LINGUIST List 12.971

Fri Apr 6 2001

Review: Geometry and Features of Tone

Editor for this issue: Terence Langendoen <>

What follows is another discussion note contributed to our Book Discussion Forum. We expect these discussions to be informal and interactive; and the author of the book discussed is cordially invited to join in. If you are interested in leading a book discussion, look for books announced on LINGUIST as "available for discussion." (This means that the publisher has sent us a review copy.) Then contact Simin Karimi at or Terry Langendoen at


  1. tae-jin yoon, Review of The Geometry and Features of Tone by Snider (1999)

Message 1: Review of The Geometry and Features of Tone by Snider (1999)

Date: Fri, 6 Apr 2001 16:45:05 -0500 (CDT)
From: tae-jin yoon <>
Subject: Review of The Geometry and Features of Tone by Snider (1999)

Snider, Keith (1999) The Geometry and Features of Tone, Summer Institute
of Linguistics and The University of Texas at Arlington, 173 pp., $ 25.00,
Publications in Linguistics 133.

Tae-Jin Yoon, University of Illinois at Urbana-Champaign

1. Introduction

'The Geometry and Features of Tone' by Snider (1999) presents a model for
the phonological representation of tone, called Register Tier Theory
(henceforth, RTT). A tone is a specific pitch levels and/or contours to
mark lexical contrast. And tone has been treated and represented as a
feature independent of the segmental features which constitute consonants
and vowels. In addition to the tone features, there are register tones
which are relatively high or low set of lexical tones. Thus RTT is so-
called because, among other things, upper register tones and lower
register tone occupy separate tiers from the tone tiers such as high (H)
and low tone (L) in the geometrically organized phonological

The proposal is said to be guided by the following fundamental criteria
for the model for the tonal presentation (pp. xii-xiii): (1) a model
needs to be able to uniquely represent each tone phoneme in any given
language, (2) it needs to account for the different types of tonal
alternation found in natural language, (3) it needs to be able to
adequately characterize the different types of contour tones found in
natural language, and finally (4) the proposed model needs to be able to
exclude from formal descriptions phenomena that never occur in natural
language. In sum, an adequate model for the phonological representation
of tone must not only be expressive enough to describe all the tonal
features and alternations found in natural language, but also be
restricted enough to exclude unattested phenomena from description.

Of all things, it seems that what is underlying in the development of RTT
is to account for one of the natural tonal phenomena called 'downstep'.
Downstep refers to the lowering in pitch of the high or low tones
relative to the preceding high or low tones. Since the process is quite
natural in tone languages, there have been some controversial issues in
treating the downstep. Some (eg. Pierrehumbert 1980) argue that the pitch
lowering can be accounted for in terms of phonetic implementation, while
others (Inkelas 1989) including Snider (1999) argue that it can better be
accounted for in terms of a phonological representation.

As for the intended readers, the book is accessible by anyone who has a
minimal background of generative phonology, as well as by linguists who
have interest in tone and who are working in feature geometry. Especially,
this book seems to be very helpful to those who have had the chance to
know a bit of what the autosegmental phonology is and how it is developed
from the research on suprasegments such as tone, stress and intonation,
but who have not been introduced to the upstep or downstep phenomena.

2. Synopsis

Broadly speaking, the book consists of four parts: (1) an overview of
autosegmental phonology, (2) the conceptual modeling and predictions of
Register Tier Theory (RTT), (3) the application of the proposed model to
a number of tone languages, and (4) some critique of alternative

Chapter 1 presents a brief overview of autosegmental theory, and
demonstrates some advantages by representing tone in terms of
autosegmental phonology. Also included are the basic concepts and
properties of the Theory of Lexical Phonology (Kiparsky 1982).

Chapters 2 and 3 present the RTT model and discuss what predictions the
model can make. What is distinguished from other standardly assumed
version of feature geometry is that Snider (1999) incorporates register
tier into the geometry of tone, in addition to the generally assumed
tonal tier. To the extent that the model incorporates register tier into
the presentation, it is somewhat similar to the model of Yip (1988).
However, the two models differ in that in the Snider's model, the
register tier and tone tier are independently linked to the tone bearing
unit, whereas in Yip's model, the tone tier is dependent on the register
tier. After modeling the phonological representation of tones, he
presents assimilation, upstep and downstep, contour tones (such as
falling tones and rising tones), and multiple tone height as some of the
predictions the model can make.

In Chapters 4 through 7, the RTT model is applied to a number of
languages such as Chumburung (a Kwa language spoken in Ghana), Engenni (a
Kwa language spoken in Nigeria), Acatlan Mixtec (a dialect of Otomanguean
language spoken in Mexico), and Bamilek?Dschang (a Bantu language spoken
in Cameroon). Among the languages, Acatlan Mixtec manifests the upstep
phenomenon, while others show downstep cases.

Finally, Chapter 8 concludes with critique of some alternative approaches.
The author argues that the previous work either fails to provide a way of
changing a tonal register independent of the tones, or were unable to
account for the cumulative nature of successive downsteps and upstep due
to the lack of the relative value of register features.

3. Critical evaluation

Overall, the treatment of the features and representation of tone by
Snider is coherent and illustrative. The author's providing definitions
of almost all technical terms, and his presenting derivations of
phonological operations were very helpful to understand the properties of
tonal phenomena. Especially the diagrams provided for the phonetic
implementation of the derived phonological representation were

Nevertheless, some assumptions made in the modeling of the tonal
representation seem to raise some questions. Two of such questions are
concerned with the definition of the register features and the
interpretation of the last Lo tone in the Hi-Lo-Hi-Lo sequence. (Hi
refers to High tone; Lo, Low, hereafter.)

The first question is concerned with the definition of the register
features h and l. The acoustic and auditory properties of tone are
plausibly emphasized to be relative in nature (p. 24), thus Snider
defines the register feature h as higher relative to the preceding
register setting, and the register feature l as lower relative to the
preceding register setting. Given the definition, Snider seems to be
concerned with the question of how the register feature of the first tone
bearing unit in an utterance is interpreted, which has no preceding tone
bearing unit. Thus, with regard to this question, he assumes that native
speakers begin all utterances with reference point in mind (p. 25). What
I am curious about are: what kind of mental representation can be
constructed which serves as the reference point? And, how can the
listener access the speaker's mentally constructed reference point?

The second question is concerned with the interpretation of the pitch of
the tone bearing unit which follows downstepped Hi that is followed by Lo.
That is, in Hi-Lo-Hi-Lo sequence, the interpretation of the second Lo
seems somewhat arbitrary. Let us first consider a typical downstep
situation. The typical downstep occurs when the second Hi of a Hi-Lo-Hi
sequence is downstepped relative to the first Hi. Snider describes the
typical downstep as in (1), where o indicates tonal root node tier, H and
L indicate tonal features, and h and l indicate register features:

(1) a. H L H b. H L H c. H L H
 | | | | | | | | |
 o o o o o o o o o
 | | | | | / | | | /
 h l h h l h h l h

In input (1.a), Hi is linked to H tone feature and h register feature and
Lo is linked to L tone feature and l register feature. By spreading l
register feature of the Lo to the following Hi in (1.b), the second Hi
loses its h register feature, as the unlinked h in (1.c) shows. As a
result, while the first Hi is linked to the original H tone feature and h
register feature, the second Hi is linked to the H tone feature and a
newly linked l register feature. Thus, in Hi-Lo-Hi sequence, the
spreading of the l register feature by Lo and the delinking of the h
register feature from the following Hi can be interpreted as the typical
downstep phenomenon. Consequently, the RTT model can represent the
downstep phenomenon in a coherent and illustrative way.

However, when Lo is followed by the sequence where the typical downstep
occurs, as in the Hi-Lo-Hi-Lo sequence, Snider interprets the Lo as an
automatically downstepped situation. The representation of the Hi-Lo-Hi-
Lo sequence is in (2).

(2) a. H L H L b. H L H L c. H L H L
 | | | | | | | | | | | |
 o o o o o o o o o o o o
 | | | | | | / | | | | / |
 h l h l h l h l h l h l

In the input (2.a), the sequence Hi-Lo-Hi-Lo is represented. As in (1.b-
c), the l register feature of the first Lo spreads to the following Hi,
and the h register feature of the second Hi loses its association line.
The natural interpretation of the representation in (2.c) is that on one
hand, the first Hi tone is the highest, and the second Hi tone is
relatively lower than the first Hi, and on the other hand, the first low
tone is relatively lower than the surrounding High tone, and the second
low tone is the same as the first low tone. However, Sinder stipulates
that the second Low tone must be interpreted as automatically downstepped.
Even though he provides some explanations with regard to the
interpretation, they seem to be somewhat unconvincing.

Clements, G.N. 1996. Review of "The phonology of tone: The representation
 of tone register, edited by Harry Van der Hulst and Keith Snider,"
 Language 72.4:847-52.
Fromkin, V. 1978. Tone: A Linguistic Survey. New York: Academic Press.
Goldsmith, John. 1976. An Overview of Autosegmental Phonology. Linguistic
 Analysis 2: 23-68.
Inkelas, S. 1989. Register tone and the phonological representation of
 downstep. In Laurie Tuller and Isabelle Haik (eds.), Current
 approaches to African linguistics 6: 65-82. Dordrecht: Foris.
Kiparsky, P. 1982. Lexical phonology and morphology. In I.S. Yang (ed.),
 Linguistics in the morning calm, 3-91. Seoul: Hanshin.
Pierrehumbert, J. 1980. The phonology and phonetics of English intonation.
 Ph.D. dissertation, MIT.

Tae-Jin Yoon is a graduate student in the Department of Linguistics at
the University of Illinois at Urbana-Champaign (UIUC). He has completed
his M.A. at the University of Seoul, Korea, in the Spring 2000 and began
to pursue his Ph.D. at the UIUC in the Fall 2000. His research interest
lies in phonetics and phonology, as well as other aspects of both applied
and theoretical (both functional and formal) linguistics.
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue