant segments.
Assuming that A and //// are mutually exclusive has the following advantages:
1) Spirantization is an ordinary assimilation process where less sonorant objects
become more sonorant in a sonorant environment (e.g. Spanish). Unless (22) is
assumed, there is no apparent reason why the incorporation of A into stops should
produce fricatives (e.g. Grimm's Law). The Spanish type can be viewed as a true
spirantization, whereas the Grimm-type turns out to be a “false spirantization”:
spirantization here is only a secondary consequence of the fact that A and //// meet
within the same segment. A changes the Place of articulation. As a consequence of
its presence, //// is expelled from the internal structure of the segment. The following
example illustrates [ph
] > [f].
(23) x x
|
B > B
|
A A
/
|
h h
[ph
] [f]
2) The absence of stops for certain Places of articulation is a straightforward
consequence of the contribution of A to the articulation of the corresponding
fricatives ([f,v], [T,D], [S,Z], [<,÷]).
3) The most common affricates [pÉf], [tÉs, dÉz], [tÉS,dÉZ] match exactly the ‘holes’ of
lacking stops: [pÉf] is the labio-dental stop missing in its simplex form for the reasons
stated above, [tÉs, dÉz] are the stops corresponding to [s,z] ([t,d] do not have this
function since their fricative versions are [T,D], not [s,z], as demonstrated by
spirantization), and [tÉS,dÉZ] fill the empty stop-slot facing
219
[S,Z]. Affricates illustrate another possible result of what might happen when A
and //// are projected onto one another. In the case of Grimm's Law, //// has been
completely expelled. In the case of affricates, //// is expelled from the phonological
expression A contributes to. It is nevertheless maintained in a secondary articulation
of a contour segment. Consider for example the creation of [tÉS] resulting from the
projection of A and I onto [k]:
(24) x x
|
v I v I
| |
U A A
|
/
|
h h h
[k] [tÉS]
This analysis can contribute to the understanding of the most common pattern of
spirantization found in diachrony, that is stop > affricate > fricative. As an example,
the evolution of Latin velar stops through affricates to French [S,Z] occurring before
[a] can be partly accounted for: Latin gamba, carru > Gallo-Romance dÉZãmb´, tÉSar
> French Zãb, SaX ‘leg, tank’. Even though the source of the palatal agent remains
mysterious, the role played by the triggering context “__a” is better understood:
“__a” does not cause palatalization; it triggers affrication /spirantization.
5.7 Internal structure of Obstruents
In the preceding sections I examined those properties of the consonantal identities
advocated here that diverge from standard assumptions. The remaining
representations of Obstruents are more consensual. For instance, the labial primitive
B is present in all labials, the vector carrying palatality I is present in palatals
[c,Ô,ç,∆] and alveo-palatals [S,Z] (but also, more controversially, in [s,z]), the velar
primitive U contributes to the articulation of velars and uvulars, and the apertureprimitive
A is common to gutturals. Moreover, the primitive representing ATRness –I
is viewed as an activator of the buccal cavity. Thus, it is present in buccal, and
absent from non-buccal, i.e. guttural articulations. For example, –I distinguishes
between the buccal epenthetic consonants [t,d] and the non-buccal epenthetic
consonant [/], both being identical as far as their melodic properties are concerned.
The table in (25) summarizes the consonantal identities I propose (for further
discussion, see Scheer 1996; 1998c).
220
(25) B I U ////
H p pÉf t tÉs c tÉS k q /
? + h
L b — d dÉz Ô dÉZ g G
H ∏ f T s ç S x X < h
h
L β v D z ∆ Z ƒ “ ÷ ˙
A
v v v v v v
| |
B B I I I U U U
| | | |
A A A A A A A
| | | | |
I- I- I- I- I- I- I- I- I- I|
| | | | | | | | |
h h h h h h h h h h h h h
| | | | |
(/) (/) (/) / / /
p f c S T s k X q < h / t
b v Ô Z D z g “ G ÷ ˙ d
∏ ç x
β ∆ ƒ
5.8 Liquids and Nasals
5.8.1 [r] is made of A and I
The evolution of Germanic languages evidences the tight relation between [r] and
low vowels. In (standard) German for instance, the former MHG apical [r] is
distributed as follows (see e.g. Drosdowski 1984:35f, 57 for more illustration).
(26) a. [r] > [å] / V __ # fooå vor ‘before’
nuå nur ‘only’
hç“çå Horror ‘horror’
mawå Mauer ‘wall’
bææå Bär ‘bear’
biiå Bier ‘beer’
leeå leer ‘empty’
fçjå Feuer ‘fire’
b. [r] > [å] / V __ C luåç Lurch ‘amphibian’
g´bIåg´ Gebirge ‘mountain’
lçåt Lord ‘Lord’
221
c. [r] > [a] / a __ {C,#} baat Bart ‘beard’
baaS Barsch ‘perch’
faat Fahrt ‘trip’
gaa gar ‘done, cooked’
d. [r] > [“/X] / C __ d“aj, *dåaj drei ‘three’
g“ajs, *gåajs Greis ‘old man’
pXajs, *påajs Preis ‘price’
e. [r] > [“] / V __ V pI“aat Pirat ‘pirate’
ka“aat Karat ‘carat’
oo“aan Oran Algerian city
In word-final position after a vowel (other than [a]) (vor) and after a vowel (other
than [a]) followed by a consonant (Gebirge), MHG [r] has become a low vowel
different from [a] that Drosdowski (1984:57) transcribes central [å]. In the same
context, if the vowel preceding MHG [r] is [a], the result is a long [aa] (Bart, gar).
Elsewhere, that is intervocalically (Pirat) and after a consonant (drei, Preis), the
former [r] is realized [X,“] according to the voice-value of the preceding segment.
As (26) shows, the vocalic variant of [r] seems to be [a] or [å]. I therefore propose
that [r] is A-headed.
Lowering-activity of [r] is also reported for English where only mid and low
vowels occur before etymological [r] (cf. Harris 1994:244). Cyran (1994:210) and
Broadbent (1991) adduce more evidence supporting the presence of A in [r].
Furthermore, the following evidence suggests that I also contributes to the
articulation of [r]. In Southern Dutch (Rotterdam, Leiden), [j] can replace [r] wordfinally
and before a consonant:
(27) standard Dutch Southern Dutch
daar daaj daar ‘over there’
kaart kaajt kaart ‘card’
stççrt stççjt stoort ‘disturb 2SG.PRES’
karn´n kajn´n karnen ‘make buttermilk’
vErpt vEjpt werpt ‘throw 3SG.PRES’
In the same context, Caribbean Spanish exhibits [j] where standard Spanish shows
[r,l] (data from Harris 1983):
(28) standard Spanish Caribbean Spanish
revolver revojvej ‘revolver’
karta kajta ‘card’
papel papej ‘paper’
algo ajgo ‘something’
This kind of phenomenon occurring in the typical lenition-context “__C,#” is
analysed as a segmental decomposition by Harris (1992). If this way of interpret-
222
ing the above data is correct, the result of the decomposition [j]=I must also be
present in the original [r,l].
5.8.2 [r], [l] and [n] Are Variants of the Same Phonological Object
Several genetically non-related languages present alternations of [r], [l] and [n]
without apparent segmental conditioning.
In Chaha (Ethio-Semitic language) for instance, [r] and [n] are allophones, [n]
occurring word-initially and before obstruents, [r] elsewhere (see Leslau 1950:118
for a more detailed description).
(29) PRETERITE PRESENT JUSSIVE √
1SG nädäf-xw
ˆm ä-rädˆf nˆ-ndˆf Rdf ‘card (wool)’
1SG näky
äm-xw
ˆm ä-räky
ˆm nˆ-räkˆm Rky
m ‘ride (horse)’
In Korean, [l] and [r] are allophones. [r] is found intervocalically, whereas [l] occurs
word-finally and in consonantal environments (data from Labrune 1993:336, 342):
(30) √aR ‘know’
/aR + ta/ --> aal-ta CITATION FORM
/aR + ¨pnita/ --> ar-¨pnita POLITENESS FORM
/aR + ¨o/ --> ar-¨o EXHORTATIVE FORM
/aR + a/ --> ar-a DECLARATIVE FORM
Moreover, [l]/[r] have a third allophonic variant word-initially, that is [n] (but not
every Korean [n] is an allophone of [l]/[r]):
(31) √Rak
/o + Rak/ --> o-rak ‘diversion’
/ø + Rak/ --> nak ‘pleasure’
/Rak + wçn/ --> nak-wçn ‘paradise’
Like in Chaha, the allophonic relationship between [n], [r] and [l] suggests a unique
melodic identity for all three segments.
In MHG, finally, there are numerous doublets of the same word involving [l] and
[r] (cf. Paul et al. 1989:144).
(32) [r] [l] NHG
Herke Helche Helke female first name
smieren smielen — ‘smile’
prior priol Prior ‘prior’
murmern murmeln murmeln ‘murmur’
Canterbury Candelberc Canterbury Canterbury
marmor marmel Marmor ‘marble’
marter martel Marter ‘torture’
mörter mörtel Mörtel ‘mortar’
turter turtel Turteltaube ‘turtledove’
môrber mûlber Maulbeere ‘mulberry’
223
MHA dörper ‘farmer’ > dörpel > törpel > NHG Tölpel ‘dolt’
This “confusion” also suggests that [l] and [r] are made of the same phonological
substance.
5.8.3 [l] and [n] contain I
In (standard) German, [X] and [ç] are in complementary distribution. [ç] occurs after
front vowels, [X] after [a,o,u]:
(33) a. [X] after [u,o,a] [ç] after [y,ø,i,e]
Absence of I Presence of I
buuX byyçå ‘book SG./PL.’
kçX kœçIn ‘cook MASC./FEM.’
baX bEç´ ‘creek SG./PL.’
Iç ‘I’
b. mIlç ‘milk’
manç´ ‘some’
As can be seen in (33a), [ç] is the palatalized version of underlying /X/: in [kçX] vs.
[kœçIn] ‘cook MASC./FEM.’, the I contained in the feminine suffix umlauts the rootvowel
which, in turn, palatalizes /X/ into [ç]. However, [ç] also occurs after [n] and
[l] as shown in (33b). Since the eventual palatality of /X/ is induced by the preceding
segment, these consonants must be I-providers.
Italian also evidences the presence of I in [l]. As a second member of branching
Onsets, [j] is the regular reflex of former [l], as evidenced for example by più < plus.
5.8.4 Lowering Properties of Nasals
MHG high vowels followed by a (geminated) Nasal regularly surface as mid vowels
in NHG (cf. Paul et al. 1989:79).
(34) MHG NHG
sunne Sonne ‘sun’
sumer Sommer ‘summer’
kumen kommen ‘come’
münech Mönch ‘monk’
sun Sohn ‘son’
künec König ‘king’
gewunnen gewonnen ‘won’
geswummen geschwommen ‘swum’
This lowering ability supposes the presence of A in Nasals. In the same way,
nasalization of vowels by following Nasals generally yields mid nasal vowels, even
if the starting point was a high vowel. E.g. in the history of French, the process VN
> V˜ / __{C,#} gave rise to the three French nasal vowels [E),ä˜,ç)], none of which is
high (cf. alternations such as [fE)] fin ‘thin MASC.’ vs. [fin] fine ‘thin FEM.’
224
where the input is [i]). The same evolution under identical contextual conditions has
taken place in the history of Slavic. And again, the resulting nasal vowels were all
mid or low, even if high vowels were nasalized (see e.g. Trávníček 1935:44ff;
Vondrák 1906:335ff).
5.8.5 Summary: Internal structure of Nasals and Liquids
The table in (35) summarizes the results of the preceding sections.
(35) 6.8.1. [r] is A-headed (German, English)
6.8.1. I contributes to the articulation of [r] (Spanish, Dutch)
6.8.2. [l], [n] and [r] have the same melodic identity (Chaha, Korean,
MHG)
6.8.3. [l] and [n] contain I (German, Italian)
6.8.4. Nasals contain A (MHG > NHG, French and Slavic nasal vowels)
I conclude therefore that [l,r,n] contain A and I, the former being the head of the
structure. This situation raises the question as to what makes these three articulations
different. The apex trills when [r] is produced, whereas it does not for [l] and [n]. [n]
is nasal, the others are not. Given these properties, I propose that [r] is a trilling [l],
whereas [n] is a nasal [l]. Hence, [n] is made of the same primitives as [l] plus N.
And besides the Elements defining [l], a primitive responsible for the trill of the
apex contributes to the production of [r]. It is therefore necessary to introduce a new
Element, T “trilling apex”, that contributes to the articulation of trills and is absent
from the internal structure of non-trilling consonants. T belongs to a category of
marginal primitives that define non-delayed secondary properties of consonants such
as ingression, trill or retroflexion. Their distribution is mostly physiologically
restricted (e.g. to areas available for the apex as far as trill and retroflexion are
concerned).
The phonological identities given in (36) integrate the results arrived at so far and
the identities for the various Places of articulation discussed above (I contributes
palatality, B labiality, U is head in uvulars, operator in velars).
(36) Liquids Nasals
B
A A A A A A A A
I I U I I U U
T N N N N N
r l ł m n ¯ N ≤
Under these provisos, the following generalizations can be stated, allowing for a
purely formal definition of the objects ‘Nasal’ and ‘Liquid’:
225
(37) a. Liquids are A-headed
b. Nasals contain A and N
5.9 Sonority
Within the model of consonantal representation introduced, the set of observations
referred to as sonority is a mere function of three parameters: 1) the linking of the
phonological expression at hand to an Onset or a Nucleus, 2) the presence of the
consonantal primitives /////h and 3) the role played by A within the expression (head,
operator, absent). The following table summarizes the definition of sonority on
purely formal grounds:
(38) segment Nucleus/Onset h// role of A
a N — head
e,o N — operator
i,u N — absent
Liquids O — head
Nasals O — head/operator
Glides O — absent
s,z O h head
gutturals O h head/operator
fricatives O h operator/absent
stops O h and / absent
For example, a sonorant is an object residing in an Onset from which //// and h are
absent. A vowel is an object residing in a Nucleus. An Obstruent is a segment
residing in an Onset to whose articulation //// and/or h contribute(s).
6. Consonantal interaction: Paradigmatic aspect
Having introduced consonantal representations where the set of observations called
sonority receives a phonological expression, let us reconsider the questions on wordinitial
clusters raised in §4, repeated hereafter for convenience.
(39) a. Syntagmatic restrictions
#CCs that do or do not occur depending on the syntagmatic order of their
members: #TR is okay, but #RT out. In clusters of this type, the consonants
always contrast in sonority.
b. Segmental/paradigmatic restrictions
There are also CCs of non-contrasting sonority that do not occur wordinitially:
e.g. *#lr, rl, nl, ln, tp. In these cases, the syntagmatic order of the
members does not matter: they are unattested in any order.
Let us first address the latter issue. Consider the situation obtaining for the melodic
lines hosting A, I and U within the model of consonantal representation developed
above when two consonants occur in a row. “ ” indicates the ab-
226
sence of any primitive. The juxtaposition of an empty and a filled position is noted
by “⇐”.
(40) a. p r t r k l f r
I/U ⇐ I ⇐ I U I ⇐ I
| | | | | | | |
A ⇐A ⇐A ⇐ A A A
b. n r s r ł r t p
I/U I I I I U I
| | | | | | | |
A A A A A A A
Typical branching Onsets as in (40a) oppose at least one empty and one filled
position on a given line. By contrast, (40b) shows consonant clusters that do not
occur word-initially for the paradigmatic reasons mentioned (they are *#__ in any
order). These never oppose an empty and a filled position on a given line. The
representations in (40) of course are only a choice of possible combinations of two
consonants. Place limitations preclude an exhaustive survey. The reader may verify
that the above statements have general value in Scheer (1996: 320ff).
This distribution relating possible #CCs to the opposition between filled and
empty positions leads me to propose the following definition of consonantal
interaction:
(41) Infrasegmental Government (IG)
Iff a phonological primitive faces an empty position on a given autosegmental
line, it may govern this position.
According to (41), the consonant clusters of (40b) can contract no infrasegmental
governing relation because either both positions on a given line are occupied ([nr],
[sr], [łr]) or both are empty ([tp]).
IG is a development of Harris’s (1990) notion of segmental complexity. Harris
argues that interconsonantal relations depend on the number of phonological
primitives the head and the dependent are made of. The more complex a consonant
(i.e. the more primitives it is made of), the better head of a consonantal domain it is,
and vice versa. Obviously, IG relies on this notion: a consonant C1 may govern
another consonant C2 iff C1 is more complex than C2. Within the representational
framework developed in the previous sections, complexity mainly depends on the
presence/absence of A, i.e. the vector of sonority.
However, this approach would not be any better than the one based on sonority
constraints if it could not say WHY possible #CCs should depend on consonantal
interaction. The distributional relation between the possible consonantal interaction
of CCs occurring word-initially and the impossible interaction of CCs that do not
occur in this context is nothing more than an observation.
Therefore, the next section examines how IG interacts with lateral relations
holding between phonological categories.
227
7. The Phonological ECP, CVCV and the Beginning of the Word
In recent work, the hypothesis assuming a strict CVCV syllabic structure has been
evaluated for particular analyses in various languages.12
The CVCV-model
(Lowenstamm 1996) views syllabic structure as a strict consecution of nonbranching
Onsets and non-branching Nuclei (i.e. no branching constituents, no
Codas). For the sake of clarity, consider the representation of closed syllables,
geminates, long vowels and the right edge of consonant-final words within this
frame:
(42) closed syllable geminate long vowel […C#]
O N O N O N O N O N O N O N
C V C ø C V C V …C ø #
All structural information contained in traditional syllabic approaches is preserved.
For instance, the site of “closed-syllable” phenomena such as devoicing, lenition,
shortening etc. that occur word-finally and before consonants usually receives the
uniform description “Coda”. In a CVCV approach, these phenomena are referred to
as occurring “before an empty Nucleus”. The difference between both descriptively
equivalent statements is the causal relation between the relevant environment and
the observed phenomena. Apart from the general observation that Codas are “weak”
because e.g. they admit only a subset of possible consonants, there is no reason why
segments should devoice, deaspirate, lenite, in short decomplexify in this special
position. The correct cross-linguistic observation pointing to the weakness of Codas
can only lead to a less surprised reaction when devoicing etc. occurs once more in a
Coda-position. It can hardly explain this fact. By contrast, if the Onset is universally
viewed as a dependent of the Nucleus like e.g. in Government Phonology, then the
fact that objects decomplexify before an empty Nucleus stands in a direct causal
relation with the emptiness of the latter. That is, the licensing power of an empty
category is smaller than that of a filled category.
A CVCV structure multiplies the number of empty categories, namely of empty
Nuclei. This situation raises the more general question of the status of empty
categories in linguistics. It seems to be consensual that “you cannot get an empty
category for free”. This idea is encoded within the Empty Category Principle saying
that an empty category may remain unexpressed if and only if precise conditions are
met. These conditions are defined in terms of the relation the empty category
contracts with a filled position that is laterally distant. In syntax, it was proposed that
movement could only take place if the trace of the moved object in its now empty
base-position is properly governed by this object in its new position. Proper
Government was defined by the structural relation the filled and the empty position
contracted (c-command, barriers). This example provides the kind of motivation
typical for the existence of empty catego-
228
ries. If there were no structure preservation, i.e. if the category the object was moved
from were deleted or not even present lexically, no explanation along the above lines
would be available.
Empty categories do burden the grammar because they request special care
(defined e.g. as Proper Government). Nevertheless, their existence is a necessary
condition for an explanatory account. Hence, the burdening of the grammar by more
empty categories should not be viewed as an undesirable overload, but on the
contrary as a welcome source of explanation. If grammar is not free in its moves
because it must create or maintain the conditions requested for the existence of
empty categories, this constitutes a step towards a more constrained model. The
challenge, as for any other scientific theory, is to propose a model that is as
constrained as possible while covering all relevant data.
The same reasoning holds for phonology. KLV (1990:219) proposed phonological
Proper Government based on the same kind of lateral long-distance phenomena
involving an empty and a filled category that led to syntactic Proper Government. In
their view, empty categories are subject to the ECP in phonology as well as in
syntax. An adapted version of their phonological ECP is given in (43).
(43) Empty Category Principle
An empty Nucleus may remain unexpressed iff it is properly governed.
The mentioned long-distance phenomena are vowel-zero alternations that are
typically sensitive to the object(s) occurring between the zero (empty Nucleus) and
the vowel (filled Nucleus) to its right.13
Consider e.g. Czech hudøb-a ‘music
NOM.SG.’ vs. hudeb-ní, *hudøb-ní ‘musical ADJ.’ or Moroccan Arabic køtˆb-ø ‘he
writes PF.’ vs. kˆttˆb-ø, *køtti-b-ø ‘he causes to write’. If both are separated by more
than one consonant, the expected zero surfaces as a vowel. The intervening CCcluster
is viewed as a barrier that does not allow the filled Nucleus to properly
govern the empty Nucleus, which must therefore appear on the surface.14
However, the blocking effect of the “barrier” CC is a pure observational fact that
does not follow from anything, see (44a). By contrast, the multiplication of empty
Nuclei when assuming a CVCV structure offers a straight forward answer to the
question “why do intervening CCs block Proper Government (PG)?”:
229
(44) a. Non-CVCV: Why do intervening CCs block PG?
PG
R R
O N C O N C
x x x x x x
k ˆ t t ˆ b
b. CVCV: why do intervening CCs block PG? Because the [CC] encloses an
empty Nucleus N☺, /CN☺C/, that seeks PG. PG is not blocked, it simply
cannot reach the first [ˆ].
PG
O N O N☺ O N O N
k ˆ t t ˆ b
However, intervening CCs do not “block” PG in all circumstances. In Czech
prefixes for instance, -e- alternates with zero: pode-brat ‘seize from below’ vs. podøbradek
‘double chin’. The alternation corresponds to a contrast in the lexical
structure of the roots involved: /ber/ for [-br-at] where -e- is properly governable (cf.
1sg [-ber-u]) vs. /brad/ for [-brad-ek] where -a- is not properly governable (Scheer
1996; 1997 for a complete demonstration). In the former case, -a- properly governs
the -e- of /ber/. As a consequence, the prefixal -e- fails to undergo Proper
Government. It therefore appears on the surface (unassociated segments are
inaudible).
(45) PG
O N O N – O N☺ O N O
p o d e b e r a t
By contrast in the case of podø-bradek, -a- properly governs the prefixal -øalthough
the CC [-br-] stands in between the governor and the governee.
(46) PG
O N O N – O N☺ O N O N O N
p o d e b r a d e k
So far, only two phonological operations that are able to satisfy the ECP have been
identified: 1) Proper Government and, in extension of the above definition, 2)
Licensing of final empty Nuclei (cf. Kaye 1990). Assuming a CVCV struc|
|
230
ture, the empty Nucleus N☺ in cases such as (46) is neither final nor concerned by
Proper Government (the -a- governs the prefixal -e-). Nevertheless, it does not
surface. Hence, the ECP must be satisfied by another phonological operation. I
propose that consonantal interaction as described above can close its domain to the
effect that the ECP is satisfied:
(47) Phonological operations that are able to satisfy the ECP15
a. Proper Government
b. Licensing of final empty Nuclei
c. Infrasegmental Government:
The empty Nucleus N☺ of a domain [CN☺C] may remain unexpressed if a
relation of Infrasegmental Government holds between its surrounding
consonants.
Under these provisos, PG can apply over /bN☺r/ (podø-bradek) because the cluster
constitutes a domain of IG. By contrast, no domain of IG can be established within
the cluster /ber/ (pode-brat) since the properly governable /-e-/ prohibits consonantal
communication. In this case, PG applies to the nearest target available, that is the
/-e-/ mentioned. As a consequence, the prefixal Nucleus fails to undergo PG and
receives phonetic interpretation.
In cases like podø-bradek, PG by -a- cannot be held responsible for the muteness
of N☺ in /-bN☺r-/ because its effect can be seen on the prefix. Lowenstamm (in
press) argues that this situation in fact is general even if there is no prefix involved:
the first vowel of a word governs what is generally referred to as “#”. The
phonological identity of the non-linguistic object “#” is an empty Onset followed by
an empty Nucleus:
(48) The beginning of the word: “#” is an empty CV
PG
O N O N☺ O N…
b r a…
The initial Nucleus being subject to the ECP, it seeks PG from the first vowel of the
word. Hence, the first vowel can never properly govern N☺.
We have now reached the point where an answer to the question raised at the end
of the previous section can be provided. It had been shown that impossible wordinitial
clusters such as #nl, #tp are precisely the ones for which the consonantal
identities developed predict that no relation of Infrasegmental Government may
hold. Hence, it was tempting to establish a causal relation between both facts saying
“a word-initial CC can exist only if it constitutes a domain of Infrasegmental
Government”. Nevertheless, there was no apparent reason why Infrasegmental
Government should be a condition on possible word-initial CCs. Assuming CVCV
and the phonological identity of “#”=CV, the question “why
231
are CCs within which no IG holds no possible word-initial clusters?” receives the
answer “because the empty Nucleus they enclose is subject to the ECP and IG the
only way to satisfy it”. As an example, consider the situation for possible #tr as
opposed to impossible *#nr:
(49) a. #tr b. *#nr
PG
IG PG
O N O N☺ O N… O N O N☺ O N…
⇐ I a… I I a
⇐ A A A
t r n r
In both cases (49a–b), the ECP concerning the initial empty Nucleus is satisfied
through PG by the first vowel of the word. In contrast, only the ECP applying to the
N☺ of (49a) is satisfied: [tr] can interact and close their domain, whereas [nr] cannot.
(49b) is ruled out because it contains an empty Nucleus, N☺, that is not licensed by
any of the phonological operations that may satisfy the ECP.
8. Consonantal Interaction: Syntagmatic Aspect
Up to this point, it has been shown that the set of CCs that do not occur wordinitially
for paradigmatic reasons coincide with the set of CCs within which
consonantal identities predict that no Infrasegmental Government can hold.
Moreover, a causal relation between impossible IG and the non-occurrence of CCs
in word-initial position has been established.
I now wish to address the syntagmatic aspect of the restrictions on word-initial
clusters: why do consonant clusters of the #TR-kind occur word-initially, but not
their mirror-image *#RT? The particular identities of the consonants involved do not
matter here because they are identical for the occurring as well as for the nonoccurring
CCs. In theory, Infrasegmental Government is possible for both #TR
(right-to-left) and *#RT (left-to-right):
(50) Head-final IG Head-initial IG
O N☺ O N O N☺ O N
⇐ I I ⇒
⇐ A A ⇒
t r r t
232
Based on independent evidence, Charette (1990) has noticed that interconsonantal
relations depend on the availability of a vocalic support for the head of the
consonantal domain. Consider the universal claim Charette makes about
interconsonantal relations:
(51) Government Licensing (adapted from Charette 1990)
A consonant C1 can govern another consonant C2 only if C1 is licensed to do so
by its Nucleus.
Government Licensing was developed within a non-CVCV frame where
interconsonantal relations are expressed by means other than Infrasegmental
Government and the consonantal identities differ from the ones proposed above.
However, the idea expressed by Government Licensing is theory-neutral: in order
for a consonantal cluster to exist, its head needs vocalic support. Let us see hereafter
which are the predictions made by Government Licensing as to word-initial clusters:
(52) a. Well-formed structure b. Ill-formed structure
PG PG
O N O N☺ O N… O N O N☺ O N…
T ⇐ R V… I ⇒ I a
IG IG
Under the assumption of Infrasegmental Government, R is always the head of the
domain of consonantal interaction, and T is the dependent. According to
Government Licensing, R needs to be licensed by its Nucleus in order to be able to
govern T. In (52a), the Nucleus following R hosts the first vowel of the word. This
vowel can license R so that R is able to establish IG over T. In (52b), however, the
Nucleus N☺ following R is empty in any event. Hence, it can never license R which,
in turn, is unable to govern T. Although the two members of #RT fulfil segmental
requirements for an Infrasegmental Government relation, no such relation can be
established for syntagmatic reasons. As a consequence, N☺ does not satisfy the ECP,
and (52b) is ill-formed.
Charette's Government Licensing thus correctly predicts the non-occurrence of
initial *#RT clusters.16
9. Conclusion
In this article, I have tried to develop a non-circular alternative to the standard way
of handling the distributional restrictions that apply to word-initial consonant
clusters. At no point of the argumentation, a constraint intervenes. Rather, the set of
observations expressed by commonly used constraints such as
Lic Lic
|
233
“within a branching Onset, sonority must increase” follow from more general
principles.
(53) Restrictions on word-initial consonant clusters follow from
a. Government Licensing (Charette 1990),
b. segmental complexity (Harris 1990),
c. the phonological ECP (KLV 1990),
d. CVCV and “#”=CV (Lowenstamm 1996; in press),
e. the consonantal identities developed.
All of these devices are assumed to operate in Phonology generally. None of them
makes special reference to the particular issue discussed, that is word-initial CCs.
For this reason, the approach presented avoids circularity.
Finally, the presented theory of consonantal interaction, unlike constraint-based
models, makes the prediction that a world where only word-initial clusters of
decreasing sonority occur could not possibly exist.
Address of the Author
Tobias Scheer
1467, Chemin des Adrets
Les Bucoliques
FR-06370 Mouans-Sartoux (France)
email: scheer@naxos.unice.fr
Notes
* I am grateful to Krisztina Polgárdi, Honoré Kamany, Patricia Cabredo-Hofherr, Joaquim Branda˜o
de Carvalho and an anonymous reviewer for valuable comments.
1
For more discussion of this approach, see for example Clements (1990), Selkirk (1984) and
references therein.
2
Cf. Harris’s (1994:119) statement about the twofold properties of A: Articulations involving A
“are produced by lowering and retracting the tongue body.”
3
In more recent work, efforts have been made in order to account for ATR contrasts in terms of
structural oppositions rather than using an independent prime (see for instance the discussion in
Harris & Lindsey 1995:62ff). The representation of ATR is not crucial for the purpose of the
present paper. For this reason, I will provide no further discussion.
4
More detailed discussion of the vocalic part of this prediction can be found in Scheer (1996:151ff).
5
West-Atlantic language spoken in Guinea. Data from Klingenheben (1941:17).
6
“I propose introducing |µµµµ| ‘velarity’ and |ω| ‘labiality’/’roundness’, and dispensing with |u|
completely. |u| seems to conflate too many properties anyhow. […] Overall I think it’s a good idea
for ALL markedness considerations to be excluded from phonological characterizations. […]
Segments ought to code only their own properties, not statistics of cross-language distribution”.
7
Rennison defines “U — pull towards high back tongue body position (but with no implication of
lip rounding!)”.
8
Abbreviations used in this article: IE = Indo-European, Germ = Common Germanic, Got = Gothic
(about 4th century A.D.), OHG = Old High German (about 850–1050 A.D.), MHG = Middle High
German (about 1050–1350 A.D.), NHG = New High German (since about the 16th century),
Lat = Latin, Gr = Greek.
9
Spirantization occurs in any context except sC-clusters (Got sp,sk,st) and ht,ft (e.g. Lat stella,
OHG stërno) and IE [pt,kt] (e.g. Lat captus, noctis, Got haft, nahts (OHG naht > NHG Nacht)). Cf.
Paul et al. (1989:113f).
234
10
There is debate on the status of labials, see Braune & Ebbinghaus (1981:49), Jellinek (1892), Paul
et al. (1989:113f, 124). The voicing of resulting fricatives is controlled by Verner’s Law: iff the
fricative is followed by a voiced articulation (=vowel, sonorant, voiced Obstruent) and the
preceding vowel it is unstressed in IE, then the fricative is voiced. Otherwise, it is unvoiced (see
e.g. Paul et al. 1989:123f for illustration).
11
Of course this does not imply that A is the sole defining feature of aspiration, see e.g. Harris
(1994:135) for discussion.
12
See e.g. Lowenstamm (1988; 1996), Guerssel & Lowenstamm (in prep.), Bendjaballah (1995),
Creissels (1989), Bonvino (1995), Ségéral (1995), Hérault (1989), Nikiema (1989), Ségéral &
Scheer (in press), Larsen (1994; 1995), Heo (1994), Scheer (1996; 1997; 1998a,b).
13
See e.g. Kaye (1989; 1990), Charette (1990), Scheer (1996; 1997; 1998a,b) for data and analyses
concerning vowel-zero alternations.
14
See e.g. KLV (1990), Kaye (1990), Charette (1990), Scheer (1996; 1997; 1998a,b) on Proper
Government.
15
Other proposals such as Interonset Government and magic licensing, which are discussed in
Gussmann & Kaye (1993), Cyran & Gussmann (forthcoming) and Kaye (1992), are not relevant
for the purposes of this paper. See Scheer (1998c) for discussion.
16
Domains of IG are defined in the lexicon. Accordingly, in languages where consonants are
lexically unrelated to syllabic constituents, IG does not occur. It is interesting to note that
languages allowing for both #TR and #RT clusters are precisely representatives of Afro-Asiatic.
These languages have a templatic structure, that is syllabic constituents and segmental information
do not co-habit in the lexicon. Lowenstamm (in press) argues for a different status of the initial CV
in this kind of languages. According to his analysis, the initial empty Nucleus is not always subject
to PG in templatic systems. If this view is correct, then the possible occurrence of both #TR and
#RT clusters in templatic languages follows: in #CN☺CV, V properly governs N☺, not the initial
CV. Consequently, the surrounding consonants are subject to no co-occurrence restrictions.
References
Anderson, J.M. & C.J. Ewen 1987. Principles of Dependency Phonology. Cambridge: Cambridge
University Press.
Angoujard, J-P. 1992. Quelques “Elements” pour la représentation des gutturales. Talk given at the
First Conference on Afro-Asiatic Languages, Paris.
Archangeli, D. 1988. Aspects of underspecification theory. Phonology 5, 183–208.
Backley, P. 1993. Coronal: "The undesirable element." UCL Working Papers in Linguistics 5, 301-
323.
Bendjaballah, S. 1995. "Aspects apophoniques de la vocalisation du verbe berbère (Kabyle)." In: L.
Nash, G. Tsoulas & Anne Zribi-Hertz (eds.), Actes du deuxième colloque ‘Langues et
Grammaire’, 33–47.
Bonvino, E. 1995. Aspects de l'article défini en romanesco. MA thesis, Université Paris 7.
Braune, W. & E.A. Ebbinghaus (1981). Gotische Grammatik. 19th edition, Tübingen: Niemeyer.
Broadbent, J. 1991. "Linking and intrusive r in English". UCL Working Papers in Linguistics 3, 281–
302.
Brockhaus, W. 1994. Segmental representations without [R°]? Some evidence from German. Talk
given at the Government Phonology Workshop, Vienna, April 1994.
235
de Carvalho, J. Branda˜o 1997. "Primitives et Naturalité". In: B. Laks (ed.), Nouvelles Phonologies,
Languages 125, 14–34.
de Carvalho, J. Branda˜o & M. Klein 1996. "A subsymbolic approach to phonological primitives." In:
J. Durand & B. Laks (eds.), Current trends in Phonology: Models and Methods. Salford: ESRI,
99–123.
Charette, M. 1990. "Licence to govern." Phonology Yearbook 7, 233–253.
Clements, G.N. 1990. "The role of the sonority cycle in core syllabification." In: J. Kingston & M.
Beckman (eds.), Papers in Laboratory Phonology, I: Between the Grammar and Physics of
Speech. Cambridge: CUP, 283–333.
Clements, G.N. 1993. "Lieu d'articulation des consonnes et des voyelles: une théorie unifiée." In: B.
Laks & A. Rialland (eds.), L'architecture des représentations phonologiques. Paris: CNRS
Editions, 101–145.
Collinge, N.E. 1985. The Laws of Indo-European. Amsterdam - Philadelphia: Benjamins.
Creissels, D. 1989. "Structures syllabiques du dialonké de Faléya." Mandekan 18, 1–30.
Cyran, E. 1994. Vocalic Elements in Phonology: A study in Munster Irish. Doctoral dissertation,
University of Lublin.
Cyran, E. & Gussmann, E. Forthcoming. Consonant Clusters and Governing Relations: Polish Initial
Consonant Sequences. Ms., Catholic University of Lublin.
Drosdowski, G. (ed.) 1984. Grammatik der deutschen Gegenwartssprache. 4th edition, Mannheim.
Elmedlaoui, M. 1993. "Gemination and spirantization in Hebrew, Berber and Tigrinya: A “FortisLenis
Module” analysis." Linguistica Communicatio 5, 121–176.
Ettajani, N. In prep. Les pluriels brisés en arabe marocain. Ms., Université Paris 7.
Guerssel, M. & J. Lowenstamm. In prep. The Derivational Morphology of the Classical Arabic Verb.
Ms., Université Paris 7 & UQAM.
Gussmann, E. & J. Kaye 1993. "Polish Notes from a Dubrovnik Café." SOAS Working Papers in
Linguistics and Phonetics 3, 427–462.
Harris, J. 1983. Syllable structure and stress in Spanish. Cambridge, Mass.
Harris, J. 1990. "Segmental complexity and phonological government." Phonology Yearbook 7, 255–
300.
Harris, J. 1992. "Licensing Inheritance." UCL Working Papers in Linguistics 4, 359–406.
Harris, J. 1994. English sound structure. Oxford: Blackwell.
Harris, J. 1996. "Full interpretability in phonology." In: J. Durand & B. Laks (eds.), Current trends in
Phonology: Models and Methods, Salford: ESRI.
Harris, J. & G. Lindsey 1990. "Phonetic interpretation in generative grammar." UCL Working Papers
in Linguistics 2, 355–369.
Harris, J. & G. Lindsey 1995. "The elements of phonological representation." In: J. Durand & F.
Katamba (eds.), Frontiers of Phonology. London: Longman, 34–79.
Heo, Y. 1994. Empty categories and Korean Phonology. Doctoral dissertation, SOAS, London.
Hérault, G. 1989. "Les rections syllabiques en soninké." Linguistique Africaine 3, 43–90.
Hooper, J.B. 1976. An introduction to natural generative Phonology. New York: Academic Press.
Hyman, L. 1975. Phonology: theory and analysis. New York.
Jellinek, M.H. 1892. "Zur Frage der Verschiebung der Gutturale." Zeitschrift für deutsches Alterthum
36, 77–88.
236
Kahn, D. 1976. Syllable based generalizations in English Phonology. Doctoral dissertation, MIT.
Kaye, J. 1989. "Government in Phonology: The case of Moroccan Arabic." The Linguistic Review 6,
131–159.
Kaye, J. 1990. "'Coda' licensing." Phonology Yearbook 7, 301–330.
Kaye, J. 1992. "Do You Believe in Magic? The Story of s+C Sequences." SOAS Working Papers in
Linguistics and Phonetics 2, 293–313.
Kaye, J., J. Lowenstamm & J-R. Vergnaud (=KLV). 1985. "The internal structure of phonological
representations: A theory of Charm and Government." Phonology Yearbook 2, 305–328.
Kaye, J., J. Lowenstamm & J-R. Vergnaud (=KLV). 1990. "Constituent structure and government in
phonology." Phonology Yearbook 7, 193–231.
Klingenheben, A. 1927. "Die Laute des Ful." Zeitschrift für Eingeborenen-Sprachen, Beiheft 9.
Klingenheben, A. 1941. "Die Suffixklassen des Ful." Zeitschrift für Eingeborenen-Sprachen, Beiheft
23.
Labrune, L. 1993. Le statut de /r/ en japonais et en coréen: Histoire, typologie, structure interne des
segments. Doctoral dissertation, Université Paris 7.
Larsen, B.U. 1994. Some Aspects of Vowel Length and Stød in Modern Danish. MA thesis,
Université Paris 7.
Larsen, B.U. 1995. "Italian vowel length, Raddoppiamento Sintattico and the definite article: A
unified analysis in Government Phonology." In: L. Nash, G. Tsoulas & A. Zribi-Hertz (eds.),
Actes du deuxième colloque ‘Langues et Grammaire’, 110–124.
Lass, R. 1984. Phonology: An introduction to basic concepts. Cambridge: CUP.
Leslau, W. 1950. "Chaha: An Outline of the Grammar." In: W. Leslau (ed.), Gurage studies: Collected
articles, Wiesbaden, 117–137.
Lowenstamm, J. 1981. "On the maximal cluster approach to syllable structure." Linguistic Inquiry 12.
Lowenstamm, J. 1988. Another look at Ngizim syllable structure. Talk given at the 19ème conférence
de linguistique africaniste, Boston University, April 1988.
Lowenstamm, J. 1996. "CV as the only syllable type: Full interpretability in phonology." In: J.
Durand & B. Laks (eds.), Current trends in Phonology: Models and Methods. Salford: ESRI,
419–441.
Lowenstamm, J. In press. "The beginning of the Word." In: J. Rennison (ed).
Maddieson, I. 1984. Patterns of Sounds. Cambridge.
McCarthy, J. 1988. "Feature Geometry and Dependency: A review." Phonetica 43, 84–108.
Nartey, J.N.A 1979. "A Study in Phonemic Universals — Especially Concerning Fricatives and
Stops." UCLA Working Papers in Phonetics, 46.
Nikiema, E. 1989. "Gouvernement Propre et licenciement en phonologie: le cas du Tangale. Langues
Orientales Anciennes, Philologie et Linguistique 2, 225–251.
Paradis, C. & J-F. Prunet 1991. The special status of Coronals: Internal and external evidence. San
Diego.
Paul, H., P. Wiehl & S. Grosse 1989. Mittelhochdeutsche Grammatik. 23rd edition, Tübingen:
Niemeyer.
Rennison, J. 1990. "On the elements of phonological representations: The evidence from vowel
systems and vowel processes." Folia Linguistica 24, 175–244.
Rennison, J. (ed.). In press. Syllables!? La Hague: Holland Academic Graphics.
237
Rennison, J. In press. "Can there be empty elements? On empty heads and empty operators." In: J.
Rennison (ed.).
Rice, K. 1992. "On Deriving Sonority: a Structural Account of Sonority Relationships." Phonology 9,
61–99.
Ritter, N. 1997. Encoding Sonority in a Revised Theory of Segmental Representations. Talk given at
HILP 3, Vrije Universiteit Amsterdam, January 1997.
Roca, I. 1994. Generative Phonology. London: Routledge.
Sagey, E. 1986. The Representation of Features and Relations in Non-linear Phonology. PhD dissertation,
MIT.
Schane, S.S. 1984. The Fundamentals of Particle Phonology. Phonology Yearbook 1, 129–155.
Scheer, T. 1996. Une théorie de l'interaction directe entre consonnes. Doctoral dissertation, Université
Paris 7.
Scheer, T. 1997. "Vowel-zero Alternations and Their Support for a Theory of Consonantal
Interaction." In: P.M. Bertinetto, L. Gaeta, G. Jetchev & D. Michaels (eds.), Certamen
Phonologicum III. Torino: Rosenberg & Sellier, 67-88.
Scheer, T. 1998a. "A unified model of Proper Government." The Linguistic Review 15,1, 41-67.
Scheer, T. 1998b. "Governing Domains are Head-final." In: E. Cyran (ed.), Structure and
Interpretation. Studies in Phonology. Lublin: Folium, 261-285.
Scheer, T. 1998c. "La structure interne des consonnes." In: P. Sauzet (ed.), Langues et Grammaire IIIII,
Phonologie. Paris: Université Paris 8, 140-172.
Ségéral, P. 1995. Une théorie généralisée de l'apophonie. Doctoral dissertation, Université Paris 7.
Ségéral, P. & T. Scheer 1998. "A Generalized Theory of Ablaut: The case of Modern German Strong
Verbs." In: A. Ortmann, R. Fabri & T. Parodi (eds.), Models of Inflection. Tübingen: Niemeyer,
28-59.
Selkirk, E. 1984. "On the Major Class Features and Syllable Theory." In: M. Aronoff & R.T. Oehrle
(eds.), Language Sound Structure. Cambridge, Mass.: MIT Press, 107–136.
Smith, N. 1988. "Consonant Place Features." In: H. van der Hulst & N. Smith (eds.), Features,
segmental structure and harmony processes, Part I. Dordrecht: Foris, 209–236.
Szigetvári, P. 1994. "Coronality, Velarity and why They Are Special." The Even Yearbook, 185–224.
Trávníček, F. 1935. Historická mluvnice cÿeskoslovenská. Praha: Úvod, Hláskosloví a Tvarosloví.
Vondrák, W. 1906. Vergleichende Slavische Grammatik. I. Band Lautlehre und Stammbildungslehre.
Göttingen.
van de Weijer, J.M. 1994. Segmental structure and complex segments. Doctoral dissertation,
Rijksuniversiteit te Leiden.