Date: Tue, 8 Apr 2003 19:38:17 -0700
From: Ferdinand de Haan <email@example.com>
Subject: Typology and Universals
Croft, William (2003) Typology and Universals, 2nd ed., Cambridge
Ferdinand de Haan, University of Arizona
This book is the second edition of a book that first appeared in 1990.
Compared to the first edition many changes were made, reflecting the ever
evolving field of typology. The first edition of the book was already very
good, but the second edition is even better. The aim of the book is not to
provide an analysis of selected topics in typology (such as Comrie 1989),
but is meant as an introduction to how to think like a typologist and how to
formulate a convincing typological analysis. It is written with admirable
clarity and each topic is introduced with a simple example as a starting
point and Croft gives successively deeper and deeper explanations for a
given phenomenon. In that, he succeeds very well.
Chapter 1 ('Introduction') is, almost inevitably, an introduction to the
field of typology, which is defined as an empirical-scientific approach to
the study of language. It is compared to formal approaches. The main
difference between typology and formal approaches is the fact that typology
aims to uncover universals through cross-linguistic research. However,
typology has in common with the formal approaches the fact that all seek to
answer the question: What is a possible language? The reader is introduced
to typology through a cross-linguistic comparison of the use of articles in
English and French. The question that arises is: how can we compare
categories cross-linguistically, for instance the category 'subject' or
'relative clause'? Quite often one has to apply external criteria, such as
semantic criteria, in order to be able to do cross-linguistic research. The
chapter concludes with a section on sampling problems (a typological study
requires an adequate sample of the world's languages), and a section on the
reliability of data sources.
Chapter 2 ('Typological classification') concerns the various linguistic
types (or strategies) for a given parameter. This is exemplified with the
example of possession constructions. For this parameter, a number of
strategies are given, ranging from the simplest strategy (juxtaposition of
possessum and possessor) to highly grammaticalized strategies (possession
marked with clitics). Languages may have more than one way of marking a
given parameter, which leads to the statement that it is not languages that
are classified, but linguistic types. If a language has more than one type,
how can one determine which type (if any) is the basic one? Croft gives a
number of criteria; a type is less basic if it (a) is restricted to a
grammatical subclass, (b) pragmatically specialized, (c) structurally
unusual (i.e., more complex), or (d) less frequently found in texts. From
these linguistic types, complex language universals can be constructed. The
chapter concludes with a section on morphological typology, which is
essentially a historical overview on morphological classification, starting
with Schlegel's early 19th century classification.
We now come to the meat of the book. Chapter 3 ('Implicational universals
and competing motivations') discusses the nature of typological universals.
The basic problem is how to account for the language types that actually
exist. The answer is: with universals. There are two types of universals,
the unrestricted universal ("all languages have property X"), and the
implicational universal ("if a language has property X, it will also have
property Y"). There is also a discussion on the logical structure of
implicational universals. A universal like the one above predicts that one
linguistic type should not occur, namely a language that has property X, but
not property Y, because that is excluded on logical grounds. It is sometimes
thought that the discovery of universals is the goal of typology, but in
fact they are only a necessary prerequisite for deeper explanations. The
next section looks at competing motivations for universals. Sometimes the
logically excluded type does occur, but in just a very few languages. This
can be explained by recognizing that possibly more than one factor (or
"motivation" plays a role). Two concepts are introduced, that of "dominance"
and that of "harmony". These are hard concepts to get across but Croft
manages very well. The dominant type occurs in the implicans of an
implicational universal, while harmony links two values of different
parameters if these two only occur with each other (thus in an implicational
universal "if X then Y" the values X and Y are harmonic, as are the values
~X and ~Y). The existence of competing motivations such as dominance and
harmony can explain why some linguistic types do exist, even though they seem
to be excluded by the implicational universal. Thus, all types that are
partially motivated should exist, but the proportion of attested types
should reflect the degree of motivation for a given type.
The next section looks at explanations for dominance and harmony. Examined
are Hawkins' (1983) Heaviness parameter which states that heavier modifiers
(such as relative clauses) tend to follow the modified word (in this case
the noun), while lighter modifiers such as demonstratives tend to precede it.
This can be explained through ease of parsing. For harmony Hawkins' (1994)
Early Immediate Constituents or Dryer's (1992) Branching Direction Theory are
discussed. The last section deals with ways in which typology can help test
hypotheses from generative approaches with a discussion on typological
approaches to the pro-drop parameter (see Gilligan 1987).
Chapter 4 ('Grammatical categories: typological markedness, economy and
iconicity') discusses the notion of typological markedness, an explanation
for the asymmetrical properties of otherwise equal linguistic elements. In
the category of number marking, 'plural' is marked typologically with
respect to 'singular' because whenever the singular is overtly marked, the
plural is overtly marked as well. This is the principle of 'structural
coding' which states that the marked value of a category has at least as
many morphemes as the unmarked value. A problem is that it is not always
easy to count morphemes (how to count portmanteau morphemes, for instance?)
and so markedness is often assessed based on the principle of 'behavioral
potential' which is looked at according to either inflectional potential
(the marked value has fewer or equal formal distinctions as the unmarked
one) or distributional potential (the marked value occurs in fewer
environments than the marked value). Explanations for markedness are given,
such as 'economy' (expressions should be minimized wherever possible) and
'iconicity' (the structure of language mirrors the structure of experience).
These are competing motivations and are subject to the observations of
chapter 3. Even deeper explanations are searched for such as 'frequency', a
marked value occurs less frequently than the corresponding unmarked value.
Chapter 5 ('Grammatical hierarchies and the semantic map model') discusses
grammatical hierarchies, or chains of implicational universals. Hierarchies
discussed are the sonority hierarchy, animacy hierarchy, person hierarchy,
definiteness hierarchy and the referentiality hierarchy. An example from the
realm of word order is Hawkins' (1983) Prepositional Noun Modifier Hierarchy
which goes: NNum > NDem > NA > NG > NRel. The hierarchy is read as follows:
if a language has a given word order on the hierarchy, it will also have the
word orders to the right of this word order on the hierarchy. Such
hierarchies may be explained by appealing to a semantic map, which maps the
language-specific distribution of a category onto the conceptual space. This
language-specific distribution occupies a continuous region on the map. An
example is Haspelmath's (1997) semantic map for Indefinite Pronouns or
Keenan and Comrie's (1977) Accessibility Hierarchy.
Chapter 6 ('Prototypes and interaction of typological patterns') introduces
the notion of prototype analysis in typology. Quite often the precise nature
of a hierarchy or a markedness relation depends on the precise category
surveyed, because many factors play a role at the same time. For instance,
the person hierarchy depends on what is studied, since the hierarchy differs
in the case of politeness (2 < 3 < 1) and number (1 < 2 < 3). The notion of
'prototype' takes a number of such cases and fits them into an overall
cluster with core and peripheral elements. The most famous prototype is the
transitivity analysis of Hopper and Thompson (1980).
Chapter 7 ('Syntactic argumentation and syntactic structure in typology')
applies typological solutions to syntactic problems. The way distributional
analysis is applied in typology is compared to applications in other
linguistic theories. One such problem is the notion of 'subject' as a
cross-linguistic category. Distributional facts of English are not adequate
in determining (or even assuming) the validity of a universal notion
'subject'. One answer to this and similar problems is to appeal to
hierarchies and to the semantic map model. This chapter pulls together a lot
of theoretical strands from the previous couple of chapters and it is by far
the most difficult one in terms of topics discussed. Nevertheless, the
argumentation is exceptionally clear.
Chapter 8 ('Diachronic typology') examines the role of typology in
diachronic research. Languages can and do change their word order patterns
and it is the job of typology to account for the way such changes occur and
to constrain any attested changes somehow. From attested states typologists
derive processes that drive language change (for instance, the change from a
demonstrative to a definite article) and at the same time try to explain
attestations of word order patterns that are not allowed by implicational
universals. A large section of the chapter introduces the reader to the
framework of grammaticalization. This topic, of all the ones in the book, is
the one that is currently most hotly debated. Grammaticalization is driven
by the notion of 'unidirectionality', or the notion that the change in a
grammatical category occurs one way only. Unidirectionality accounts have
been posited for syntactic and semantic features. It is as yet by no means
an accepted part of linguistic diachronic theory since there are
counterexamples to unidirectional processes. The unresolved problem is how
to treat these exceptions. See Newmeyer (1998), Haspelmath (1999), and the
papers in Campbell (2001), among others, for a discussion of the current
state of the art in grammaticalization.
Chapter 9 ('Typology as an approach to language'), is a somewhat
philosophical chapter to conclude the book. It revisits the questions posed
in the first chapter and ties everything together in light of the previous
This book is to be used as a textbook and so the appropriate question here
is: has the book met its objectives? The answer is a resounding yes. It
complements existing works, most notably Comrie (1989), in that it places
emphasis on how to do typological research. The book introduces the core
concepts and uses them to gradually gain a deeper understanding of how to
construct a meaningful typological analysis. In addition, it is extremely
well written and there are hardly any problems of clarity. An added bonus is
the selection of typology problems which are on Croft's website and are
meant to be done in conjunction with the book (and they are not easy!). It
is hard to find faults with the book. The only wish I would have is to add
an appendix containing the universals of Greenberg (1966) and Hawkins
(1983), because Croft refers to them quite often. They are of course widely
available elsewhere, but since this is a textbook its inclusion would have
been appropriate. Still, this is hardly a fault.
There are a number of textbooks on typology such as Whaley (1997) and Song
(2001), alongside older ones such as Mallinson and Blake (1981). The choice
of textbook is a highly individualized one and depends largely on
personality, audience and goal. Nevertheless, this book would be an
excellent source for most people. Intuitively, I feel it would be most
appropriate in graduate courses on typology, because of the level of
You may even give a copy to your formalist friends as an example of solid
Campbell, Lyle, ed. (2001). Grammaticalization, a critical assessment.
Special issue of Language Sciences (vol 23.2-3).
Comrie, Bernard (1989). Language Universals and Linguistic Theory, second
edition. Chicago, University of Chicago Press.
Dryer, Matthew (1992). The Greenbergian word order correlations. Language
Gilligan, Gary (1987). A cross-linguistic approach to the pro-drop
parameter. Ph.D. thesis, University of Southern California.
Greenberg, Joseph (1966). Some universals of grammar, with particular
reference to the order of meaningful elements. In Greenberg (ed.) Universals
of Language. Cambridge, MA: MIT Press, 73-113.
Haspelmath, Martin (1997). Indefinite Pronouns. Oxford: OUP.
Haspelmath, Martin (1999). Why is grammaticalization irreversible?
Linguistics 37, 1043-68.
Hawkins, John A. (1983). Word Order Universals. New York, Academic Press.
Hawkins, John A. (1994) A performance theory of word order and constituency.
Mallinson, Graham, Barry Blake (1981). Language typology. Amsterdam: North
Newmeyer, Frederick (1998). Language form and language function.
Cambridge,MA: MIT Press.
Song, Jae Jung (2001). Linguistic Typology: Morphology and Syntax. London:
Whaley, Lindsay (1997). Introduction to Typology: The Unity and Diversity
of Language. Sage Publications.