EDITORS: Clements, G. Nick and Ridouane, Rachid
TITLE: Where Do Phonological Features Come From?
SUBTITLE: Cognitive, Physical and Developmental Bases of Distinctive Speech
SERIES: Language Faculty and Beyond 6
PUBLISHER: John Benjamins
André Zampaulo, Department of Spanish and Portuguese, The Ohio State University
The edited volume Where Do Phonological Features Come From? presents first an
obituary note on the first editor, phonologist G. Nick Clements, and his
outstanding contributions to research on phonology and the phonetics-phonology
interface. The book is subsequently divided into five parts: Introduction,
General and Cognitive Issues, Acoustic and Articulatory Bases of Features,
Extracting Features from the Signal and Features in Phonological Development.
Part I of Where Do Phonological Features Come From? offers the editors’ general
overview of the contents of the volume. Since the early groundbreaking works of
Jakobson, Fant & Halle (1952) and Chomsky & Halle (1968), research on feature
theory has centered on discovering the intrinsic characteristics of sounds that
enable speakers to establish phonological contrasts that convey distinct
meanings on the lexical, morphological and grammatical levels. The present
collection of papers contributes to this search by addressing the mental status
of features, their role in speech production and perception, how their physical
properties may be measured and what part they play in the development of
language. Thus, the answer(s) to the question of ‘where do phonological features
come from?’ speak(s) to the core of how they arise in human oral communication
and how speakers cognitively extract, organize and implement them as units
within sounds. While these issues are explored from different standpoints --
ranging from general linguistics to phonetics and speech sciences and language
acquisition --, the editors are adamant in establishing the principal goal of
the volume, which is to provide scholars with the most current state of the
research on feature theory.
Part II focuses on the general and cognitive issues of distinctive features,
going beyond their traditional roles in distinguishing words or defining natural
classes of sound patterns. Current investigation of the cognitive status of
features offers evidence contrary to the hypothesis that they are stored in
speakers’ brains as part of a Universal Grammar; rather, the papers in this part
of the volume argue that features are learned and language-specific.
Abigail C. Cohn investigates the conflict between those two viewpoints in the
first paper, “Features, segments, and the sources of phonological primitives,”
by examining their arguments regarding issues such as the nature of phonological
primitives, their phonetic implementation, their occurrence in all human
languages and their relation to language acquisition. The author reviews work on
distinctive features within classic generative phonology and concludes that it
provides imprecise results. Evidence from language-specific phonetics
demonstrates certain similarities in the phonetic correlates of features across
languages, but at the same time, it also reveals that features do not
necessarily establish the same phonological categories cross-linguistically. As
for the role of features in language acquisition, Cohn maintains that speakers
can learn and establish categorical systems out of more gradient ones.
In the second paper, “Feature economy in natural, random, and synthetic
inventories,” Scott Mackie and Jeff Mielke test Clement’s (2003) notion of
feature economy (the maximization of the ratio of sounds in an inventory to the
features needed to define them) in a large-scale investigation using P-base, a
database of inventories and sound patterns of 549 languages. Although the
results support Clement’s concept of feature economy by suggesting that natural
inventories are indeed more economical, they also reveal that features are not
necessary for an inventory to incur feature economy effects, as de Boer’s (2001)
agent-based simulations of vowel inventories without features are also shown to
be at least as economical as natural vowel inventories.
Part III features three papers which investigate the acoustic and articulatory
foundations of distinctive features and present a reflection upon how these
phonological units can be abstracted from physical properties.
The third paper, entitled “Sound systems are shaped by their users: The
recombination of phonetic substance,” by Björn Lindblom, Randy Diehl, Sang-Hoon
Park and Giampiero Salvi, extends previous research carried out under the
framework of dispersion theory, according to which languages tend to favor
phoneme inventories that maximize acoustic distinctiveness while minimizing
articulatory effort. The authors set out to test the explanatory adequacy of
this framework regarding the preference of languages for labial, dental/alveolar
and velar places of articulation. Through computational experiments centered on
the place of articulation of stop+vowel syllables from diverse language
inventories, these authors are able to determine the perceptual cost, the
articulatory cost and the mode of learning of features. Subsequently, they apply
such constraints to the observed phonetics of stop+vowel inventories and argue
that phonological facts are better explained through user-based accounts rather
than abstract ones.
Hyunsoon Kim presents acoustic, articulatory and aerodynamic data to investigate
the phonetic implementation of distinctive features in Korean lenis and fortis
fricatives /s, s’/ in the fourth paper, “What features underline the /s/ vs.
/s’/ contrast in Korean? Phonetic and phonological evidence.” The author argues
that these two voiceless segments should be specified for the feature [-spread
glottis] regarding the opening of the glottis, while they are different from
each other as far as the primary articulator of the tongue blade, i.e. while
lenis /s/ is [-tense], fortis /s’/ is [+tense].
The fifth paper, “Automaticity vs. feature-enhancement in the control of
segmental F0,” by Phil Hoole and Kiyoshi Honda, offers a fine-grained
examination of enhancement theory (Stevens & Keyser 2010), which argues that
some features can be enhanced by others that are not necessarily phonetically
related to them. The authors analyze electromyographic data from the
cricothyroid muscle in order to investigate the role played by F0 in the
features of voicing and vowel height. Regarding the former, results indicate
that the articulation of consonants is what determines the different patterns of
F0, and the articulation of vowels establishes their different pitch patterns.
Thus, speakers find themselves in control of choosing whether or not to enhance
features, which suggests that these do not represent abstract immutable entities.
The sixth, seventh and eighth papers, which correspond to Part IV of the volume,
are concerned with the extraction of features from the signal and the subsequent
categorization of sounds.
Diana Archangeli, Adam Baker and Jeff Mielke present three studies regarding the
articulation and perception of American English /ɹ/ in the sixth paper,
“Categorization and features: Evidence from American English /ɹ/.” While the
first study explores the different articulations of this sound, the second study
assesses its perception by native speakers, and the third evaluates the
co-articulatory effects in /str/ clusters. The results indicate that speakers
are inclined to categorize sounds and extract sound patterns from the signal,
though the latter might not always be clear or consistent. The authors also
argue that these results support the hypothesis that features are learned and
emergent in the inventory of speakers and rely upon how these parse acoustic
data. Thus, this paper presents evidence against the idea that features are
Bob McMurray, Jennifer S. Cole and Cheyenne Munson contribute the seventh paper,
“Features as an emergent product of computing perceptual cues relative to
expectations.” This paper, too, is concerned with how speakers are able to
extract distinctive features from the signal. The authors put forth a model that
builds upon earlier proposals (e.g. Fowler 1984; Gow 2003) and make an argument
for the idea that the listener, by parsing the data to which s/he is exposed, is
then able to filter out the varying character of sound articulation and to
utilize it as relevant information. The authors exemplify this model with a case
study of vowel-to-vowel co-articulation (V-to-V), which shows that, by parsing
the variable acoustic formant measures of vowels, speakers are able to correctly
identify vowels and most accurately predict the vowel to appear in the next
In the eighth paper, entitled “Features are phonological transforms of natural
boundaries,” Willy Serniclaes presents a different view of how speakers
categorize sounds and extract distinctive features. The author argues that
features are not to be identified at the stable state of the signal, but rather
at the boundaries between sounds, such as in vowel formant transitions or stop
bursts. Features represent then contrastive units found in the differences
between categories. The author supports this approach by providing
psychoacoustic evidence for the enhanced auditory sensitivity that characterizes
these boundary regions. Moreover, he demonstrates that the perception of vowel
and consonant place-of-articulation distinctions coincide after rotation of the
acoustic space. This indicates that speakers' place-of-articulation perception
acts in accordance with a 'radial' representation of the vocal tract, with
psychoacoustic boundaries standing as the central reference point.
The relationship between language acquisition and the study of distinctive
features is reserved for the final section of the volume (Part V), which
includes the last three contributions.
In the ninth paper, “Features in child phonology: Inherent, emergent, or
artefacts of analysis?,” Lise Menn and Marilyn Vihman offer their view on what
the actual role distinctive features play in child phonology. In tune with most
of the previous papers, the authors argue against the notion that features are
innate and universal units present in all children’s minds. According to their
account, features are instead inherent in the sense that they represent a
cohesive system composed of the acoustic-auditory input signal and children’s
cognitive and articulatory capacities. Features are thus phonetically grounded
and emergent during language acquisition, forming part of a child’s mental
grammar as s/he increases her/his use of the language over time.
Alejandrina Cristià, Amanda Seidl and Alexander L. Francis contribute the tenth
paper, “Phonological features in infancy,” in which they carry out experiments
with groups of 7-8-month-old and 14-month-old infants in order to address the
availability of features to the young learner and how these features help shape
human language. Using the Headturn Preference method, the authors find that when
features are grouped into natural classes, they end up helping the learning
experience and the creation of sound patterns. Additionally, children are able
to identify the operating features in their language when they are between 8 and
14 months old. These results provide evidence that distinctive features emerge
from an interaction of cognitive patterns during first language acquisition and
are thus not universally innate.
In the eleventh and final paper, “Acoustic cues to stop-coda voicing contrasts
in the speech of 2-3-year-olds learning American English,” Stefanie
Shattuck-Hufnagel, Katherine Demuth, Helen M. Hanson and Kenneth N. Stevens
approach the variation in the production and perception of cues by children and
the possibly different use of features that they carry out in comparison with
adults. By analyzing two children’s use of acoustic cues to voicing contrast in
word-final stops, the authors find a higher occurrence of epenthetic vowels
after voiced rather than voiceless codas, which indicates that a possible lack
of total gesture control, planning inability and non-adult patterns of enhancing
feature cues by children may still be present in their production of stops, even
though adult listeners are able to filter and recognize them as such.
The compilation of scholarly work in ''Where Do Phonological Features Come From?''
meets the objective that its editors set out, by establishing where the research
on feature theory stands nowadays. Relying on accounts from general linguistics
to speech production and perception and language acquisition, its
multi-perspective nature renders it one of the richest contributions yet for our
understanding of the foundations of distinctive features. For this sole reason,
it surely belongs in every phonologist’s scholarly collection and research
universities’ libraries around the world.
From the pioneering work of classical generative phonology to more recent
publications such as Clements & Hume (1995), Mielke (2008) and Clements (2009),
feature theory has always attempted to offer an explanation for the way sounds
are extracted from the acoustic signal and how their composing units are
organized and stored in the brains of language users, so as to enable
inter-speaker oral communication. The present volume speaks to the core of this
issue. It provides a solid set of groundbreaking papers of which the underlying
thesis builds upon experimental data to reveal that, rather than universally
innate, features are emergent and learned through the course of language
acquisition, in the interplay between sound articulation and the perception of
the acoustic signal.
The excellent scientific quality of the volume's papers reveals different
directions for further research within the field. The role of distinctive
features in sound and language change represents one of the interface areas to
which current research on feature theory can contribute and is nonetheless
absent in this volume. Assuming that sound variation and change is pervasive and
inherent in natural human languages, what actual role do emergent phonological
features play in the constant (re)shaping of sound patterns? If features, as the
units of which sounds are composed, develop from the phonetics of sound
articulation and perception during inter-speaker oral communication, would
‘feature change’ be a more revealing term than ‘sound change’? While research on
feature theory continues to advance, it provides us with better tools to
understand the components of the linguistic knowledge generated and stored in
the brains of language users. The volume under review represents an excellent
guide to such tools and the state of our knowledge on these questions today.
de Boer, Bart. 2001. The origins of vowel systems. Oxford: Oxford University Press.
Clements, G. Nick. 2003. Feature economy in sound systems. Phonology 20:287-333.
Clements, G. Nick. 2009. The role of features in speech sound inventories. In E.
Raimy & C. Cairns (eds.), Contemporary views on architecture and representations
in Phonological Theory, 19-68. Cambridge, MA: MIT Press.
Clements, G. Nick & Elizabeth V. Hume. 1995. The internal organization of speech
sounds. In J. Goldsmith (ed.), The Handbook of Phonological Theory, 245-306.
Cambridge, MA: Blackwell.
Chomsky, Noam & Morris Halle. 1968. The sound pattern of English. Cambridge, MA:
Fowler, Carol A. 1984. Segmentation of coarticulated speech in perception.
Perception & Psychophysics, 36:359-368.
Gow, David W. 2003. Feature parsing: Feature cue mapping in spoken word
recognition. Perception & Psychophysics, 65:575-590.
Jakobson, Roman C., Gunnar M. Fant & Morris Halle. 1952. Preliminaries to speech
analysis: The distinctive features and their correlates. Cambridge, MA: MIT Press.
Mielke, Jeff. 2008. The emergence of distinctive features. Oxford: Oxford
Stevens, Kenneth & Samuel J. Keyser. 2010. Quantal theory, enhancement, and
overlap. Journal of Phonetics 38:10-19.
ABOUT THE REVIEWER