16.1457, Review: Phonology/Morphology/Hispanic Ling: Eddington
LINGUIST List
linguist at linguistlist.org
Mon May 9 09:11:37 UTC 2005
LINGUIST List: Vol-16-1457. Mon May 09 2005. ISSN: 1068 - 4875.
Subject: 16.1457, Review: Phonology/Morphology/Hispanic Ling: Eddington
Moderators: Anthony Aristar, Wayne State U <aristar at linguistlist.org>
Helen Aristar-Dry, Eastern Michigan U <hdry at linguistlist.org>
Reviews (reviews at linguistlist.org)
Sheila Dooley, U of Arizona
Terry Langendoen, U of Arizona
Homepage: http://linguistlist.org/
The LINGUIST List is funded by Eastern Michigan University, Wayne
State University, and donations from subscribers and publishers.
Editor for this issue: Naomi Ogasawara <naomi at linguistlist.org>
================================================================
What follows is a review or discussion note contributed to our
Book Discussion Forum. We expect 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 review." Then contact Sheila Dooley at collberg at linguistlist.org.
===========================Directory==============================
1)
Date: 09-May-2005
From: Matthew Carlson < mtc173 at psu.edu >
Subject: Spanish Phonology and Morphology
-------------------------Message 1 ----------------------------------
Date: Mon, 09 May 2005 05:08:40
From: Matthew Carlson < mtc173 at psu.edu >
Subject: Spanish Phonology and Morphology
AUTHOR: Eddington, David
TITLE: Spanish Phonology and Morphology
SUBTITLE: Experimental and quantitative perspectives
SERIES: Studies in Functional and Structural Linguistics 53
PUBLISHER: John Benjamins
YEAR: 2004
Announced at http://linguistlist.org/issues/16/16-136.html
Matthew T. Carlson, Department of Spanish, Italian, and Portuguese,
Pennsylvania State University
DESCRIPTION
Eddington presents this volume as an argument for quantitative and
experimental approaches to the study of linguistics in general, and
Spanish phonology and morphology in particular. Such approaches have
become more popular in recent years, shifting attention to more
experimental methodology as an alternative to formal linguistic analysis.
In contrast to the latter, Eddington paints a picture of experimental
approaches as marginal, or in "left field" (xiii), and therefore
potentially controversial, but he takes pains to point out that the
research he discusses here is but one part of a rich and varied field in
which formal linguistics also has a role to play. The book is directed
toward linguists and students with limited experience with what Eddington
classifies as empirical approaches to linguistics. It is therefore not a
comprehensive review of the literature on Spanish falling into this
category. It is instead an introduction, presenting and exemplifying the
unique contribution of this type of experimental methodology, the kind of
questions it may begin to answer, and what techniques may be appropriate
for answering them. In the following I will summarize the contents of each
chapter, followed by an assessment of the book's merits and weaknesses.
Chapter 1: This chapter sets the stage for the rest of the book by drawing
a sharp dividing line between experimental approaches to language and
grammar and formal linguistic analysis. The crucial distinction is based
on the psychological reality of the analyses resulting from each approach.
Eddington distinguishes between formal linguistics as a non-empirical
science and experimental and quantitative approaches, which he classifies
as empirical. He discusses the notion of falsifiability and the role of
idealizations or heuristics (e.g. the ideal speaker-listener), and
explains why these are not in themselves falsifiable theoretical
constructs. He follows with the types of evidence that formal and
experimental approaches may draw on, paying particular attention to the
weaknesses of autonomous methods, that is, methods that attempt to isolate
analyses from particular individual speakers, a practice Eddington
attributes primarily to formal approaches. While the bulk of the chapter
appears devoted to pointing out that traditional formal linguistic methods
are incapable of determining the psychological relevance of their
resulting analyses, Eddington is careful to point out that both formal and
empirical approaches are valuable, provided their domains are kept
separate. Thus, Eddington dichotomizes formal and experimental approaches
to linguistics, the first of which may be understood to concern the
consequences of laws and representations, and the second to concern the
substance and content of those laws (Mohanan 1997).
Chapter 2: In this chapter Eddington argues for the use of experimentation
in linguistics and defends its use against several common objections. He
makes it clear that experiments, understood to include also more
naturalistic data that is analyzed statistically to test an explicit
hypothesis, provide the kind of non-autonomous, empirical, and
spatiotemporal data that he argues are required to determine the
psychological relevance of linguistic realities. The chapter is structured
primarily around a list of several criticisms of experimental methodology,
including the knowledge base on which it rests, its relevance to
linguistic competence (vs. performance), external validity, and its
ability to distinguish between competing analyses. This discussion
clarifies the domain of empirical approaches as pertaining to actual
language processing and performance, and points out that idealizations
(e.g. competence) and notational artifacts (e.g. diacritics in underlying
representations) fall outside the domain of experimental research.
Eddington closes the chapter with brief discussions of experimental
approaches to English phonology, including a series of studies testing the
psychological reality of the English vowel shift, exemplifying the
experimental rigor needed to make specific conclusions in light of
potentially conflicting results.
Chapter 3: The goal of this chapter is to show that analyses based on
small data sets and the intuitions of a few individuals, as Eddington
argues is often the case, frequently do not capture the psychological
pertinence, or even the descriptive facts of linguistic phenomena. The
discussion is centered on studies addressing several phenomena in Spanish
morphophonology, including vowel opening following syllable final /s/-
deletion, coronal and velar softening (e.g. dividir~divisi-n,
divide~division), depalatalization (e.g. do-a~don, Mrs.~Mr.), intonation,
and change-of-state verbs. While the chapter makes no attempt at a
comprehensive review of this literature, Eddington uses a series of
examples to show that historical evidence, detailed phonetic analysis,
perceptual and productive experiments, and corpus analyses can reveal
highly complex and nuanced patterns of language behavior that defy
attempts to arrive at simple or elegant generalizations. He argues,
however, that psychological pertinence, and not elegance, is the goal of
the approach advocated in this book, and that evidence from the methods
exemplified here uncovers a level of detail that is missing in studies
that rely on the speaker intuitions and the small data sets on which most
traditional formal analyses rely.
Chapter 4: Eddington devotes this chapter to demonstrating that frequency
must be considered as a factor in determining the psychological reality of
linguistic phenomena. Two different measures of frequency are discussed,
type frequency (the number of units that participate in a given pattern)
and token frequency (the number of times a unit appears). Eddington then
gives an example of a study (Perez 1998) in which frequency appears as an
uncontrolled confounding variable, and shows how frequency may be used to
clarify an otherwise odd result. The importance of measuring frequency in
empirical studies of language is further illustrated at the level of
phoneme clusters, words, and collocations. While again not providing a
comprehensive review of the phenomena in question, Eddington presents data
showing that the frequency of certain phoneme clusters may be used to
provide a more grounded account of phenomena such as /e/-epenthesis and
the tendency of /VsC-/ onsets to be produced as /esC-/. Frequency is also
implicated in the tendency for frequent word combinations to become
lexicalized as single units. However, this chapter is not an exhaustive
review of the literature on frequency, and provides only a taste of the
problems that arise in attempts to examine the role of this complex but
crucial variable in language processing.
Chapter 5: Whereas in the majority of the book Eddington discusses
experimental methods and evidence, in this chapter he presents an
alternative way of constructing theory to that of formal analysis, one
that produces empirically testable predictions. He spends the bulk of the
chapter discussing Skousen's (1989, 1992) Analogical Model as a
representative of exemplar-based models of language processing in general.
The discussion is framed in terms of the debate concerning the roles of
computation and the lexicon in grammar, with this class of models
representing the maximal involvement of the lexicon, in contrast to models
that "require speakers to glean generalizations from the data and
formulate them into systems of rules or constraints" (p. 98). Eddington
goes on to discuss the level of detail in the lexicon as well as its
structure within contextual space, and gives examples of how this
structure may be modeled in computer algorithms in order to test various
predictions. Particular attention is given to the fact that exemplar-based
models are able to account for frequency effects and gradience. The use of
such models is supported by reviews of the literature on gender
assignment, the formation of nouns ending in -i-n, and dialectal
differences in diminutive formation, drawing on corpus-based computational
simulations as well as studies of child language acquisition, markedness,
and speech errors.
Chapter 6: In contrast to the presentation of a theoretical approach that
is empirically testable in Chapter 5, Chapter 6 contains a review of
experimental and empirical studies of three highly studied phenomena in
Spanish, diphthongization, syllable structure, and stress. The chapter
does not contain a thorough review of earlier research that would
presumably fall into the non-empirical category, but rather a discussion
of a number of studies that have first attempted to establish the
psychological relevance of these phenomena to actual language processing
in native speakers, and then to probe their origins and structure. The
evidence presented in these studies includes historical sound change,
nonce probe tasks, computational simulations, forced choice and lexical
decision experiments, as well as corpus surveys. These inquiries revealed
that some phonological phenomena have a distinct psychological reality
(e.g. the syllable) whereas others have more limited effects on language
processing (e.g. diphthongization, syllable weight). This chapter thus
attempts to show a highly nuanced view of several phenomena that have
traditionally received a large amount of attention in the literature.
Chapter 7: Eddington more specifically addresses the question of
morphological processing in this chapter. In doing so he focuses again on
the question of psychological reality, in particular the question of
whether morphological processing is a separate kind of processing in
itself or whether it is a combination of the effects of
orthographic/phonemic and semantic processing, given that morphemes are
relatively stable phonological units with strong semantic links across
lexical items. To elucidate this problem, Eddington discusses evidence
from lexical decision task studies that employed priming across modalities
and between scripts, where the degree of orthographic and semantic overlap
of primes was manipulated. He discusses the contradictory nature of some
of these results, and their implications for the interaction between
orthographic, semantic, and morphological processing. These studies relied
on several different languages, and Eddington turns in the following
section to similar work conducted on Spanish. He then discusses a group of
models of morphology in which "morphemes can be viewed as interconnected
patterns that exist in two or more words that are both semantically and
orthographically/phonologically similar" (p. 135). He closes the chapter
by reviewing several studies that investigated whether words in Spanish
and English are stored with or without their gender (Spanish only) and
plural morphemes (for plural forms), thus probing the psychological
reality of separate gender and plural morphemes.
Chapter 8: The final chapter of the book serves as a brief summary of the
contents of each chapter and of the main thrust of the book, as discussed
above.
Appendix: Based on the assumption that many linguists receive only limited
training in experimental design, data collection, and statistical
analysis, Eddington devotes a substantial appendix to a discussion of the
tools necessary for quantitative and experimental inquiry into Spanish
phonology and morphology. He first discusses four common types of
statistical analysis, the correlation, chi-square tests, logistic
regression, and analysis of variance, outlining the specific purpose of
each, the type of data required, statistical significance, and other
issues specific to each kind of test. The following section contains a
discussion of basic experimental design, focusing on the lexical decision
task and on questions of number of subjects and of obtaining approval for
the use of human subjects. He then proceeds to a discussion of various
threats to the internal and external validity of experiments, briefly
offering examples of solutions to each one. The appendix also contains a
list of software resources, including statistical packages, software for
running experiments, electronic corpora and computer language simulations.
Eddington provides references to these resources and to user's manuals,
especially when these are available electronically.
CRITICAL EVALUATION
As mentioned above, this book is a concise introduction to and defense of
the use of experimental and quantitative methodology in endeavoring to
account for language structure in a psychologically relevant way. It
provides a careful and enlightening tour of the ways that an experimental
approach may make a unique and valuable contribution to the study of
language, and makes a compelling argument that the problem of
psychological reality falls outside the domain of formal linguistic
analysis. In doing so, it sets up empirical approaches in opposition to
the methodology of formal linguistics, charging that the latter belongs to
the realm of non-empirical sciences such as pure mathematics and logic.
The relationship between these two approaches, however, is somewhat
ambiguous, and this is a primary weakness of the book. Eddington
explicitly stresses the inherent value in formal approaches, but provides
only the briefest indication of what this value might be, only that the
domain of formal analyses must be kept separate from that which concerns
psychological reality. In later chapters, however, Eddington at times uses
formal analyses to elucidate the linguistic structure that may be
subjected to tests of psychological reality, suggesting a more symbiotic
relationship between the approaches. This possibility, while potentially
interesting and helpful, is only briefly alluded to (p. 28), and formal
analyses are, in the main, not included in the reviews of literature.
As an overview of the application of empirical methodology to problems in
Spanish phonology and morphology, however, the book provides a valuable
introduction. It covers experimental studies of a variety of levels of
phonology and morphology, touching on variation in the realization of
individual phonemes, prosodic features such as stress, morphophonological
alternations, and the behavior of classes of words such as change-of-state
verbs. Eddington also includes a wide variety of experimental methods,
from computational simulations and corpus analyses through questionnaires
to studies of online lexical processing using the lexical decision task.
The book is nonetheless not a treatise on experimental methodology, and a
number of domains of research into the psychological reality of linguistic
structures are excluded. One particularly salient absence, given the
emphasis on frequency effects, is the research on frequency effects and
phonotactic probability in lexical access, which relies on such online
measures as the lexical decision (e.g. Vitevitch & Luce, 1998; e.g.
Vitevitch & Luce, 1999).
Due to this lack of discussion of some areas of research, the boundaries
of empirical methodology are left somewhat unclear. For instance,
Eddington includes naturalistic data collection among these approaches
(Ch. 2), particularly stressing its value as a way of bolstering the
external validity of findings obtained in the laboratory, but he provides
little guidance about how a researcher might obtain or use such evidence
in this way. He also does not go into detail about the contrast between
questionnaires, which provide data about metalinguistic judgments and
intuitions, and online measures of processing. This distinction is
crucial, in that intuitions may depend on different mechanisms than online
processing, regardless of whether the intuitions are the investigators, or
are collected from a large sample of speakers.
Nonetheless, this book is an argument for the value of experimental and
quantitative methods, and not a comprehensive treatise on methodology. As
such, it provides enough information to set out some of the primary
questions that may be addressed from an experimental approach, as well as
some of the problems with using such a methodology. To this end, the
appendix may be particularly helpful to readers who desire to begin to
apply experimental methods. Eddington includes copious references to
larger works on statistical analysis, experimental design, and so forth,
and provides web addresses where, at least at the time of publishing,
researchers can obtain software and support for conducting research on a
variety of empirical questions concerning morphological and phonological
structure. This book is thus both a valuable contribution to the debate
over the psychological reality of linguistic analyses, and a helpful
resource for those who wish to explore what experimental and quantitative
methodology can reveal about language structure and processing, pointing
them to more extensive resources that may provide a solid foundation for
experimental design and empirical investigation.
REFERENCES
Mohanan, K. P. (1997). LINGUIST List posting, April 23, 1997.
http://linguistlist.org/issues/8/8-575.html
Pérez, H. E. (1998). "Incidencia de dos rasgos acústicos en la percepción
de la correlación /p-t-k/ vs. /b-d-g/". Revista de Lingüística Teórica y
Aplicada. 36, 113-125.
Skousen, R, (1989). Analogical modeling of language. Dordrecht: Kluwer.
--------------- (1992). Analogy and structure. Dordrecht: Kluwer.
Vitevitch, M. S., & Luce, P. A. (1998). "When words compete: Levels of
processing in perception of spoken words". Psychological Science, 9(4),
325-329.
Vitevitch, M. S., & Luce, P. A. (1999). "Probabilistic phonotactics and
neighborhood activation in spoken word recognition". Journal of Memory and
Language, 40, 374-408.
ABOUT THE REVIEWER
Matthew Carlson is currently a Ph.D. candidate in Hispanic Linguistics at
the Penn State University. His primary research is on the role of
frequency and usage in the adult second language acquisition of Spanish
phonology. Other interests include usage-based approaches to grammar and
to phonology in particular, working memory and phonological memory, and
the effects of literacy and orthography on SLA.
-----------------------------------------------------------
LINGUIST List: Vol-16-1457
More information about the LINGUIST
mailing list