LINGUIST List 13.622
Thu Mar 7 2002
Review: Inflectional Morphology: Theory of Paradigm
Editor for this issue: Terence Langendoen <terrylinguistlist.org>
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 simin
linguistlist.org
or Terry Langendoen at terry
linguistlist.org.
Subscribe to Blackwell's LL+ at
http://www.linguistlistplus.com/ and
donate 20% of your subscription to LINGUIST! You get 30% off on
Blackwells books, and free shipping and postage!
Directory
Ida Toivonen, Stump (2001) (second review)
Message 1: Stump (2001) (second review)
Date: Wed, 6 Mar 2002 15:41:35 -0800 (PST)
From: Ida Toivonen <toivonencsli.Stanford.EDU>
Subject: Stump (2001) (second review)
Stump, Gregory T. (2001) Inflectional Morphology: A Theory of Paradigm
Structure. Cambridge University Press, xvi+308pp hardback ISBN
0-521-78047-0, Cambridge Studies in Linguistics 93.
Ida Toivonen, University of Rochester
OVERVIEW
A fully realized morphological theory is an important contribution to
linguistic theory. Gregory Stump's 'Inflectional Morphology: A theory of
Paradigm Structure' gives us exactly that: an explicit and general theory
of inflectional morphology (it is suggested that the theory can be
extended to cover derivational morphology as well). The theory Stump
develops is called Paradigm Function Morphology (PFM). PFM takes the
paradigm to be the primary object of study, rather than the word. Another
important characteristic of the theory is that it treats morphology as an
autonomous module of the grammar, separate from syntax and phonology.
The book consists of a list of abbreviations, eight chapters, footnotes,
references and a general index.
In what follows, I will summarize each individual chapter of the book (see
also Avgustinova 2001). After that, I will comment on some of the issues
raised in the chapters.
Chapter 1. (Inferential-realizational morphology)
The first chapter sets up a typology of different types of morphological
theories. First, lexical and inferential theories are juxtaposed. In a
lexical theory of inflectional morphology, associations between
morphosyntactic properties and morphemes are listed in the lexicon. In an
inferential theory, relations between a root and its various inflected
forms are expressed by rules or formulas. PFM is inferential, not lexical.
Stump also distinguishes incremental theories from realizational
theories. In incremental theories, morphosyntactic information gets added
incrementally as morphemes are added to a stem. In a realizational
theory, a word's association with certain morphosyntactic properties
licenses the appropriate affixes. PFM is realizational.
Stump first identifies the four types of theories that emerge from
different combinations of the distinctions outlined above, and then
provides empirical and theoretical arguments for inferential-realizational
theories. The chapter ends with an outline of the main characteristics of
PFM.
Chapter 2. (Paradigm functions)
This chapter introduces and makes explicit the basic claims and principles
of PFM.
First, the notions of 'paradigm' and 'paradigm function' are explained.
In PFM, paradigms are taken to be a central object of analysis and not
epiphenomenal, as they are in many other theories. The word in any given
cell of a paradigm is determined by a paradigm function, which applies to
a root paired with the morphosyntactic properties associated with the cell
in question. An example Stump gives is German 'Buch' (book). A paradigm
function applies to <Buch-,{'genitive', 'singular'}> to yield <Buches,
{'genitive', 'singular'}>.
In a given language, a paradigm function is defined in terms of more
specific realization rules, which are organized into blocks. Another type
of rule is morphological metageneralizations: a rule about
morphophonological regularities associated with several different
morphological rules.
These foundational concepts are illustrated in an analysis of Bulgarian
verb inflection.
Chapter 3. (Rule competition)
Sometimes the situation occurs where two different rules can apply.
Chapter 3 argues that such competition is always solved by Panini's
principle, according to which the the narrowest rule takes precedence.
Some cases exist where no one member of a rule block can be singled out as
the narrowest applicable rule. Stump solves this problem by postulating
that realization rules may apply in two different modes: unexpanded and
expanded. Detailed analyses of Potawatomi and Georgian motivate these
assumptions.
Chapter 4. (Headedness)
Chapter 4 is concerned with headed words and the different types of
morphological marking they may exhibit. A headed word is defined as a
word which arises through the application of a category-preserving rule of
derivation or compounding, where a category-preserving rule is a rule that
allows a morphosyntactic property to persist from the base to the derived
form. An example is the Russian diminutivization rule which adds the
ending -iSka': the gender of the base is maintained in the derived
diminutive form.
Headed words fall into three subclasses: Some inflect on the head ('head
marking'), some inflect externally, i.e., not on the head ('external
marking'), and some inflect both on the head and externally
('double-marking'). An example of double-marking comes from Breton:
'bag-ig' is a headed root, where 'bag' is the head. The plural is
'bag-ou-ig-ou': the plural marker 'ou' appears on the head and also
externally. This is handled by dividing category-preserving rules into
three types. One type of rule apply to a root to give rise to another
root, i.e., root-to-root rules. The expression generated by that rule is
external marking. Word-to-word rules are head-marking, and word-to-stem
rules are double-marking.
After these concepts have been established, two generalizations are
introduced. 1) If you consider two headed coderivatives (i.e. words which
have arisen through the same derivation rule), either both exhibit head
marking or neither does. 2) If a root ever exhibits head marking in its
inflected paradigm, it always does. These generalizations are captured by
the 'head-application principle'. Most of the chapter consists of
motivating these generalizations and explaining the principle, but a
discussion of some apparent counterexamples is also included.
Chapter 5. (Rule blocks)
In PFM, rules are assumed to be organized into blocks. The rules are not
ordered within a block, but are selected in accordance with Panini's
principle. Chapter 5 shows how three important types of morphology can be
dealt with in terms of rule blocks: portmanteau morphology, parallel rule
blocks (classes of morphemes which overlap in their membership), and
reversible morphology (classes of morphemes whose relative position
varies according to the set of morphosyntactic properties realized). Stump
argues that these phenomena provide a strong motivation for 'template'
morphology, for which PFM provides an explicit theory. However, paradigm
functions are not positive output constraints, so the notion of template
in PFM differs in that way from the traditional view of templates.
The analyses laid out in this chapter crucially rely on rules of referral,
which make up a subclass of realization rules (rules of exponence being
the other subclass). Rules of referral identify the morphological
realization of one set of morphosyntactic properties with that of some
contrasting set of properties. Such rules have previously been employed
by Arnold Zwicky and others in order to account for syncretism. Stump
adopts them for syncretism as well (chapters 2 and 7), but in chapter 5,
he extends their use. Under his formalization, rules of referral can
refer the realization of a set of morphosyntactic properties to rules
which are situated in other rule blocks but realize the same syntactic
properties. They are under this definition put to use in the analyses for
portmanteau, parallel and reversible morphology. These morphological
phenomena thus provide independent motivation for the existence of rules
of referral.
Chapter 6. (Stem alternations)
A single lexeme can have a variety of distinct stems within its paradigm.
Chapter 6 is concerned with how PFM can account for when and why a
particular stem is chosen over another in a given cell of the paradigm.
An important characteristic of stem alternations is that they are often
not predictable from other factors, such as meaning, morphosyntactic
feature content or phonology. Stump therefore assumes that a lexeme's
stems carry indices to distinguish them from each other. This distinction
is morphomic, i.e. it has nothing to do with any component in the grammar
except morphology. Stem alternations thus provide an argument for the
autonomy of morphology. Stem alternations are dealt with in PFM by two
kinds of morphomic rules: stem formation rules and stem indexing rules.
Stem indexing rules are appealed to when the difference between two stems
does not follow from anything else, such as phonology or morphosyntactic
feature content. Morphomic rules constitute a separate type of
inflectional rules alongside realization rules (rules of exponence and
rules of referral).
Stems are thus distinguished from each other by the means of stem
formation rules or stem indexing rules. The choice of which stem to use
in what paradigm cell(s) is determined by stem-selection rules (a type of
rule of exponence) or morphological metageneralizations.
Chapter 7. (Syncretism)
Syncretism is mentioned in several chapters, but treated in full in
chapter 7. Syncretism is a situation where "two or more cells within a
lexeme's paradigm are occupied by the same form" (Stump, p. 212). Stump
argues that syncretism is directional or nondirectional. When some form
adopts the morphology of another form, the syncretism is said to be
directional; otherwise the syncretism is nondirectional. Directional
syncretism is unidirectional or bidirectional. When nondirectional
syncretism is can be explained as a mere lack of contrast in the system of
forms (e.g., third person verb forms are not distinguished for number), it
is referred to as unstipulated syncretism. Stipulated, nondirectional
syncretism is called symmetrical syncretism. An example of symmetrical
syncretism comes from Hua, where 2nd person singular and 1st person
plural forms are syncretized. This leaves us with four types of
syncretism: unidirectional, bidirectional, unstipulated and symmetrical.
Unstipulated syncretism simply reflects the fact that the rules of a
language are not sensitive to a certain distinction (e.g., number).
Directional syncretisms are accounted for by rules of referral.
Symmetrical syncretisms are accounted for by means of a metarule, which
creates a connection between two rules.
Chapter 8. (Conclusions, extensions, alternatives)
The final chapter summarizes the formal mechanisms employed in PFM. This
chapter also discusses some important issues that go beyond the formation
of inflected words. One issue is the semantics of inflected words;
'inflectional semantics'. Specifically, bracketing paradoxes (exemplified
by 'unhappier') are addressed. Another issue is derivational morphology,
which does not otherwise receive much attention in the book (although
derivational morphology does play an important role in chapter 4). The
book ends with a discussion of an alternative approach, Network
Morphology. Network Morphology is similar to PFM in many ways, and Stump
points out that it is compatible with most of the assumptions laid out in
the previous chapters.
COMMENTS
It follows from PFM that inflectional templates (the organization into
position classes) are defined by paradigm functions. Since all
inflectional morphology is defined by paradigm functions, there is in
effect no difference between templatic inflection and layered
(affix-to-base) inflection. Since it has been argued that reference to
templates is necessary in order to analyze some inflectional phenomena
(see, e.g., chapter 5 of the book under discussion here), the simplest
assumption seems to be that ALL morphology is templatic. However, much
work in morphology is built solely upon a layered model, and layered
morphology is often assumed in addition to template morphology even by
researchers who take templates to be necessary in order to capture
certain phenomena or to model some languages. In other words, many
morphologists seem to have a strong intuition that a layered model is a
natural and appropriate way to think about inflectional morphology.
Stump's work provides a clear and fully formalized theory of morphology
which assumes that all inflection is templatic (but recall that paradigm
functions are not positive output constraints). This is very valuable, as
it will give proponents of different views something concrete to respond
to. It is interesting to note that Stump's argumentation turns the tables
in a sense: If we grant the point that some morphological phenomena call
upon position classes for their analysis, and Stump can get all
inflectional morphology with this model, then the burden of proof falls
upon the proponents for layered morphology, and not the morphologists who
argue for position classes. References relevant to this topic include
McDonough 2000, Spencer 1991, Simpson and Withgott 1986; see also
references cited in those works.
One of the major benefits of this book is its vast empirical coverage: A
large variety of morphological phenomena from many typologically diverse
languages are considered. In fact, it is striking how little English data
is included in the discussion. This seems exotic and unfamiliar to those
of us who are used to reading books by English-speaking authors,
presenting formal *syntactic* theories. Although other languages are
frequently appealed to, new syntactic theories are generally developed
with English in mind. PFM is clearly not developed mainly with English in
mind, and one of the reasons for this seems quite obvious: PFM is a theory
of MORPHOLOGY, and English morphology is not very elaborate. This leads
to a question that is often raised by typologists: What would (Western)
linguistic theory look like if the research had not from the beginning
focussed on English and related western European languages? A linguistic
theory originally developed for Warlpiri or Inari Sami would perhaps
include an elaborate model of morphology, and syntax would just be seen as
a not-so-interesting by-product of semantics and morphology. Stump does
not elevate morphology over the other components of grammar, but he does
grant it a role that a more English-biased approach would probably not
have envisioned.
Although my reaction to this book is generally very positive, I have one
major criticism: At the end of some chapters, I was left with the feeling
that a lot of machinery had been introduced to deal with relatively
straightforward data. The theory makes use of many different rule types,
rule blocks, metageneralizations, and other mechanisms. I feel
particularly suspicious of the postulation in chapter 3 that rules can
apply in different modes: expanded or unexpanded mode. However, Stump
argues that this postulations is necessary to avoid introducing rule
ordering (a mechanism adopted by several other theories). In fact, each
formal mechanism introduced is carefully motivated. Moreover, a vast
variety of morphological phenomena are covered by the theory --- PFM would
of course look a lot more simple if important data were pushed aside.
Stump treats morphology as an autonomous module, completely independent of
other types of linguistic information. This is a good starting point and
an interesting contrast to analyses that treat morphology as nothing but
phonology and/or syntax. However, I think future research combining PFM
with work on the interfaces between morphology and other modules might
prove fruitful (Stump also hints at this in the concluding chapter), and
perhaps some of the resulting discoveries will help modify or discard
mechanisms that at present seem (to me, at least) awkward and overly
powerful.
In sum, the range of languages and linguistic phenomena treated in this
work is impressive. Stump develops a clear and precise theory of
morphology that I think will be of interest for all scholars interested in
word formation, whether their own research adopts the paradigm-based
approach or not.
REFERENCES
Avgustinova, T. 2001. Review of Stump, Inflectional Morphology.
http://linguistlist.org/issues/12/12-1861.html/.
McDonough, J. 2000. Athabaskan redux: Against the position class as a
morphological category. In: Dressler et al. (eds.), Morphological Analysis
in Comparison. Amsterdam: John Benjamins Publishing Company.
Spencer, A. 1991. Morphological Theory. Oxford: Blackwell Publishers.
Simpson, J. and Withgott, M. 1986. Pronominal clitic clusters and
templates. In: Borer (ed.), Syntax and Semantics 19: The Syntax of
Pronominal Clitics. Academic Press.
ABOUT THE REVIEWER
Ida Toivonen received her Ph.D in Linguistics at Stanford University. She
is now a Visiting Assistant Professor at the University of Rochester. Her
main research interest is syntax, and she works mainly on Swedish, Finnish
and Inari Sami.