Publishing Partner: Cambridge University Press CUP Extra Publisher Login

New from Cambridge University Press!


Revitalizing Endangered Languages

Edited by Justyna Olko & Julia Sallabank

Revitalizing Endangered Languages "This guidebook provides ideas and strategies, as well as some background, to help with the effective revitalization of endangered languages. It covers a broad scope of themes including effective planning, benefits, wellbeing, economic aspects, attitudes and ideologies."

New from Wiley!


We Have a New Site!

With the help of your donations we have been making good progress on designing and launching our new website! Check it out at!
***We are still in our beta stages for the new site--if you have any feedback, be sure to let us know at***

Review of  Determinants of Grammatical Variation in English

Reviewer: Joybrato Mukherjee
Book Title: Determinants of Grammatical Variation in English
Book Author: G√ľnter Rohdenburg Britta Mondorf
Publisher: De Gruyter Mouton
Linguistic Field(s): Phonology
Subject Language(s): English
Issue Number: 14.3136

Discuss this Review
Help on Posting
Date: Fri, 14 Nov 2003 21:36:10 +0100
From: Joybrato Mukherjee
Subject: Determinants of Grammatical Variation in English

Rohdenburg, Guenter and Britta Mondorf, eds. (2003)
Determinants of Grammatical Variation in English,
Mouton de Gruyter, Topics in English Linguistics 43.

Joybrato Mukherjee, University of Giessen.

The volume under review goes back to a symposium on
"Determinants of Grammatical Variation in English", held in
Paderborn/Germany in June 2000. On more than 560 pages, it
provides a multidimensional perspective on grammatical
variation in English, i.e. on how it can be described and
explained as well as on how it may be implemented in
linguistic models. The editors are certainly right in
pointing out in their introductory remarks that the
"sixteen contributions selected for this volume are all
based on solid empirical research" (p. 1). But what makes
the volume particularly rich and stimulating is not just
the empirical, data-oriented methodology underlying all
analyses, but the fact that, firstly, virtually all the
contributors combine descriptive research with a discussion
of theoretical implications and that, secondly, they
usually do not confine themselves to one factor of
grammatical variation alone. Without any doubt, the book is
a goldmine for all functionalists because it brings
together a wide range of interesting approaches to the
description and analysis of grammatical variation in
English. In the following, I will briefly summarise the
contents of each paper. Afterwards, I will provide a brief
critical evaluation of the volume.


In the opening paper, Manfred Krug delves closely into the
role and importance of frequency as a determinant in
grammatical variation and change. He starts off by
discussing and exemplifying traditionally established
frequency-related concepts (such as irregularity and
analogy) as well as more recent concepts (e.g. entrenchment
in cognitive grammar) that try to map frequency in language
use on models of language. From the point of view of
language change in general and grammaticalisation theory in
particular, frequency is regarded as having two different
effects: conservative and progressive. That is to say,
frequent items (e.g. the irregular verb BE) tend to resist
regularisation much more persistently than infrequent
items, but it is also frequency that helps to trigger off
the grammaticalisation of forms (such as the contraction of
"it is" > "it's"). In the second part of the paper, Krug
gives various examples of the two effects of frequency. For
example, he discusses the cliticisation of HAVE and BE onto
pronouns in present-day spoken English by drawing on data
obtained from the British National Corpus, the Bank of
English and the London-Lund Corpus. It transpires that
there is a strong correlation between the frequency of the
combination of pronoun and auxiliary (i.e. its "String
Frequency") and the tendency to use the contracted form
(e.g. "I have" > "I've"). For various other examples, Krug
reveals that not only String Frequency, but also
Transitional Probability and the combination of the two
measures (an alternative formula which gives the percentage
of the frequency of a sequence of words in relation to the
frequency of the first word) are appropriate and good tools
for the description and prediction of such phenomena of
coalescence of adjacent linguistic items.

In taking issue with the clear-cut distinction between
syntax as part of a conceptual-intentional system and
phonology as part of an articulatory-perceptual system as
proposed by Chomsky (1995), Julia Schlueter argues that
phonology is not a syntax-external factor but a system-
internal determinant of grammatical variation. After
providing some neurophysiological background information on
what a realistic and biologically viable linguistic network
model should look like, Schlueter presents data from
various corpus analyses, including forms of BE in Chaucer's
language ("be/bee" vs. "ben/been") and the forms of the
indefinite article before -initial words (e.g. "an
historical" vs. "a historical"), and discusses why specific
instances of the attributive construction "Det + not + Adj"
are admissible while others are not (e.g. "a not unhappy
person" vs. *"a not happy person"). On the basis of her
findings, she postulates an "ideal syllable structure" of
the consonant-vowel (CV) type, which may help to explain,
for example, why "an" is used in front of -initial
adjectives whenever the corresponding /h/-sound is
devoiced. Furthermore, she formulates a "principle of
rhythmic alternation" which is intended to capture "a
prosodic tendency that is equally marked by alternating
structural patterns" (p. 88). According to this principle,
sequences of stressed syllables as well as sequences of
unstressed syllables are said to be avoided, resulting, for
example, in the inadmissibility of *"a 'not 'happy person"
due to a "stress clash". In the final resort, then,
Schlueter claims that phonological factors may well
function as determinants of grammatical variation at the
levels of both morphology and syntax.

The variation in the post-verbal order of constituents is
at the heart of Thomas Wasow and Jennifer Arnold's
contribution. They provide an in-depth discussion of
various factors that have been shown to influence the
choice between variants with or without dative alternation
(e.g. "Kim handed a toy to the baby" vs. "Kim handed the
baby a toy"), between variants with different verb-particle
positions (e.g. "We figured the problem out" vs. "We
figured out the problem") and similar constructional
alternatives. These factors include weight (i.e. more
complex vs. less complex constituents), information
structure (i.e. given vs. new information), semantic
connectedness (e.g. collocational or idiomatic links
between the verb and post-verbal elements as in "take into
account") and lexical bias (i.e. the preference of
particular verbs for specific constructions). The authors
also discuss the extent to which ambiguity avoidance may
function as a determinant of grammatical variation, e.g. in
speakers' preference of "Pat saw with a telescope a man in
a funny hat" over "Pat saw a man in a funny hat with a
telescope". Throughout the paper, Wasow and Arnold discuss
how individual factors strengthen or weaken the effect of
other factors (e.g. lexical bias and ambiguity avoidance).
Not only does the paper provide a multifactorial view of
constituent ordering on the descriptive level, but it also
makes a plea for a multimethod approach to syntax in making
use of corpus data and experimental data.

The concept of multifactorial analysis is a cornerstone of
Stefan Th. Gries's research into alternative particle
placements (e.g. "John picked up the book" vs. "John picked
the book up"). In a similar vein to Wasow and Arnold, Gries
argues that previous approaches have suffered, among other
things, from too rigid a focus on one particular
determinant, be it structural (e.g. complexity) or
functional (e.g. discourse familiarity of items). In
contrast, Gries takes into account a wide range of
morphosyntactic, semantic and discourse-functional factors
and offers a linear discriminant analysis of these factors
"in order to (i) weigh their importance for the dependent
variable (here: the choice of construction) and (ii)
predict the resulting value of the dependent variable
(here: verb-particle-object ordering or verb-object-
particle ordering)" (p. 164). On the one hand, Gries thus
offers a highly innovative and statistically sound
multifactorial methodology; on the other hand, the
predictive power of his findings prove that a strict focus
on morphosyntactic factors (as for example suggested by
Hawkins (1994) in his model of Early Immediate
Constituents) does not provide a comprehensive picture of
the determinants that are involved in native speakers'
choice between the verb-particle-object ordering and the
verb-object-particle ordering.

John A. Hawkins confirms the plausibility of Rohdenburg's
(1996) complexity principle by sketching out an explanation
in terms of processing efficiency. The complexity principle
states that the more explicit variant is preferred over the
less explicit variant in cognitively more complex
environments. With regard to the variation between
explicitly-marked phrases and their zero-marked
alternatives, the complexity principle implies that zero-
marked phrases must be positioned close to their heads
(e.g. "() he had done it" in "I realized () he had done it
with sadness in my heart"), while more distant positions
are only possible for explicitly-marked variants (compare
the admissibility of "I realized with sadness in my heart
that he had done it" as opposed to the dubious status of
?"I realized with sadness in my heart () he had done it").
By drawing on his own model of Early Immediate
Constitutents, Hawkins argues that zero-marked phrases need
to be adjacent to their heads because they can be assigned
semantic and/or syntactic properties only by the
superordinate node being parsed. For example, while the
explicitly-marked phrase "that he had done it" is clearly
marked as a subordinate clause (being one essential
syntactic property), the zero-marked variant "() he had
done it" is ambiguous in this respect and is thus
cognitively more complex because the clarification of its
status "depends", as it were, on the parsing of its head.
Hence the parsing preferences as predicted by the model of
Early Immediate Constituents corroborate the applicability
of the complexity principle to alternations between
explicitly-marked phrases and their zero-marked

Unsurprisingly, the complexity principle plays a central
role in Rohdenburg's own paper as well, in which he
analyses the factors that are responsible for the variation
in the use of interrogative clause linkers after nouns,
verbs, adjectives and phrasal expressions. In this context,
Rohdenburg distinguishes between zero links (e.g. "She was
at a loss () what to do"), prepositional links (e.g. "She
was at a loss about/as to what to do") and verbal links
(e.g. "She was at a loss to know what to do"). Rohdenburg
draws on various corpora and other databases and argues on
the basis of many quantitative analyses that the
variability can be explained to a large extent by the
complexity principle and the "horror aequi" principle. As for
the complexity principle, Rohdenburg argues in particular
that "novel and more explicit structures are first
established in more complex environments" (p. 217), as for
example in the transitive use of the verb "check" (e.g. "He
checked the car to see whether ...") as opposed to the
intransitive use of the same verb, which is often marked by
zero links (e.g. "He checked whether ..."). Additionally,
Rohdenburg claims that "more explicit recessive structures
survive longer in contexts involving an increased
processing load" (p. 217). As for the verb "depend", for
example, he points out that the prepositional link tends to
be used much more frequently in discontinuous structures
(e.g. in "It entirely depends, after all, on how they go
about it") than in structures without intervening elements
on either side of the verb. The horror aequi principle
captures the tendency, among other things, to avoid the use
of formally identical grammatical structures in immediate
adjacency. This principle, Rohdenburg claims, may help
explain, for example, the frequent use of "... to wait and
see ..." instead of the double infinitive "to wait to see

Britta Mondorf's paper, too, is inspired by the complexity
principle. She draws on this principle in order to identify
the determinants of the alternation between the synthetic
comparative form of the adjective (e.g. "readier") and its
analytic variant (e.g. "more ready"). Her overall claim is
captured by the notion of "more-support: In cognitively
more demanding environments which require an increased
processing load, language users tend to make up for the
additional effort by resorting to the analytic (more)
rather than the synthetic (-er) comparative" (p. 252). By
drawing on quantitative data from the British National
Corpus and various newspaper archives for 28 adjectives,
she identifies several relevant determinants (e.g. the
length, the final segment and the frequency of the
adjective) and quantifies the contribution of each factor
to the alternation between synthetic and analytic forms.
For example, it is shown that while the synthetic option is
preferred with generally frequent adjectives, the analytic
variant with "more" tends to be used with infrequent
adjectives, which are more difficult to access: "More-
support by separating form and function then serves to
apportion the otherwise complex expression of adjective and
degree marker in one lexeme" (p. 260). Mondorf not only
takes syntactic factors into consideration, but also
discusses morphological, phonological, lexical, semantic
and pragmatic factors. She is thus able to offer an
unprecedentedly comprehensive picture of what may lead
language users to opt for the analytic comparative in
specific contexts.

Uwe Vosberg is concerned with diachronic effects of the
complexity principle and the horror aequi principle on the
evolution of clausal verb complements. In this context, he
places special emphasis on extractions, i.e. non-canonical
syntactic structures that produce filler-gap dependencies
and, accordingly, traces (cf. i < ti) as in "Now, how
many(i) do you remember to have heard named(ti)?". On the
basis of several case studies (e.g. the complementation of
the verb "remember" in Early Modern and Modern English),
evidence is provided for Vosberg's "Extraction principle"
(p. 308), which states that infinitival complementation
tends to be preferred over gerundial complementation in the
cognitively complex environments of syntactic extractions.
Vosberg also shows that gerundial complementation after
ing-forms of the verb are generally avoided, thus
corroborating the relevance of the horror aequi principle
as another determinant in the variation between variants of
clausal verb complementation.

The alternation between infinitival and gerundial
complementation is also discussed by Christian Mair, though
with an explicit reference to the verbs "begin" and
"start". Mair starts off from an outline of some
explanations that have so far been offered for the
complementation of the two verbs at hand. For example, he
refers to the well-known constraints on gerund complements
(e.g. when an adverbial is placed after "begin" and
"start", resulting in the inadmissibility of *"began in the
following years selling"), and to the different levels of
formality (i.e. the preference for "start" in informal
contexts and for "begin" in formal contexts), which is also
corroborated by corpus findings (cf. Biber et al. 1999:
373, 747). By referring to various corpora of British and
American English in the 1960s and 1990s as well as other
databases, Mair argues that so far the importance of
variation as a determinant in its own right has been
largely ignored. For example, while the infinitive after
"begin" has remained the default case and statistical norm
in British English, a significant shift towards the
gerundial complement can be noted in American English from
the early 1960s to the early 1990s. In general, it seems
that complement choice after "begin" and "start" is a good
example of on-going language-change and, accordingly, a
field in which a clear distributional pattern of use has
not (yet?) emerged. Rather, the situation is characterised
by synchronic - regional and stylistic - variation amidst a
process of diachronic change.

The variation between complement clauses is also the topic
of Dirk Noel's paper. In particular, he focuses on the
alternation between infinitival and finite complements
after verbs of the type of 'believe', e.g. "I believe Mary
to be dishonest" vs. "I believe that Mary is dishonest". In
the light of data obtained from the British National
Corpus, he takes issue with many semantico-syntactic
approaches to grammar (which he considers to be, in his own
words, examples of "semantic extremism"). For example,
Wierzbicka (1988: 26) claims that "ALL contrasts between
TO, ING and THAT can be accounted for in terms of meaning".
However, Noel argues forcefully and convincingly that the
different semantic representations of the different clause
types that are offered by Wierzbicka (1988) do not allow
for a semantically clear-cut distinction between, say, "I
believe Mary to be dishonest" and "I believe that Mary is
dishonest": the variation between the fused construction
with the to-infinitive on the one hand and the that-clause
on the other may display a difference in the use of the two
constructions, but this pragmatic difference is not
primarily semantically motivated. Furthermore, Noel shows
that the structures that are deemed dubious by Wierzbicka
(1988) on semantic grounds (for example, she claims that
"John knows Mary to be a Mormon" is not fully acceptable
because "KNOW (X) to be" expresses "personal experiential
knowledge" and is thus used with first-person subjects
only) are far from being rare in the British National
Corpus: in actual fact and contrary to Wierzbicka's
predictions, in the case of "KNOW (X) to be", 58% of all
instances have a third-person subject. It thus seems that
many semantic determinants of grammatical variation in the
field of clausal complementation that have so far been
offered on purely intuitive grounds cannot be shown to be
relevant to the distribution of forms and structures in
real corpus data.

The alternation between the s-genitive and the of-genitive
in English (e.g. "the boy's eyes" vs. "the eyes of the
boy") is scrutinised in Anette Rosenbach's paper. She takes
into consideration the following three factors: (1) the
animacy of the possessor, (2) the topicality of the
possessor, (3) the type of possessive relation. In order to
assess in an elicitation experiment with British and
American speakers to what extent these factors influence
the choice between the two genitives, she first of all
identifies the features of genuine "choice contexts" (e.g.
the realisation of the possessor as a full lexical noun
phrase) and - within these choice contexts - the features
of truly "comparable contexts" (e.g. with a possessor not
ending in /s/ or /z/ because in these phonological contexts
the s-genitive is usually avoided). Finally, in designing
the experiment she defines the above-mentioned three
factors in terms of prototypical binary choices, e.g.
personal nouns vs. concrete nouns for animate (+a) vs.
inanimate (-a) possessors. The experiment, in which native
speakers were asked to spontaneously choose between s-
genitive and of-genitive in given, comparable choice
contexts, reveals that in general the importance of the
three factors can be ranked as follows: animacy >
topicality > possessive relation. Rosenbach also infers
from the data that there is an "ongoing change towards an
increasing use of the s-genitive with (-animate)
possessors" which "is not lexically restricted" (p. 399).
She concludes her paper by explaining the findings in terms
of increasing cognitive economy, which may or may not be
linked to iconicity in grammar.

Interestingly enough, the article by Anatol Stefanowitsch
also deals with the two genitives in English. However, his
approach - including basic assumptions, the underlying
theory, methodology and the overall results - is
fundamentally different. To begin with, Stefanowitsch takes
issue with what he calls the "discourse-functional
hypothesis", which views the two genitives as discourse-
pragmatic or information-structural alternatives but
regards them as more or less semantically equivalent. He
rejects this approach also on grounds of the outcome of a
quantitative analysis of 100 examples obtained from the
Corpus of Spoken Professional American English which do not
bear out the predictions of the discourse-functional
hypothesis. In contrast, he argues by putting forward a
"semantic hypothesis" that "the two genitives are
semantically distinct constructions, whose primary function
is the assigning of semantic roles to their head and
modifier slots" (p. 414). Stefanowitsch's approach is thus
firmly based on construction grammar (cf. Goldberg 1995).
Accordingly, he argues that the two genitive constructions
assign different semantic roles to the two components:
while the s-genitive is said to assign the roles of
POSSESSEE and POSSESSOR to its head and modifier, the of-
genitive is understood to assign the roles of ENTITY and
INTRINSIC ENTITY respectively. He sketches out an
elaborated model which, among other things, also takes into
account that specific lexical items can override the
typical semantic roles. For example, Stefanowitsch notes
that while *"the shoes of Kate" is not possible because
"shoes" does not evoke an ownership relation, "the budget
of the university" is admissible because "'budget' already
evokes a relation of ownership and can thus override the
semantics of the of-genitive" (p. 432).

The determinants of grammaticalisation both in formal and
functional theories are discussed in detail by Olga C.M.
Fischer. In the first part, she critically reviews some key
issues and concepts in current grammaticalisation theory:
the idea of unidirectionality, conceptual chains (i.e. the
semantic motivation of grammaticalisation), the question of
whether grammaticalisation itself is a determinant of
language change or whether it denotes an epiphenomenon, and
the question of what the parameters of grammaticalisation
are. In the second part, Fischer discusses two case studies
(i.e. the grammaticalisation of infinitival "to" and of
"have to") in order to show that processes of
grammaticalisation are not only complex but also highly
individual so that it is difficult - if not impossible - to
abstract away general principles or determinants. For
example, she shows that while semantic bleaching of "have
to" was a prerequisite for its grammaticalisation, semantic
bleaching occurred only simultaneously with other factors
in the early stages of the grammaticalisation of
infinitival "to". In a wider setting, Fischer also calls
into question the universal nature of other principles of
grammaticalisation that have been repeatedly suggested in
the literature, especially unidirectionality. For example,
in late Middle English and in Early Modern English, the
grammaticalisation of "to" to an infinitival marker was to
some extent reversed: the appearance of split infinitives,
for example, is indicative of a process of partial
degrammaticalisation. It is thus evident that not all
general parameters of grammaticalisation are present in all
actual processes of grammaticalisation because, as Fischer
pointedly concludes, "(i)n the real linguistic world many
rules are no more than tendencies" (p. 469).

Peter Siemund deals with grammatical variation in the field
of intensifiers and reflexives from both a cross-linguistic
and a dialectological perspective. By combining the
perspectives of linguistic typology and dialectology,
Siemund sets out to show that synchronic and diachronic
grammatical variation in English follows cross-linguistic
generalisations to a very large extent. He thus maps
general processes of the grammaticalisation of reflexive
marking on the diachronic change of the English language
and, what is more, on the synchronic present-day variation
between dialects of English. For example, the variation
between diachronic forms of English without reflexives and
those with reflexives (e.g. Old English vs. Modern Standard
English) can still be found in the synchronic variation
between Standard English (e.g. "He has cut himself") and
non-standard dialects (e.g. "He has cut him" in Yorkshire)
and also in the variation between Standard English on the
one hand and English-based pidgins and creoles like Sranan
(e.g. "a kil hem" = "he has killed himself"), which reflect
the original Old English situation, on the other. The paper
also provides some interesting comments on - and examples
of - "free self-forms", i.e. self-forms that are not in the
same local domain as their antecedent, as for example in
"Of course most of us, including myself, will accept the
democratic decision". Siemund convincingly suggests an
analysis of such free self-forms as headless adnominal

Principles of linguistic typology and the description of
dialects of English are also fruitfully combined by
Lieselotte Anderwald. Her focus is on negation in non-
standard English from the point of view of markedness
theory. Following Greenberg (1966) and Croft (1990), she
takes into consideration the following markedness criteria:
zero value (S1: unmarked = zero), syncretisation (S2:
unmarked = more forms), irregularity (S5: unmarked = more
irregular) and frequency (S8: unmarked = more frequent). In
Standard English negation, criteria S1 and S8 are
satisfied, while S2 and S5 are not. Anderwald then
discusses the negation in non-standard English by zooming
in on the frequency and use of "don't", "ain't" and
"wasn't/weren't". Interestingly, some dialects display a
crossover pattern and turn out to generalise "was" in
positive contexts but "weren't" in negative contexts. In
general, Anderwald reveals that while Standard English
satisfies only two of the above-mentioned markedness
criteria, many non-standard varieties conform to all four
markedness criteria because the marked negative paradigm
can be shown to have fewer forms (criterion S2) and to be
less irregular through simplification (criterion S5):
often, "ain't" is the only negative counterpart to
"am/are/is" and to "have/has", "don't" the only negative
counterpart to "do/does", and "weren't" the only negative
counterpart to "was/were".

In the final paper, Sali A. Tagliamonte explores the
variation between "have", "have got" and "got" for stative
possessive meaning in three varieties of English: Buckie
(Northern Scotland), Wheatley Hill (Northeast England) and
the city of York. Tagliamonte starts off by discussing the
extent to which determinants such as contraction, negation
and the distinction between abstract and concrete objects
exert an influence on the choice of one of the three
variants in spoken language. On the basis of a multivariate
statistical analysis of the data, the conclusion is drawn
that the factors that are responsible for the variation are
very much the same across the three communities at hand.
This finding is interpreted by Tagliamonte as an indication
of "persistence" and as a vindication of Kroch's (1989)
"constant rate hypothesis which holds that grammatical
constraints will hold constant over time despite the fact
that a form or construction may be taking over from
another" (p. 550).

The book is concluded with an author index and a subject


As stated at the beginning, Rohdenburg and Mondorf have no
doubt succeeded in putting together a very interesting
collection of high-quality articles. As the synopsis
reveals, the book includes a wide range of quite different
- complementary as well as conflicting - approaches to the
description and analysis of grammatical variation in
English. What they all have in common is an affinity for
functional explanations of variation rather than formalist
and rule-governed accounts. It is a particular strength of
the present volume that practical studies are complemented
throughout with theoretical discussions and linguistic
model-building (e.g. in Fischer's discussion of
grammaticalisation theory, in Rohdenburg's abstraction of
the complexity principle and the horror aequi principle,
and in Noel's hilarious and convincing attack on extreme
semantico-syntactic approaches to grammatical variation).
The proof-reading is almost perfect. There are less than a
handful of blunders that have come to my notice (e.g. the
wrong number of words given for the BNC on p. 298, which is
not 10+900 million words in size but 100 million words; *"3
instance" on p. 552). The contributors and the editors have
ensured an easily accessible style and a clear and easy-to-
follow line of argumentation in all articles.

It goes without saying that a collection of articles, in
particular whenever it is based on and/or inspired by
papers previously read at a conference, cannot cover all
aspects that would be relevant to the overall topic in a
systematic and comprehensive way. Nevertheless, I would
like to point up four issues that could have been addressed
in more detail in general and in individual papers in

Firstly, it is not always explicitly stated why which data
were used for linguistic analysis. For example, it is not
straightforwardly clear why phonological claims and
conclusions should be based on spoken data that were
orthographically encoded in the written medium (e.g. the
British National Corpus in Krug's paper) or on written
newspaper archives (e.g. The Guardian in Schlueter's
paper). The question arises, for example, whether the
orthographic representation of spoken data is the best
source for a quantitative analysis of contractions in
spoken English if the original spoken data had not been
directly accessed (if they had, this should have been
explicitly mentioned). Also, some articles provide only
vague information on the database that has been used. For
example, it is not made clear by Stefanowitsch on what
grounds he chose the 100 examples from the Corpus of Spoken
Professional American English and, accordingly, what the
quantitative analysis of these examples is intended to

Secondly, one sometimes wonders whether the implicit
assumption of semantic equivalence between grammatical
variants is entirely true to the facts. A good case in
point is Rohdenburg's discussion of "wait to see" and "wait
and see". It is at least worth-discussing whether the two
constructions are genuine variants - or whether they are
semantically non-equivalent in the first place (thus
picking up, in a sense, on Stefanowitsch's semantic
hypothesis). In this context, it should also be mentioned
that not all of the many principles suggested by various
authors are convincing in the light of slight variations of
the examples that the authors themselves provide. For
example, Schlueter's principle of rhythmic alternation may
explain why *"a 'not 'happy person" is blocked due to a
stress clash (unlike "a 'not un'happy person"), but it
fails to account for the admissibility of "a 'not 'impolite
person", marked by a similar stress clash.

Thirdly, it is not always clear how the actual and
individual speaker (and his/her choice of grammatical
variants) and the more abstract population (and equally
abstract determinants of grammatical variation at this
level) are related to each other. It is only Fischer in her
discussion of grammaticalisation developments who goes into
detail about the problem of how to reconcile the focus on
what individual speakers do on the one hand (and their
competence/performance) and on what happens at the level of
the supra-individual speech community (and the abstracted
competence/performance of a population) on the other.
Whenever determinants of grammatical variation are
described, it might be useful to also discuss whether and
to what extent a determinant is relevant to all, some or
only specific individidual speakers. It is strange that
idiolectal variation has been largely ignored in many
papers to which this level of variation might have been

Fourthly, there are only few indirect hints at chance as a
determinant of language variation (cf. Butters 2001),
including the possibility of a linguistic butterfly effect
(cf. Schneider 1997). Only Mair, in discussing the
complementation of "begin" and "start", explicitly refers
to the problem of how to deal with variation at a stage at
which functional motivations for variation are yet to
emerge. But the question of why there should be variation
in the first place is not discussed in detail. The crucial
point in this context seems to me that above and beyond
highly abstract principles such as Rohdenburg's complexity
principle and more fine-grained patterns of language-
internal (regional or stylistic) variation as discussed by
Mair there always is variation in language not only because
of principles, rules and regularities but just because
language is by nature variable (cf. Sapir 1921: 147). And
this very nature of language, based on chance, chaos or
other pre-linguistic driving forces, should probably be
regarded as one of the most basic determinants of any kind
of variation in language.

In spite of these critical remarks on several aspects that
could have been covered in more detail in the book under
review (and to which more attention might be paid in future
research in this field), I am sure that the volume will
prove stimulating and thought-provoking for all
functionally-oriented linguists who are interested in how
to come to grips with the variability of forms and
structures in English grammar.


Biber, D., S. Johansson, G. Leech, S. Conrad and E. Finegan
(1999): The Longman Grammar of Spoken and Written English.
Harlow: Pearson Education.

Butters, R.R. (2001): "Chance as cause of language
variation and change", Journal of English Linguistics 29,

Chomsky, N. (1995): The Minimalist Program. Cambridge, MA:
MIT Press.

Croft, W. (1990): Typology and Universals. Cambridge:
Cambridge University Press.

Goldberg, A. (1995): Constructions: A Construction Grammar
Approach to Argument Structure. Chicago, IL: The University
of Chicago Press.

Greenberg, J. (1966): Language Universals. The Hague:

Hawkins, J.A. (1994): A Performance Theory of Order and
Constituency. Cambridge: Cambridge University Press.

Kroch, A.S. (1989): "Reflexes of grammar in patterns of
language change", Language Variation and Change 1, 199-244.

Rohdenburg, G. (1996): "Cognitive complexity and increased
grammatical explicitness in English", Cognitive Linguistics
7, 149-182.

Sapir, E. (1921): Language: An Introduction to the Study of
Speech. New York, NY: Harcourt.

Schneider, E.W. (1997): "Chaos theory as a model for
dialect variability and change?", Issues and Methods in
Dialectology, ed. A.R. Thomas. Bangor: University of Wales
Bangor, Department of Linguistics. 22-36.

Wierzbicka, A. (1988): The Semantics of Grammar. Amsterdam:
John Benjamins.
About the REVIEWER

Joybrato Mukherjee is Professor of English Linguistics at
the University of Giessen, Germany. His research interests
include applied linguistics, corpus linguistics,
intonation, stylistics and syntax.