Review of Productivity in English Word-formation
|AUTHOR: Fernández-Domínguez, Jesús
TITLE: Productivity in English Word-formation
SUBTITLE: An approach to N+N compounding
SERIES: European University Studies, Series 21: Linguistics. Vol. 341
PUBLISHER: Peter Lang, Bern
Reviewed by Pius ten Hacken, Swansea University
In the field of morphology, productivity is one of the most discussed concepts.
It can be described as the possibility of creating new words. Discussion has
focused on the analysis of the nature of what we call productivity into
different aspects as well as on the degree to which rules have this property.
This book works towards a new measure for productivity. It differs from most
other studies in this domain in concentrating on compounding rather than
affixation. The author refers to his MA thesis (2006) and to his PhD thesis
(2008), both of which are about measuring productivity, but I have not found any
explicit statement of how the book relates to these theses.
The book consists of five chapters. Chapter 1 (1-8) is a brief introduction,
giving some methodological background and a chapter overview. A set of compounds
was extracted from the British National Corpus (BNC) and analysed semantically
in terms of Levi's (1978) Recoverably Deletable Predicates (RDPs).
Chapter 2 (9-44) introduces compounding. After introducing headedness and
various systems of characterizing and classifying the relationship between the
components of a compound, the author gives an overview of English compounds
following Bauer & Huddleston (2002). Section 2.4, roughly half of the chapter,
is devoted to the question of how to distinguish compounding from syntax. After
an overview of various proposals, the conclusion is that no clear criteria have
been found. Some authors model the transition as a cline.
Chapter 3 (45-99) is devoted to morphological productivity. It starts by
introducing a number of dichotomies. The first is between the process
(''Wortbilding'') and the outcome (''Wortgebildetheit''). Then productivity is
opposed to creativity, in which intentionality plays a role, and finally
availability, a qualitative notion, to profitability, which is quantitative.
After this, a section is devoted to factors influencing productivity, both
enhancers and constraints. Among the latter, blocking has been studied a lot.
Finally, attempts to distinguish different degrees of productivity are
presented. A large part of this section is devoted to the stages in the life of
a word until it is lexicalized.
Chapter 4 (101-169) presents different approaches to measuring productivity and
applies some of them to the set of compounds retrieved from the BNC and
classified as to their RDP. The simplest method is to count types and tokens of
the result of a word formation rule in a large corpus. Aronoff (1976) proposed
to relate the number of actual types to the number of potential types, but
realized that such calculations are not feasible in practice. Harald Baayen
proposed different measures based on the number of hapaxes in a corpus.
Stekauer's onomasiological approach concentrates on the process of naming a
concept. As this process is always successful, the total productivity must be
100%. Ingo Plag and Laurie Bauer proposed measures based on neologisms found in
a dictionary, but this is not so useful for compounds. Jennifer Hay proposed to
look at the frequency of the derived item in relation to the frequency of the
base and to whether the phonotactic sequence at the boundary of the base and the
affix can occur also morpheme-internally. The idea is that if the derivation is
relatively rare and phonotactically marked, it is more likely to be processed
online. Calculating the relative frequency for compounds is not straightforward,
because compounds have two bases.
In the final section of chapter 4 (147-168), the author presents his own
proposal for a productivity measure. He proposes an indicator of profitability
(lower case pi) which is the number of types in a corpus (V) divided by the
number of tokens in the same corpus (N). Pi should only be calculated if the
number of types is sufficiently high, but the threshold (which he calls the
''Minimal V-Input'') depends on the corpus. At the moment there is not enough
evidence to give any specific value. A second measure is the trend of
profitability (upper case PI). It is a table of V and N, in which the
development of the increase of the two is plotted when V increases. These two
measures, pi and PI, are applied to the nine RDPs.
Chapter 5 (171-176) briefly summarizes the conclusions of the study.
Before I address the content of the book, I would like to make some observations
about its form and style. The book is marked first of all by a rather
idiosyncratic use of English lexis and word order. Although in most cases this
does not affect understanding, the book would have been generally more pleasant
to read if it had been properly proofread by a native speaker. A second point
concerns the way references are built into the text. There are many references,
but often their main purpose seems to be to impress the reader by suggesting a
very wide reading. I have no reason to doubt the author's wide reading, but I
find it irritating to read lists of references preceded by ''see also'' without
any further discussion. In many cases, the references following ''see also'' do
not seem to support the preceding statement, but only to say something about the
topic. It would have been preferable if a selection of these references had been
properly discussed. A final point concerns the structure of sections, in
particular in chapters 2 and 3. These sections give a catalogue of alternative
views without developing an argument as to which of them is to be preferred. If
it is not necessary to make such a point, one wonders why these sections are
there in the first place.
In fact, large parts of the book are devoted to contextualizing the main
argument. The author's own contribution is limited to the final section of
chapter 4 and the evaluation for compounding of three of the other productivity
measures. These sections build on an uncritical and apparently unreflected use
of Levi's (1978) set of RDPs. There are two major problems with this. First,
Levi's system of RDPs has been heavily criticized and this criticism should have
been addressed. Secondly, the RDPs are used in a way that does not seem to be
compatible with Levi's original intention.
Levi proposed RDPs as a characterization of the relation between the two
components of a compound (actually a ''complex nominal'', but I will ignore the
difference here). Her choice of predicates has been criticized for at least four
reasons. First, some predicates are excessively vague, e.g. both ''fertility
pills'' and ''headache pills'' are characterized as FOR. Second, there is a
significant overlap between predicates, e.g. ''party members'' can be HAVE or IN.
Third, RDPs are designated by English words, and tend to exploit the range of
meanings these words have in English. Thus, HAVE is used for ''horse leg'' as well
as for ''student problems''. Finally, the predicates are not exhaustive, as
Downing (1977) illustrated with the famous example of ''apple-juice seat''.
Apart from their deficiencies in characterizing the relationship between the
components of a compound, RDPs are not used in the originally intended way when
their productivity is compared. Levi's (1978) theory was based on Generative
Semantics. This does not exclude its use in a different framework, but the
problems raised by transplanting it to a new framework should have been
discussed. Levi intended the RDPs as a way to reconcile the transformational
derivation of compounds with the requirement that deleted material should be
recoverable. They are meant to operate together with a system where deep
structure predicates that are part of a deverbal head are also available. In
Generative Semantics, this not only includes cases such as ''truck driver'', but
also cases such as ''car thief'', where ''thief'' is equivalent to ''stealer'' in deep
structure. Therefore, RDPs do not have to account for these cases. For
productive, non-lexicalized compounds, there is a sequence of transformations
deleting parts of the underlying sentence, including the RDPs. Levi assumes that
a compound is at least 12-ways ambiguous, because there are nine RDPs and three
of them can be used in active and passive readings. If there is a deverbal head,
further readings are added. Only lexicalized compounds have a single meaning,
which can of course be much more specialized than an RDP.
The author of this book also fails to mention that three RDPs can be used in
active and passive readings and seems to assume that there are nine compounding
rules in competition with each other. Disambiguation only occurs on the basis of
context, however, or when the compound is lexicalized. Therefore, there must
have been many classification problems. However, all of the assumptions relating
to RDPs and problems in applying them remain entirely implicit. It is as if the
author takes Levi's RDPs to be a mainstream device for which no discussion is
necessary. In the absence of any such discussion, however, it is hard to
interpret the figures that emerge from the experiments. While a lot of context
of the research is given, most of chapters 2 and 3, for which the lack of a
conclusion suggest that it is not essential for the actual research, crucial
issues relating to the origin and use of RDPs are not mentioned at all.
In conclusion, the book shows that the author is familiar with a large amount of
literature in the field of morphology. In presenting this material, he does not
integrate the various views and for most of the issues he discusses he does not
develop a view of his own. The main point of the book is a study of productivity
of compounding. This study suffers from serious methodological problems in its
Aronoff, Mark H. (1976). Word Formation in Generative Grammar. Cambridge
(Mass.): MIT Press.
Bauer, Laurie & Huddleston, Rodney. (2002). Lexical word-formation. In
Huddleston, Rodney & Pullum, Geoffrey K. (eds.), The Cambridge Grammar of the
English Language. Cambridge: Cambridge University Press, pp. 1621-1721.
Downing, Pamela Ann. (1977). On the Creation and Use of English Compound Nouns.
Levi, Judith N. (1978). The Syntax and Semantics of Complex Nominals. New York:
ABOUT THE REVIEWER
| ABOUT THE REVIEWER:
Pius ten Hacken is senior lecturer in linguistics and translation at the
Department of Translation and Digital Communication of Swansea University.
His research interests include morphology, terminology, and the lexicon. He
published his PhD 'Defining Morphology' in 1994. His latest book is
'Chomskyan Linguistics and its Competitors' (London: Equinox, 2007).