Contrastive Hierarchy Theory and the Nature of

Features
B. Elan Dresher

1. Introduction*

I will present a brief introduction to a theory of contrastive features in phonology. I will set out the
main tenets of this theory, and consider what implications it has for understanding phonological
features. I propose that the language learner's task is to arrive at a set of hierarchically-ordered
contrastive features that account for the phonological patterning of the input language. That is, I
assume that the contrastive feature hierarchy is universal, not the features themselves or their ordering.
This requirement puts strong constraints on phonological representations, and accounts for why
phonological systems resemble each other, without assuming that features are innate. This approach
also makes it possible for us to consider possible parallels between phonological and morphosyntactic
features in a new light.
In section 2, I briefly review why there is a growing consensus that phonological features are not
innate, and why it is necessary to have something else that will account for why phonological
representations are the way they are. In section 3, I present the main tenets of contrastive hierarchy
theory, and show how it imposes strong constraints on feature systems, and on the relation between
contrast and phonological activity. Section 4 gives an extended example that illustrates how a
contrastive feature hierarchy can contribute to the synchronic analysis of the Oroqen vowel system. In
section 5, I review recent proposals by Cowper and Hall that morphosyntactic features are also
organized into contrastive hierarchies. Section 6 is a brief conclusion.

2. Phonological features: Not innate, but why emergent?

There is a growing consensus that phonological features are not innate, but rather ‘emerge’ in the
course of acquisition. Most of the contributions to a volume titled Where do phonological features
come from? (Clements & Ridouane 2011) take an emergentist position; none argue for innate features.
Mielke (2008) and Samuels (2011) summarize the arguments against innate features. From a
biolinguistic perspective, phonological features are too specific, and exclude sign languages (van der
Hulst 1993; Sandler 1993). Empirically, no one set of features have been discovered that ‘do all tricks’
(as Hyman 2011 writes with respect to tone features, but the remark applies more generally). Finally,
since at least some features have to be acquired based on evidence of language-specific phonological
activity, a prespecified list of features becomes less useful in learning than had once been thought.
But if features are not innate, what compels them to emerge at all? It is not enough to assert that
features may emerge, or that they are a useful way to capture phonological generalizations. Assuming
that phonological representations are indeed composed of distinctive features, we need to explain why
features inevitably emerge, and why they have the properties that they do.1 In particular, we have to

*
University of Toronto, [email protected]. This is a slightly revised version of portions of my talk at
WCCFL 35. I would like to thank the audience for their comments and questions. For discussions, ideas, and
analyses I would also like to thank Elizabeth Cowper, Daniel Currie Hall, Christopher Harvey, Peter Jurgec, Ross
Krekoski, Andrew Nevins, Will Oxford, Keren Rice, Christopher Spahr, and Zhang Xi.
1
Though I assume here that phonological primes are binary features, it is an empirical hypothesis that the learner
creates binary features and not other sorts of entities, such as privative elements or dependency structures of
various kinds. I have argued (Dresher 2014) that such elements should also be organized into contrastive
hierarchies (pace Scheer 2010).

© 2018 B. Elan Dresher. Proceedings of the 35th West Coast Conference on Formal Linguistics,
ed. Wm. G. Bennett et al., 18-29. Somerville, MA: Cascadilla Proceedings Project.
 19

explain why learners, or some learners, do not simply posit segment-level representations. Further, are
there limits to how broad or narrow features are, or how many features can be associated with a given
phonological inventory? How many features should learners posit for a three-vowel system, for
example?
Contrastive hierarchy theory provides an answer to these questions: learners must arrive at a set of
hierarchically ordered features that distinguish between all the phonemes of their language. We will
see that this requirement imposes strong constraints on the number of features that can be posited, and
on what feature systems can look like.

3. A theory of phonological contrast

The theory of contrast presented here has roots in the work of the Prague School phonologists N.
S. Trubetzkoy and Roman Jakobson and their colleagues and students. These phonologists emphasized
the importance of contrastive features in the patterning of phonological systems; over time, contrastive
features came to be associated with ‘branching trees’. This style of contrastive underspecification
entered the literature rather inconspicuously, as I have tried to show in a number of publications
(Dresher 2009, 2015, 2016). Early, though inexplicit, examples can be found in the work of Jakobson
(1962 [1931]) and Trubetzkoy (1939) in the 1930s, and continuing with Jakobson 1941 and Jakobson
& Lotz 1949. More explicit examples can be found in Jakobson, Fant & Halle 1952, Cherry, Halle &
Jakobson 1953, Jakobson & Halle 1956, and Halle 1959. This approach was imported into early
versions of the theory of Generative Phonology; it is featured prominently in Harms 1968, the first
Generative Phonology textbook. However, branching trees (and contrastive underspecification, more
generally) were omitted from Chomsky & Halle 1968, and disappeared from mainstream phonological
theory for the rest of the century.
As a general theory of phonological representations, branching trees were revived, under other
names, by Clements (2001, 2003, 2009), and independently at the University of Toronto, where they
are called contrastive feature hierarchies (Dresher, Piggott & Rice 1994; Dyck 1995; Zhang 1996;
Dresher 1998; Dresher & Rice 2007; Hall 2007; Dresher 2009; etc.). It is the latter approach I will be
presenting here. It has gone under various names: Modified Contrastive Specification (MCS), ‘Toronto
School’ phonology, Contrast and Enhancement Theory, or just Contrastive Hierarchy Theory.2

3.1. Main tenets of the theory

The first major building block of our theory is that contrasts are computed hierarchically by
ordered features that can be expressed as a branching tree. Branching trees are generated by what I call
the Successive Division Algorithm (Dresher 1998, 2003, 2009):

(1) The Successive Division Algorithm

Assign contrastive features by successively dividing the inventory until every phoneme has been
distinguished.

What are the criteria for selecting and ordering the features? Phonetics is clearly important, in that
the selected features must be consistent with the phonetic properties of the phonemes. For example, a
contrast between /i/ and /a/ would most likely involve a height feature like [low] or [high], though
other choices are possible, such as [front/back] or [advanced/retracted tongue root].
Of course, the contrastive specification of a phoneme can sometimes deviate from its surface
phonetics. In some dialects of Inuktitut, for example, an underlying contrast between /i/ and /ɨ/ is
neutralized at the surface, with both /i/ and /ɨ/ being realized as phonetic [i] (Compton & Dresher

2
See Dresher 2009: 163f. for references and a detailed review of early work in the MCS framework. For more
recent applications and developments, see Hall 2011, 2017, Mackenzie 2011, 2013, Ko 2012, Spahr 2014, 2016,
Oxford 2015, Hall & Hall 2016, and Krekoski 2016.
20

2011). In this case, /i/ and /ɨ/ would be distinguished by a contrastive feature, even though their surface
phonetics are identical.3
As the above example shows, the way a sound patterns can override its phonetics (Sapir 1925).
Thus, we consider as most fundamental that features should be selected and ordered so as to reflect the
phonological activity in a language, where activity is defined as follows (adapted from Clements 2001:
77):

(2) Phonological activity

A feature can be said to be active if it plays a role in the phonological computation; that is, if it is
required for the expression of phonological regularities in a language, including both static
phonotactic patterns and patterns of alternation.

The second major tenet has been formulated by Hall (2007) as the Contrastivist Hypothesis:

(3) The Contrastivist Hypothesis

The phonological component of a language L operates only on those features which are necessary
to distinguish the phonemes of L from one another.

That is, only contrastive features can be phonologically active. If this hypothesis is correct, then
(4) follows as a corollary:

(4) Corollary to the Contrastivist Hypothesis

If a feature is phonologically active, then it must be contrastive.

One final assumption is that features are binary, and that every feature has a marked and
unmarked value. I assume that markedness is language particular (Rice 2003, 2007) and accounts for
asymmetries between the two values of a feature, where these exist. I will designate the marked value
of a feature F as [F], and the unmarked value as (non-F). I will refer to the two values together as [±F].

3.2. Restrictions on feature systems

The theory proposed above imposes major restrictions on feature systems. For example, if a
language has three vowel phonemes /i, a, u/, and if the vowels are split off from the rest of the
inventory so that they form a sub-inventory, then they must be assigned a contrastive hierarchy with
exactly two vowel features. Though the features and their ordering may vary, the limit of two features
constrains what the hierarchies can be. Two possible contrastive hierarchies using the features [back]
and [low] are shown in (5).

(5) Two contrastive hierarchies for a three-vowel system with [back] and [low]
a. [back] > [low] b. [low] > [back]
wo wo
[back] (non-back) [low] (non-low)
ru g g ru
[low] (non-low) /i/ /a/ [back] (non-back)
g g g g
/a/ /u/ /u/ /i/

In (5a), [back] is ordered over [low] (indicated as [back] > [low]); therefore, all the vowels are
divided on the basis of [±back]. There is only one (non-back) vowel, /i/; as it is already distinguished
from the other vowels, it receives no further features. In this ordering, [±low] is contrastive only

3
While the realizations of the two phonemes are identical, their effects on neighbouring segments differ, which is
how we can tell them apart.
 21

among the [back] vowels, and distinguishes /a/ from /u/. In (5b), all the vowels are distinguished by
[±low], with [±back] being contrastive only in the (non-low) domain. Though the two systems in (5)
make use of the same features and appear to have the ‘same’ phonemes, we expect them to pattern
differently. In (5a), /u/ is most closely connected to /a/, whereas in (5b) it is closer to /i/; we predict
that these differences in organization will be reflected in patterns of merger and neutralization. In (5a),
both /a/ and /u/ have a contrastive [back] feature that could participate in phonological processes, such
as vowel harmony or consonant backing; in (5b), only /u/ is contrastively [back], and we would not
expect /a/ to trigger backing.
Another difference between the systems in (5) is that in (5a), the marked value of the higher
feature is expanded, with the consequence that one vowel (/a/) has two marked features. In (5b), the
unmarked value of the higher feature is expanded, and no vowel has more than one marked feature.
Though the content of the features may vary from language to language, every three-vowel system will
have a contrastive feature tree that is isomorphic to either (5a) (if the marked value of the higher
feature is expanded) or (5b) (if the unmarked value is expanded).
A four-phoneme inventory can have a minimum of two features and a maximum of three. Four
such feature systems are shown in (6), where Fi represent features whose phonetic content is to be
supplied. The tree in (6a) represents a system with maximum economy, wherein the two features
completely cross-classify the four phonemes, which are represented by /1/–/4/. In this system there are
four marked specifications: /1/ has marked values of [F1] and /F2/; /2/ is marked for [F1]; /3/ is
marked for [F2]; and /4/ has no marked specifications.
The trees in (6b–d) represent systems that use three features for four phonemes. They differ in
how many marked specifications they have. The system in (6b) has three marked feature
specifications: /1/, /2/, and /3/ are marked for [F1], [F2], and [F3], respectively, and /4/ is unmarked.
(6c) is a system with five marked specifications, and (6d) has six, which is the maximum limit. Not
shown is one more three-feature system with a symmetrical tree like (6a), where [F1] expands to [±F2]
and (non-F1) expands to [±F3]). Though marked features may vary from three to six, the limit of three
features may not be exceeded.

(6) Contrastive hierarchies for a four-phoneme system

a. Two features, markedness = 4 b. Three features, markedness = 3
wo wo
[F1] (non-F1) [F1] (non-F1)
ru ru g ei
[F2] (non-F2) [F2] (non-F2) /1/ [F2] (non-F2)
g g g g g ru
/1/ /2/ /3/ /4/ /2/ [F3] (non-F3)
g g
/3/ /4/

c. Three features, markedness = 5 d. Three features, markedness = 6

wo wo
[F1] (non-F1) [F1] (non-F1)
ei g ei g
[F2] (non-F2) /4/ [F2] (non-F2) /4/
g ru ru g
/1/ [F3] (non-F3) [F3] (non-F3) /3/
g g g g
/2/ /3/ /1/ /2/

In general, the number of features required by an inventory of n elements will fall in the following
ranges: the minimum number of features is equal to the smallest integer that is greater or equal to
log2n; and the maximum number of features is equal to n–1. Some sample values of feature minima
and maxima for inventories of different sizes are shown in Table 1.
22

Phonemes log2n min max Phonemes log2n min max

2 1 1 1 10 3.32 4 9
3 1.58 2 2 12 3.58 4 11
4 2 2 3 16 4 4 15
5 2.32 3 4 20 4.32 5 19
6 2.58 3 5 25 4.64 5 24
7 2.81 3 6 32 5 5 31
8 3 3 7 36 5.17 6 35
Table 1: Minimum and maximum number of features for various-sized phoneme inventories

As can be observed in Table 1, the minimum number of features goes up very slowly as phonemes
are added; the upper limit rises with n. However, inventories that approach the upper limit are
extremely uneconomical. At the maximum limit, each new segment is distinguished by a unique
contrastive feature that is unshared by any other phoneme. For practical reasons, this upper limit is
almost unattainable for larger inventories. Thus, the contrastive hierarchy and Contrastivist Hypothesis
account for why phonological systems resemble each other in terms of representations, without
requiring individual features to be innate.

3.3. Enhancement of contrastive features

Unless a vowel is further specified by other contrastive features (originating in another vowel or in
the consonants), it is made more specific only in a post-phonological component. Stevens, Keyser &
Kawasaki (1986) propose that feature contrasts can be enhanced by other features with similar acoustic
effects (see also Stevens & Keyser 1989; Keyser & Stevens 2001, 2006). Hall (2011) shows how the
enhancement of contrastive features can result in configurations predicted by Dispersion Theory
(Liljencrants & Lindblom 1972; Lindblom 1986; Flemming 2002). Thus, a vowel that is [back] and
(non-low), like /u/ in (5), can enhance these features by adding {[round]} and {[high]}, becoming [u]. I
follow Purnell & Raimy (2015) in indicating non-contrastive properties contributed by enhancement in
curly brackets. These enhancements are not necessary, however, and other realizations of /u/ in (5) are
possible (Dyck 1995; Hall 2011).

4. Contrastive feature hierarchies in synchronic phonology: The Oroqen vowel

system

The Xunke dialect of Oroqen (Tungusic, Northern China and Inner Mongolia) has nine vowel
phonemes (length contrasts are omitted; they are not relevant here):4

(7) Xunke Oroqen vowel system (length distinctions omitted)

/i/ /u/
/ʊ/
/e/ /ə/ /o/
/ɛ/ /ɔ/
/a/

For insight into what features might be contrastive in this language, we need to look at how the
vowels pattern, that is, at the types of phonological activity they exhibit. The three most notable kinds
of phonological activity involving vowels are RTR (retracted tongue root) harmony, labial (rounding)
harmony, and palatalization. We will consider each of these in turn.

4
This inventory is based on the accounts of Hu (1986), Zhang, Li & Zhang (1989), Li (1996), and Zhang (1996).
Every vowel except /e/ and /ɛ/ has a long counterpart. Sources differ as to whether /e/ and /ɛ/ are short, long, or
diphthongs, but these details are not relevant here.
 23

4.1. RTR harmony

Vowels fall into two sets: yin, or non-RTR, vowels include /u, e, ə, o/; yang, or RTR, vowels
include /ʊ, ɛ, a, ɔ/. Only vowels from the same set may co-occur in a word, as shown in (8):

(8) [RTR] and (non-RTR) vowels do not co-occur in a word (Zhang 1996: 155–156)
a. RTR words b. Non-RTR words
ʊla ‘quill’ ulə ‘meat’
ɔɔjalɛɛ ‘sisters-in-law’ ujəlee ‘cousin’
kɔɔsʊn ‘empty’ kosuun ‘pond’

The vowel /i/ is neutral and may co-occur with either set (9):

(9) /i/ occurs in both [RTR] and (non-RTR) words (Zhang 1996: 157)
a. RTR words b. Non-RTR words
mʊrin- ‘horse’ nəkin- ‘sweat’
tari- ‘that’ ulin- ‘gifts’
birakan- ‘river’ bitə- ‘letter’

Except for /i/, every non-RTR vowel has an RTR counterpart with which it alternates: /u/
alternates with /ʊ/, /e/ with /ɛ/, /ə/ with /a/, and /o/ with /ɔ/. The RTR vowels /ʊ/, /ɛ/, /a/, and /ɔ/ trigger
RTR stem-to-suffix harmony within a word, creating alternations in suffix vowels. An example is the
definite object, which alternates between -wə in non-RTR words and -wa in RTR words; in the words
in (10), /w/ appears as m following a nasal consonant. Another example is the dative suffix -du/dʊ.

(10) Stem-to-suffix RTR harmony (Zhang 1996: 180–181)

a. RTR words b. Non-RTR words
kɔɔkan-ma ‘child DEF.OBJ’ bəjun-mə ‘moose DEF.OBJ’
bʊwa-dʊ ‘place DAT’ utə-du ‘son DAT’

The vowel /i/ is neutral and transparent to harmony (11): it does not disrupt the RTR or non-RTR
character of a word:

(11) /i/ is neutral and transparent to RTR harmony (Zhang 1996: 158, 180–181, 190)
a. RTR words b. Non-RTR words
mʊrin-sal ‘horse PL’ dəji-səl ‘bird PL’
wargi-tʃara ‘salty DIM’ toŋgorin-tʃərə ‘round DIM’
tʊkala-dʒi ‘clay INST’ sukə-dʒi ‘axe INST’

The evidence from activity, therefore, is that every vowel except /i/ has a plus or minus value of
an active feature that I have been calling [RTR]; by hypothesis, this feature must be contrastive. Note
that /i/ is phonetically {(non-RTR)}, so it is not immediately obvious on phonetic grounds that /i/ is not
contrastive for this feature.5 Nevertheless, in order to account for the facts of RTR harmony, the
Oroqen learner will have to find a feature ordering in which the feature [±RTR] does not apply to /i/.

5
On the contrary, it has been asserted that high front vowels are particularly compatible with an advanced tongue
root and antagonistic to a retracted tongue root (Archangeli & Pulleyblank 1994); one might therefore suppose
that /i/ would be the vowel most likely to occur exclusively with (non-RTR) vowels, but this is not the case. It
appears that it is the contrastive status of /i/, and not its phonetic characteristics, that determines its non-
participation in RTR harmony.
24

4.2. Labial (rounding) harmony

Only the low vowels /o/ and /ɔ/ trigger labial harmony, and only /ə/ and /a/ undergo rounding: /ə/
alternates with /o/, and /a/ alternates with /ɔ/. Two successive /ɔ/ or /o/ vowels cause a suffix /a/ or /ə/
to become round.6 In RTR words, the present tense suffix -ra alternates with -rɔ, and in non-RTR
words, -rə alternates with -ro, as shown in (12). Note that /u/ and /ʊ/ do not trigger rounding, as shown
in (13).

(12) Stem-to-suffix labial harmony

a. RTR words b. Non-RTR words
baka-ra ‘get PRES.TNS’ nəkə-rə ‘weave PRES.TNS’
ɔlgɔɔ-rɔ ‘dry PRES.TNS’ mooro-ro ‘moan PRES.TNS’

(13) No labial harmony with /u, ʊ/

a. RTR words ʊrʊʊn-ma ‘hoof DEF.OBJ’ *ʊrʊʊn-mɔ
b. Non-RTR words ulgulu-wə ‘moose DEF.OBJ’ *ulgulu-wo

The evidence from activity here, then, is that /o/ and /ɔ/ must have an active, therefore contrastive,
feature that causes rounding. [round] (or [labial]) is an obvious candidate. These vowels alternate with
/ə/ and /a/, the only vowels that undergo rounding, suggesting they are contrastively (non-round). As
with [RTR], the restriction of the contrastive feature [±round] to these particular vowels is not obvious
from the phonetics, because /u/ and /ʊ/ are also phonetically {[round]}; however, there is no evidence
that that they have an active [round] feature.

4.3. Palatalization

The front vowels /i/, /e/ and /ɛ/ cause palatalization of a preceding /s/, shown in (14); this suggests
that they have a contrastive triggering feature we will call [front] (or [coronal]).

(14) Palatalization of /s/ to [ʃ] before front vowels (Zhang 1996: 171–172)
a. [s] before non-front vowels b. [ʃ] before front vowels
sukə [suxə] ‘axe’ asi [aʃi] ‘now’
sʊnta [sʊnta] ‘deep’ sɛɛn [ʃɛɛn] ‘ear’
sɔkɔ- [sɔxɔ] ‘fill’
sələ [sələ] ‘iron’
sarbʊ [sarbʊ] ‘chopsticks’

4.4. Height contrast

We still need to distinguish /ə/ ~ /u/, /a/~ /ʊ/, and /e/ ~ /i/. The alternations we have seen, /ə/ ~ /a/
~ /o/~ /ɔ/ and /u/ ~ /ʊ/ (to which we can add /e/ ~ /ɛ/), suggest that there is only one height contrast,
which we can designate [±low] ([±high] would also serve here, depending on whether we consider the
low or high vowels as marked).

4.5. Xunke Oroqen contrastive features

Putting together the evidence of phonological activity surveyed to here, we need to arrive at a
feature hierarchy that yields the required values. Zhang (1996) proposes the hierarchy: [low] >

6
On rounding harmony in Oroqen I follow Zhang 1995, 1996, Zhang & Dresher 1996, Dresher & Zhang 2005,
and Dresher & Nevins 2017; see Li 1996 and Walker 2014 for a different view. On the restriction that rounding
harmony is not triggered by a single initial round vowel, see Zhang 1996, Zhang & Dresher 1996, Walker 2001,
and Dresher & Nevins 2017.
 25

[coronal] > [labial] > [RTR]. I adopt this analysis, substituting [front] for [coronal] and [round] for
[labial]. This feature hierarchy, applied to the vowels in (7), yields the tree diagram in (15).

(15) Xunke Oroqen contrastive hierarchy: [low] > [front] > [round] > [RTR]

qp
[low] (non-low)
qp ru
[front] (non-front) [front] (non-front)
ru wo g ru
[RTR] (non-RTR) [round] (non-round) /i/ [RTR] (non-RTR)
g g ru ru g g
/ɛ/ /e/ [RTR] (non-RTR) [RTR] (non-RTR) /ʊ/ /u/
g g g g
/ɔ/ /o/ /a/ /ə/

The feature specifications in (15) satisfy the requirement that every active feature be contrastive.
Every vowel except /i/ is specified for [±RTR], thereby accounting for why only /i/ does not
participate in RTR harmony. Only vowels with contrastive [±round] participate in labial harmony, and
only vowels specified [front] cause palatalization.
To sum up, we have been able to give an account of the vowel phonology of Oroqen that is
consistent with the Contrastivist Hypothesis: all the active features are contrastive. Moreover, this
analysis explains why certain vowels participate in certain processes and others do not, in ways that
are not obvious from their phonetic descriptions.

5. Contrastive feature hierarchies in morphosyntax

In a series of papers, Elizabeth Cowper and Daniel Currie Hall have shown that language-
particular contrastive feature hierarchies are not limited to phonology, but also play an important role
in morphosyntax (Cowper & Hall 2013, 2014a, 2014b, 2017). Here I will review two of their
examples: number features in English, and non-present tenses in Bamileke-Dschang.

5.1. English number features

English nouns are classified as being either singular (as in this book, a book), plural (these books,
books, some books) or mass (this mud, mud, some mud). As shown in (16a), singular and mass nouns
share the demonstrative this against the plural, which takes these. But mass nouns and plurals share the
null article and some against the singular, which requires a (16b).

(16) Syncretisms in the English number system

SINGULAR MASS PLURAL
a. tthis book this mud these books
b. a book tØ/some mud Ø/some bookst

Cowper & Hall (2017) show that these syncretisms can be captured better with a contrastive
hierarchy than with feature geometry. They propose that the feature [number] in English is sub-divided
by two features: [atomic], which marks singular, and [discrete], for plural. These features are ordered
[atomic] > [discrete], as in (17).
26

(17) English number hierarchy

qp
[atomic] a, an (non-atomic) Ø, some
SINGULAR ei
this, -Ø (non-discrete) [discrete]
MASS PLURAL
this, -Ø these, those, -s

The first split is between [atomic] and (non-atomic). [atomic] marks singular nouns and is spelled
out by the determiner a(n); (non-atomic) is the default, spelled out by Ø or some. (non-atomic) is
divided by [discrete], which marks plural, spelled out by these, those, and the plural suffix -s. Nouns
that are not [discrete] default to this and -Ø.
Cowper & Hall (2017) point out that the expansion of the unmarked feature (non-atomic) in (17)
would not be possible if the feature [atomic] were privative, as there would be no unmarked feature to
expand. In a privative system [non-atomic] would have to be treated as the marked value, to allow for
the further expansion of that node. They argue that such an analysis is less adequate and intuitive than
the one in (17).

5.2. Bamileke temporal proximity features

The Dschang dialect of Bamileke (Grassfields Bantu) distinguishes five degrees of proximity that
cross-classify with a distinction between past and future, giving ten non-present tenses: P1–P5 and F1–
F5 (Hyman 1980, citing work by Maurice Tadadjeu). These tenses are shown in Table 2:

DEGREE INTERPRETATION: PAST INTERPRETATION: FUTURE

1 ‘immediate past’ ‘immediate future’
2 ‘same day’ ‘same day’
3 ‘previous day’ ‘next day’
4 ‘several days ago’ ‘several days in the future’
5 ‘distant past’ (a year or more) ‘distant future’ (a year or more)
Table 2: Bamileke-Dschang non-present tenses (Hyman 1980)

Cowper & Hall (2017) propose that the five degrees of temporal proximity can be specified using
three features, which create the hierarchy shown in (18):

(18) Bamileke temporal proximity feature hierarchy: [same day] > [near] > [far]
[past/future]
qp
[same day] (non-same day)
ei qp
[near] (non-near) [near] (non-near)
1 2 3 wo
‘immediate ‘same day’ ‘previous/ (non-far) [far]
pst/fut’ next day’ 4 5
‘several days ‘distant
ago/in the fut’ pst/fut’

The first feature is [±same day]. The positive value is divided by [±near]: the most proximate time
is [same day, near], which is degree 1; [same day, non-near] is degree 2. The same feature [±near] also
divides (non-same day). The closest in time one can be if not on the same day is the previous day or
the next day (degree 3). The (non-near) branch is divided by a new feature, [±far]. (non-far) is degree
4 (not the same day, and not the previous or next day), and [far] is the distant past or future.
 27

Cowper & Hall (2017) point out that both the applicability and the interpretation of a feature
depends on its place in the hierarchy. Thus, the feature [±far] is contrastive only for events that are
(non-same day, non-near). It does not apply to [same day] or [near] events. And the feature [±near] is
interpreted differently depending on whether it applies to [same day] or (non-same day) events. Here,
too, it is crucial to allow both marked and unmarked feature values to be expanded. Contrastive feature
hierarchies thus allow for elegant and economical representations of the ten non-present tenses.

6. Conclusion

It is a recurring intuition of phonological theorists that the phonological systems of the world’s
languages use a very limited set of features (Jakobson, Fant & Halle 1952; Chomsky & Halle 1968). I
have argued here that this is not because there is a limited set of innate universal features; the
impression that all languages use the same substantive features is to some extent an illusion. Rather, it
is because Universal Grammar requires speakers to construct contrastive feature hierarchies, and they
are what limit the number of features available to the phonology of a given language. And contrastive
feature hierarchies may not be limited to phonology, but play a role in other components of language,
such as morphosyntax, and perhaps in other cognitive domains as well (Nevins 2015).
In the words of Jakobson, Fant & Halle (1952: 9): “The dichotomous scale [= the contrastive
feature hierarchy/BED] is the pivotal principle of the linguistic structure. The code imposes it upon the
sound.”

References
Archangeli, Diana & Douglas Pulleyblank. 1994. Grounded phonology. Cambridge, MA: MIT Press.
Cherry, E. Colin, Morris Halle & Roman Jakobson. 1953. Toward the logical description of languages in their
phonemic aspect. Language 29(1): 34–46.
Chomsky, Noam & Morris Halle. 1968. The sound pattern of English. New York, NY: Harper & Row.
Clements, G. N. 2001. Representational economy in constraint-based phonology. In T. Alan Hall, (ed.),
Distinctive feature theory, 71–146. Berlin: Mouton de Gruyter.
Clements, G. N. 2003. Feature economy in sound systems. Phonology 20(3): 287–333.
Clements, G. N. 2009. The role of features in speech sound inventories. In Eric Raimy & Charles E. Cairns (eds.),
Contemporary views on architecture and representations in phonological theory, 19–68. Cambridge, MA:
MIT Press.
Clements, G. N. & Rachid Ridouane (eds.). 2011. Where do features come from? Cognitive, physical and
developmental bases of distinctive speech categories. Amsterdam: John Benjamins.
Compton, Richard & B. Elan Dresher. 2011. Palatalization and ‘strong i’ across Inuit dialects. Canadian Journal
of Linguistics/Revue canadienne de linguistique 56(2): 203–228.
Cowper, Elizabeth & Daniel Currie Hall. 2013. Syntactic change and the cartography of syntactic structures. In
Stefan Keine & Shayne Sloggett (eds.), NELS 42: Proceedings of the Forty-Second Annual Meeting of the
North East Linguistic Society, 129–140. Amherst, MA: GLSA.
Cowper, Elizabeth & Daniel Currie Hall. 2014a. The features and exponence of nominal number. Lingue e
Linguaggio 13(1): 63–82.
Cowper, Elizabeth & Daniel Currie Hall. 2014b. Reductiō ad discrīmen: Where features come from. Nordlyd
41(2): 145–164.
Cowper, Elizabeth & Daniel Currie Hall. 2017. Scope variation in contrastive hierarchies of morphosyntactic
features. Presented at Georgetown University Round Table, March 2017.
Dresher, B. Elan. 1998. On contrast and redundancy. Presented at the annual meeting of the Canadian Linguistic
Association, Ottawa.
Dresher, B. Elan. 2003. Contrast and asymmetries in inventories. In Anna-Maria di Sciullo (ed.), Asymmetry in
grammar, volume 2: Morphology, phonology, acquisition, 239–257. Amsterdam: John Benjamins.
Dresher, B. Elan. 2009. The contrastive hierarchy in phonology. Cambridge: Cambridge University Press.
Dresher, B. Elan. 2014. Contrastive hierarchies in Element Theory. Presented at the Conference on Theoretical
Issues in Contemporary Phonology: Reading Tobias Scheer, École des Hautes Études en Sciences Sociales
(EHESS), Paris, February 2014. https://docs.google.com/viewer?a=v&pid=sites&srcid=ZGVmYXVsdGR
vbWFpbnxyZWFkZWluZ3RvYmlhc3NjaGVlcmVuZ2xpc2h8Z3g6NmRkZWRhMTc5ZmVkMDk2ZA. Last
accessed June 25, 2017.
28

Dresher, B. Elan. 2015. The motivation for contrastive feature hierarchies in phonology. Linguistic Variation
15(1): 1–40.
Dresher, B. Elan. 2016. Contrast in phonology 1867–1967: History and development. Annual Review of
Linguistics 2: 53–73.
Dresher, B. Elan & Andrew Nevins. 2017. Conditions on iterative rounding harmony in Oroqen. Transactions of
the Philological Society 115(3): 365–394.
Dresher, B. Elan, Glyne L. Piggott & Keren Rice. 1994. Contrast in phonology: Overview. Toronto Working
Papers in Linguistics 13(1): iii-xvii.
Dresher, B. Elan & Keren Rice. 2007. Markedness and the contrastive hierarchy in phonology. http://homes.chass.
utoronto.ca/~contrast/. Last accessed June 25, 2017.
Dresher, B. Elan & Xi Zhang. 2005. Contrast and phonological activity in Manchu vowel systems. Canadian
Journal of Linguistics/Revue canadienne de linguistique 50(1/2/3/4): 45–82.
Dyck, Carrie. 1995. Constraining the phonology-phonetics interface, with exemplification from Spanish and
Italian dialects. Doctoral dissertation, University of Toronto.
Flemming, Edward S. 2002. Auditory representations in phonology. London: Routledge.
Hall, Daniel Currie. 2007. The role and representation of contrast in phonological theory. Doctoral dissertation,
University of Toronto.
Hall, Daniel Currie. 2011. Phonological contrast and its phonetic enhancement: Dispersedness without dispersion.
Phonology 28(1): 1–54.
Hall, Daniel Currie. 2017. Contrastive specification in phonology. In Mark Aronoff (ed.), Oxford Research
Encyclopedia of Linguistics. Oxford: Oxford University Press. DOI: 10.1093/acrefore/9780199384655.013.
26. Last accessed June 25, 2017.
Hall, Daniel Currie & Kathleen Currie Hall. 2016. Marginal contrasts and the Contrastivist Hypothesis. Glossa
1(1): 50. 1–23.
Halle, Morris. 1959. The sound pattern of Russian: A linguistic and acoustical investigation. The Hague: Mouton.
Second printing, 1971.
Harms, Robert T. 1968. Introduction to phonological theory. Englewood Cliffs, NJ: Prentice-Hall.
Hu, Zengyi. 1986. Elunchunyu jianzhi [A grammatical sketch of Oroqen]. Beijing: Minzu Chubanshe.
Hulst, Harry van der. 1993. Units in the analysis of signs. Phonology 10(2): 209–241.
Hyman, Larry M. 1980. Relative time reference in the Bamileke tense system. Studies in African Linguistics
11(2): 227–237.
Hyman, Larry M. 2011. Do tones have features? In John A. Goldsmith, Elizabeth Hume & W. Leo Wetzels (eds.),
Tones and features, 50–80. Berlin: Mouton De Gruyter.
Jakobson, Roman. 1941. Kindersprache, Aphasie, und allgemeine Lautgesetze. Uppsala: Uppsala Universitets
Årsskrift. Translated by A. R. Keiler as Child language, aphasia, and phonological universals. The Hague:
Mouton, 1968.
Jakobson, Roman. 1962 [1931]. Phonemic notes on Standard Slovak. In Selected writings I. Phonological studies,
221–230. The Hague: Mouton. [In Czech in Slovenská miscellanea (Studies presented to Albert Pražak).
Bratislava, 1931.]
Jakobson, Roman, C. Gunnar M. Fant & Morris Halle. 1952. Preliminaries to speech analysis. MIT Acoustics
Laboratory, Technical Report, No. 13. Reissued by MIT Press, Cambridge, Mass., Eleventh Printing, 1976.
Jakobson, Roman & Morris Halle. 1956. Fundamentals of language. The Hague: Mouton.
Jakobson, Roman & John Lotz. 1949. Notes on the French phonemic pattern. Word 5(2): 151–158.
Keyser, Samuel Jay & Kenneth N. Stevens. 2001. Enhancement revisited. In Michael J. Kenstowicz (ed.), Ken
Hale: A life in language, 271–291. Cambridge, Mass.: MIT Press.
Keyser, Samuel Jay & Kenneth N. Stevens. 2006. Enhancement and overlap in the speech chain. Language 82(1):
33–63.
Ko, Seongyeon. 2012. Tongue root harmony and vowel contrast in Northeast Asian languages. Doctoral
dissertation, Cornell University, Ithaca, N.Y.
Krekoski, Ross. 2016. Contrast and complexity in Chinese tonal systems. Doctoral dissertation, University of
Toronto.
Li, Bing. 1996. Tungusic vowel harmony. The Hague: Holland Academic Graphics.
Liljencrants, Johan & Björn Lindblom. 1972. Numerical simulation of vowel quality systems: The role of
perceptual contrast. Language 48(4): 839–862.
Lindblom, Björn. 1986. Phonetic universals in vowel systems. In John J. Ohala & Jeri J. Jaeger (eds.),
Experimental phonology, 13–44. New York: Academic Press.
Mackenzie, Sara. 2011. Contrast and the evaluation of similarity: Evidence from consonant harmony. Lingua
121(8): 1401–1423.
 29

Mackenzie, Sara. 2013. Laryngeal co-occurrence restrictions in Aymara: Contrastive representations and
constraint interaction. Phonology 30(2): 297–345.
Mielke, Jeff. 2008. The emergence of distinctive features. Oxford: Oxford University Press.
Nevins, Andrew. 2015. Triumphs and limits of the Contrastivity-Only Hypothesis. Linguistic Variation 15(1): 41–
68.
Oxford, Will. 2015. Patterns of contrast in phonological change: Evidence from Algonquian vowel systems.
Language 91(2): 308–357.
Purnell, Thomas & Eric Raimy. 2015. Distinctive features, levels of representation and historical phonology. In
Patrick Honeybone & Joseph Salmons (eds.), The handbook of historical phonology, 522–544. Oxford:
Oxford University Press.
Rice, Keren. 2003. Featural markedness in phonology: Variation. In Lisa Cheng & Rint Sybesma (eds.), The
second GLOT International state-of-the-article book: The latest in linguistics, 387–427. Berlin: Mouton de
Gruyter.
Rice, Keren. 2007. Markedness in phonology. In Paul de Lacy (ed.), The Cambridge handbook of phonology, 79–
97. Cambridge: Cambridge University Press.
Samuels, Bridget D. 2011. Phonological architecture: A biolinguistic perspective. Oxford: Oxford University
Press.
Sandler, Wendy. 1993. Sign language and modularity. Lingua 89(4): 315–351.
Sapir, Edward. 1925. Sound patterns in language. Language 1(2): 37–51.
Scheer, Tobias. 2010. How to marry (structuralist) contrast and (generative) processing (review of Dresher 2009).
Lingua 120(10): 2522–2534.
Spahr, Christopher. 2014. A contrastive hierarchical account of positional neutralization. The Linguistic Review
31(3–4): 551–585.
Spahr, Christopher. 2016. Contrastive representations in non-segmental phonology. Doctoral dissertation,
University of Toronto.
Stevens, Kenneth N. & Samuel Jay Keyser. 1989. Primary features and their enhancement in consonants.
Language 65(1): 81–106.
Stevens, Kenneth N., Samuel Jay Keyser & Haruko Kawasaki. 1986. Toward a phonetic and phonological theory
of redundant features. In Joseph S. Perkell & Dennis H. Klatt (eds.), Symposium on invariance and
variability of speech processes, 432–469. Hillsdale, NJ: Lawrence Erlbaum.
Trubetzkoy, N. S. 1939. Grundzüge der Phonologie. Göttingen: Vandenhoek & Ruprecht.
Walker, Rachel. 2001. Round licensing and bisyllabic triggers in Altaic. Natural Language and Linguistic Theory
19(4): 827–878.
Walker, Rachel. 2014. Nonlocal trigger-target relations. Linguistic Inquiry 45(3): 501–523.
Zhang, Xi. 1995. Vowel harmony in Oroqen (Tungus). Toronto Working Papers in Linguistics 14: 161–174.
Zhang, Xi. 1996. Vowel systems of the Manchu-Tungus languages of China. Doctoral dissertation, University of
Toronto.
Zhang, Xi & B. Elan Dresher. 1996. Labial harmony in written Manchu. Saksaha: A Review of Manchu Studies 1:
13–24.
Zhang, Yan-Chang, Bing Li & Xi Zhang. 1989. The Oroqen language. Changchun: Jilin University Press.
Proceedings of the 35th West Coast
Conference on Formal Linguistics
edited by Wm. G. Bennett, Lindsay Hracs,
and Dennis Ryan Storoshenko
Cascadilla Proceedings Project Somerville, MA 2018

Copyright information
Proceedings of the 35th West Coast Conference on Formal Linguistics
© 2018 Cascadilla Proceedings Project, Somerville, MA. All rights reserved
ISBN 978-1-57473-472-0 library binding

A copyright notice for each paper is located at the bottom of the first page of the paper.
Reprints for course packs can be authorized by Cascadilla Proceedings Project.

Ordering information
Orders for the library binding edition are handled by Cascadilla Press.
To place an order, go to www.lingref.com or contact:

Cascadilla Press, P.O. Box 440355, Somerville, MA 02144, USA

phone: 1-617-776-2370, fax: 1-617-776-2271, [email protected]

Web access and citation information

This entire proceedings can also be viewed on the web at www.lingref.com. Each paper has a unique document #
which can be added to citations to facilitate access. The document # should not replace the full citation.

This paper can be cited as:

Dresher, B. Elan. 2018. Contrastive Hierarchy Theory and the Nature of Features. In Proceedings of the 35th West
Coast Conference on Formal Linguistics, ed. Wm. G. Bennett et al., 18-29. Somerville, MA: Cascadilla
Proceedings Project. www.lingref.com, document #3371.