Pintzuk, Susan, George Tsoulas and Anthony Warner, eds. (2000) Diachronic
Syntax: Models and Mechanisms. Oxford University Press, paperback ISBN:
0-19-825027-4, xii+380pp, GBP24.99.
Malcolm Ross, Research School of Pacific and Asian Studies, Australian
This book is a collection of fourteen papers on the
diachronic syntax of mostly European languages, with an
introductory contribution by the editors. Except where
indicated below, the contributions are in a Principles and
It seems to me that there are two different but overlapping
enterprises that lay claim to the term 'diachronic syntax'.
The practitioners of one operate in the P&P/Minimalist
framework, the practitioners of the other in the more
conservative framework outlined, for example, by Harris and
Campbell (1995). More of this below. Most of the
contributors to this book pursue the first of these
enterprises, whilst I pursue the second. I mention this
simply to indicate that my emphases may be somewhat
different from those of a reviewer working in the first
**Susan Pintzuk, George Tsoulas and Anthony Warner,
'Syntactic change: Theory and method'
This chapter is far more than a standard introduction: it is
a good overview of the issues currently facing historical
syntacticians working in the framework of the Minimalist
The editors begin by pointing out that the standard position
within diachronic Chomskyan syntax, emphasising the role of
acquisition in change, needs to be complemented by the study
of diffusion. This study has taken off in recent years
because electronic text corpora allow statistically founded
syntactic analysis and the interpretation of syntactic
changes in 'E(xternalised) language' as evidence of changes
in 'I(nternalised) language'. The present book contains a
number of such studies (all except Vincent, Briscoe,
Martins, Whitman and Batllori & Roca). Although diffusion
consists in part of the successive acquisition by children
of a change that has already arisen, this is not the whole
story. The trick, as it were, is to find explanations for
variation and diffusion which fall out from general
syntactic theory and do not require appeal to 'historical
The editors outline the relationships between diachronic
syntax and (i) P&P and (ii) the Minimalist Program. In the
P&P framework, the acquisition task includes determining the
language's settings of a small number of parameters within
Universal Grammar (UG), and syntactic change consists in
learners analysing the input data differently from their
forebears, thereby resetting parameters for the acquired
language. In the Minimalist Program, however, the locus of
syntactic change is confined to 'operation O', which forms
lexical items out of phonological, semantic and syntactic
features: 'change can be described simply as the
reorganization of the featural content of the lexical items
of the language' (p7). Loss or gain of movement occurs when
learners revise the featural content of functional lexical
items, and grammaticalisation can be understood as a
revision that results in structural simplification (cf
Lightfoot (1979) showed that a cluster of related changes in
E-data may reflect a single change in parameter setting in
I-language, but the editors point out that other inferences
from such E-data are possible. One change may simply remove
the environment for some other feature (cf Williams below).
There is considerable support for Kroch's (1989a, 1984b,
1994) 'Constant Rate Effect', the insight that, although at
a given time an ongoing change may be at different stages in
different grammatical contexts, its rate of change in these
contexts will be the same (exemplified by Hr�arsd�ttir's
contribution to this book).
Returning to the issue of diffusion, the editors point to
two basic explanations of variation. One is that speakers
switch between two competing grammars, as proposed by Kroch
(1989a, 1994) Santorini (1992) and Taylor (1994) (see also
Batllori & Roca and Han below). The other is that the
grammar includes variants that are equally costly, as
suggested by van der Wurff (1997) (see also Delsing below).
These two possibilities may interact with sociolinguistic or
processing pressures or, as Briscoe suggests in this volume,
with the inductive bias of the learner towards one parameter
setting rather than another. The editors' discussion here
draws our attention to the complexity of the issues raised
by diffusion, and ends with a brief look at the treatment of
grammaticalisation within the Minimalist Program.
The remainder of the introduction summarises the
contributions to the volume and notes their relevance to the
***Part 1: 'Frameworks for the understanding of change'
**Nigel Vincent, 'Competition and correspondence in
syntactic change: Null arguments in Latin and Romance'.
This outstandingly erudite chapter is an essay on method.
The case study around which it is built has to do with
changes in the licensing of null and overt pronominal
arguments as Latin developed into the Romance languages --
but the data are well enough known: Vincent's objective is
to argue for the concepts of Competition (drawn from
Optimality Theory [OT]) and Correspondence (in its
Lexical-Functional Grammar [LFG] sense) and thus for an
OT/LFG-based approach to change in linguistic systems.
'Competition' captures the intuition that at any given time
a language consists of (at least potentially) competing
forms and subsystems which are the bases of change,
including grammaticalisation paths. By modelling changes in
subsystems as OT-style changes in constraint rankings,
Vincent argues that it is possible to capture these
phenomena in a precise manner -- a goal that has otherwise
largely evaded practitioners of formal approaches.
'Correspondence' in LFG refers to the relationship between
the representation of abstract grammatical relations and
their expression in linguistic form, whether syntactic or
morphological. Where the same universal set of features for
pronominal content is expressed in different ways over time
- as zero, as dependent forms with varying degrees of
boundness and varying positions, or as independent forms --
an LFG-type approach allows a cleaner account than a
configurationally based system like P&P.
**Ans van Kemenade 'Jespersen's cycle revisited: formal
properties of grammaticalization'.
Van Kemenade sets out to show that grammaticalisation is a
morphosyntactic process rather than a semantic one. Her
material is drawn from negation in English, which has seen a
cyclic process whereby a negative morpheme has undergone
morphosyntactic weakening, associated with the rise of a new
negator which subsequently also undergoes weakening. In her
P&P framework, each negator begins life as a specifier, then
becomes a functional head, and finally an inflection.
Crucial to her argument is the obervation that Old English
(OE) main-clause-initial _ne_ usually blocked
topicalisation, indicating that it occupied the topic
(specifier of CP) position itself. At the same time, it was
cliticised to an immediately following verb. That is, it
was phonologically weakened, but still a constituent with no
change in semantic content, implying that the weakening
process precedes semantic change and is not driven by it.
Van Kemenade mentions, for example, the modern English
morphosyntactic difference between 'Didn't they warn you?'
and 'Did they not warn you?', handed down from OE, where
there is no semantic contrast.
Van Kemenade (p50) attributes to grammaticalisation
theorists, and to Hopper and Traugott (1993) in particular,
the claim that grammaticalisation is semantically driven.
It lies beyond the scope of this review to discuss this
issue, but a reading of pp66-67 of Hopper and Traugott
suggests that this is not quite the position of its authors
and that they would not necessarily be at odds with van
**Ted Briscoe, 'Evolutionary perspectives on diachronic
Briscoe's subject matter is the way in which changes spread
through a speech community. He first models language
acquisition and the situation of the learner who receives
input from two conflicting grammars that are present in the
community as the result of ongoing change, then he models
the spread of a change through the community. He argues
that (E-)languages are complex adaptive systems which
'evolve', in the technical sense of the word: there is
random variation and selection from among variants, leading
to differing inheritances. This selection, however, is
biassed in the direction of innate UG.
Drawing on his own recent work, he takes the case of
Hawaiian Creole and shows that on the model he proposes
there is nothing fundamentally exceptional about its
development, as long as one accepts UG-based bias.
Importantly, this places glottogenesis within a coherent
theory of language change.
Every other contribution to the book focusses on changes in
the grammars of individual languages, and most deal with a
set of changes in a single language or a small set of
closely related languages. Although some of them touch on
the concept of competing grammars (see the Introduction),
Briscoe's contribution is radically complementary to them,
because he tackles the _mechanism_ of diffusion of change
through a population. This issue has been addressed in the
literature before, in Keller's (1990) application of the
'invisible hand' to language change and in linguistic uses
of social network theory (Le Page and Andr�e Tabouret-Keller
1985, Milroy and Milroy 1985, Milroy 1987, Ross 1997), but
Briscoe is innovatory both in offering a formal theory of
diffusion and in integrating it with a theory of language
This chapter is very different from the others not only in
its subject matter, but also in its approach. There are no
linguistic data. Instead, Briscoe builds a formal
theoretical model, using a Categorial Grammar formalism to
depict language change and a Bayesian statistical approach
to learning and diffusion. This is daunting for the reader
whose staple fare is typified by the rest of the book, but
the issues Briscoe deals with are crucial to historical
linguistics, and I am glad that writing this review made me
grapple with their presentation.
****Part 2: 'The comparative basis of diachronic syntax'
These chapters are based on corpus analysis and offer
explanations of syntactic phenomena. Given the
configurational basis of P&P, this means determining what
positions are occupied by what constituents of the rather
challenging structures of Old and Middle English.
**Eric Haeberli's 'Adjuncts and the syntax of subjects in
Old and Middle English'
Haeberli focuses on the adjacency or otherwise of the verb
and subject in V2 clauses in Germanic languges. Some of
these languages allow an adjunct to intervene between finite
verb and (non-pronominal) subject, whilst others don't.
Haeberli provisionally labels the two maximal projections
below CP as XP and YP. The fronted constituent of a normal
V2 clause is at Spec CP, the verb at C. In a V2 clause with
a pronominal subject (e.g. Wahrscheinlich [V wird] [SU er]
[A sp�ter] dieselbe Uhr kaufen), the subject (SU) is at Spec
XP, the adjunct (A) at X, and in a clause with a
non-pronominal subject (e.g. Wahrscheinlich [V wird] [A
sp�ter] [SU Hans] dieselbe Uhr kaufen) the subject is at
Spec YP. Haeberli asks, What are XP and YP? Old English (OE)
data are crucial to his answer. Like other V2 Germanic
languages, OE has verb movement to C, but in just OE the
verb instead moved under certain conditions only to X,
sometimes indicated by a preceding pronoun subject at Spec
XP. Thus the adjunct in normal Germanic V2 clauses and the
verb in the special OE clauses occupy the same position, X.
Haeberli then compares southern and northern Early Middle
English (EME). The southern dialects had the same verb
movement possibilities as OE, but the northern dialect had
only movement to C, as in modern V2 Germanic, and no adjunct
intervening between verb and subject. The subject agreement
system of verbal suffixes was impoverished in the northern
dialect, and Haeberli sees this as evidence that the AgrSP
of Old and southern Middle English (ME) had vanished,
leaving only one position for the northern verb. That is,
Old English XP was AgrSP and, by inference, YP was TP. In
Northern Middle English the next projection below CP was TP:
there was no AgrSP, only one postverbal subject position,
and no intervening adjunct.
Haeberli's account is ingenious, but leaves me with two
questions. First, a German non-pronominal subject can occur
in either of the two positions mentioned above (i.e. the
adjunct can occur before or after the subject). This
analysis requires me to believe that there is a fundamental
difference in constituent structure between these two
versions of the clause: this feels like a heavy analytic
price to pay for a small surface difference. Second, if the
impossibility of an intervening adjunct is diagnostic of
loss of AgrSP, its impossibility in Icelandic, where
verb--subject agreement remains, is unexplained.
**Anthony Kroch and Ann Taylor, 'Verb--Object order in Early
Kroch and Taylor report on an extensive quantitative
analysis of consitituent order in West Midland and Southeast
EME manuscripts. Their goal is to determine the underlying
order(s) in these data after the effects of movements like
leftward scrambling and rightward extraposition have been
taken into account. They work towards this goal using a
combination of statistical analysis and syntactic
diagnostics and show first that EME texts have a small
remnant of INFL-final word order (in non-P&P terms, the
finite auxiliary is final), even when stylistic fronting is
eliminated from the data set. They show that underlying OV
word order (where V is the finite or non-finite verb, not
the auxiliary) is hard to establish, as leftward scrambling
of the object in a VO clause mimics OV order. However, they
conclude that there is a difference in the scrambling
behaviour of quantified and non-quantified objects:
quantified objects scramble regularly, non-quantified don't.
Their data show that 30% of non-quantified objects are
preverbal, and they infer that these represent OV order.
Finally, Kroch and Taylor show that leftward scrambling of
pronouns outlasted OV word order and that a preverbal
pronoun object is not diagnostic of OV order.
The importance of Kroch and Taylor's contribution is that it
supports the notion that syntactic change is not sudden but
occurs via competition. There is continuity between Late OE
and EME syntax, with the more conservative Southeastern and
less conservative West Midlands texts differing in their
rates of change
**Alexander Williams, 'Null subjects in Middle English
Williams argues on the basis of distribution that the types
of existential sentence from which expletive 'there' is
missing can be divided into sentence types which lack
'there' altogether and sentence types where it follows the
verb but is unpronounced. Sentences where 'there' is
unpronounced are relatively common until 1250, but drop off
rapidly after that date. Williams associates this with the
fact that verb-initial sentences drop off markedly at the
same time. Since the environment of unpronounced 'there' is
post-verbal, he argues that it disappears simply because its
environment has disappeared.
****Part 3: 'Mechanisms of syntactic change'
***Section 1: 'Features and categories'
Both the papers in this section address problems which are
not readily solved by movement analyses. Instead, they
posit changes in features (sub-categorial in Martins' case,
categorial in Whitman's) at terminal nodes.
**Ana Maria Martins, 'Polarity items in Romance:
underspecification and lexical change
This paper examines two uses of negative indefinites (e.g.
Spanish _nadie_ 'nobody', _ning�n_ 'no (one)', _nada_
'nothing') in a range of Romance languages. Preverbally,
these items are always negative in meaning, and Martins
examines whether the simple negator (e.g. Spanish _no_
'not') must cooccur with them or not. Postverbally, in
'modal' contexts (questions, imperatives, conditionals,
certain subordinate clause constructions), most Romance
languages allow these items to be used with indefinite
(non-negative) meaning (e.g. Spanish _Dudo que venga
nadie/alguien_ 'I doubt that anybody is coming').
If coocurrence of the negator with a preverbal negative
indefinite is labelled as (1) and postverbal non-negative
use as (2), then in the earliest stages of Old Romance, (1)
was obligatory, and (2) was acceptable. The histories of
the Romance languages were largely independent of each
other, but drift was in the same direction, and no modern
language retains the earliest Old Romance situation. In
Modern Romanian and Venetian, (1) remains obligatory, but
(2) is unacceptable. In later Old Romance and modern
Catalan, (1) is optional, (2) acceptable. In Galician,
Spanish, Italian and French (1) is unacceptable and (2)
acceptable. And in Portuguese, the most innovative language
in this regard, (1) and (2) are both unacceptable. Martins
argues with regard to (2) that negative and positive
indefinites are in competition and that the non-negative use
of negatives loses out to the more explicit positives.
Within the Minimalist framework, Martins proposes to deal
with these changes by changes in features whose attributes
are affirmative, negative and modal, their values '+', '0'
(un[der]specified) and 'alpha'. For example, the negative
indefinites of Portuguese are always 0 aff, + neg, 0 mod (as
they are always negative in their own right), whereas those
of earliest Old Romance are always 0 aff, alpha neg, alpha
mod, as their negative reading is always dependent on the
presence of a negator and modal use determines their reading
in that context.
**John Whitman, 'Relabelling'
In the only paper in the volume to tackle a theoretical
issue in diachronic syntax from a perspective embracing
unrelated languages, John Whitman argues that reanalysis
does not entail syntactic restructuring but 'relabelling'.
His point is that in a P&P approach, reanalysis entails
changes in deep but not surface structure. In the
Minimalist approach, there is no deep structure, and so this
possibility is unavailable. Instead, he argues, reanalysis
is relabelling, i.e. change in the categorial feature of a
head, but not in syntactic structure outside the minimal
domain of the relabelled item (cf the Introduction to the
Whitman examines cases from a variety of languages. He
begins with two cases from Ewe, verb-to-preposition and
verb-to-complementiser, then moves on to a potential
counterexample, the putative [for + NP][to + verb] to [for
[NP + to + verb]] reanalysis in English. He shows that this
is a straw man -- it represents a misinterpretation of the
diachronic data. His interpretation of verb-to-preposition
reanalysis in serial verb constructions requires that only
the second, and never the first, verb may become a
preposition. Again, he deals with counterexamples (the most
important being the object-marking 'preposition' _ba_ in
Mandarin) suggesting in each case that the synchronic
analysis which posits a preposition is wrong. Finally, he
defines two kinds of structural changes that do occur within
the minimal domain of the head, illustrating them with
Mandarin and Saramaccan cases. The first is 'pruning',
which, for example, gets rid of the node from which Spec VP
(the subject of the verb) branches when the verb becomes a
preposition. The other is its converse, specifier-to-head
This is an important paper because it tackles an issue which
is central to accounts of grammaticalisation. What does
happen in the grammar when a verb becomes a preposition? In
any framework which is constituency-based and which posits
structural change to explain reanalysis, there is the
difficult issue of how this change takes place, especially
when ungrammaticalised and grammaticalised stages co-exist.
Whitman's hypothesis reduces this difficulty.
***Section 3: 'Movement'
Of the five chapters in this section, the first is concerned
with movement within the NP, the others with movement of
verbal constituents within the clause.
**Montse Batllori and Francesc Roca, 'The value of definite
determiners from Old Spanish to Modern Spanish'
The subject of this paper is differences in the behaviour of
_el_, _la_, _los_ and _las_ in Old and Modern Spanish. In
Old Spanish, these forms remained syntactically
demonstrative, like the members of the Latin paradigm of
_ille_ from which they were descended. Thus in Old Spanish,
_el_ etc do not cooccur in a NP with a demonstrative, do not
occur in a generic NP, and are attested pronominally. None
of these statements is true of the modern language.
Batllori and Roca characterise this change as the 'loss of
derivational steps involving movement'. They posit a DemP
as complement of D. In Old Spanish _el_ etc are analysed as
Dem having moved to D. In Modern Spanish they are analysed
as generated at D.
An interesting feature of this chapter is that the Old
Spanish texts used by the authors manifest two grammars with
regard to the behaviour of _el_ etc: the one briefly
described for Old Spanish and the one described for Modern
Spanish. The authors describe this as being akin to
code-switching, i.e. switching between two competing
grammars (see the Introduction).
**Lars-Olof Delsing, 'From OV to VO in Swedish'
Delsing examines Swedish texts from about 1270 to around
1580. He finds that OV order remained prevalent in the very
earliest of these but was followed quite early in the
fourteenth century by a split into what he calls Types I and
II clauses. Type I clauses were almost only VO, Type II
were OV or VO. Type II includes a bare noun with a 'light'
verb (e.g. loff vita 'give praise') and personal,
demonstrative, possessive and indefinite pronouns. The
proportion of Type II OV clauses drops during the rest of
the fourteenth century, increases again during the
fifteenth, then drops off during the sixteenth, landing at
9% in Delsing's last text (Brahe, ca 1580). Since the
mid-seventeenth century, Swedish has been a VO language.
Delsing argues that Type I objects have a filled D-position,
whereas Type II objects don't. The bare noun with a light
verb is clearly an NP rather than a DP, and Delsing cites
arguments from earlier papers to the effect that the
pronouns are generated in functional projections below
D-position. He suggests that Type II OV/VO variation is due
to a choice between two equally costly derivations. In the
VO case the head noun or pronoun moves to D, thus filling
the D-position. In the OV case, the whole phrase is moved
to Spec,AgrOP. The temporary increase in OV during the
fifteenth century is the result of heavy Low German
influence on Swedish and is a performance phenomenon, not a
change in grammar. Delsing speculates that the final shift
to pure VO is correlated with the loss of V-to-I movement.
This chapter is comparable with the previous one, in that it
seeks to explain two simultaneous grammatical behaviours,
but Delsing opts for an 'equal-cost' explanation rather than
for the 'competing grammars' hypothesis (see the
Introduction for discussion). His suggestion of Low German
influence implies, however, that at that stage there must
also have been competing grammars.
The lack of trees or labelled bracketings make this chapter
rather hard to follow: this is exacerbated by the fact that
the writer makes more specific theoretical presuppositions
than most of the book's other contributors.
**Chang-hye Han, 'The evolution of _Do_-support in English
Han asks why _do_-support developed in a major way in
negative imperatives only after 1600, whereas it has
developed earlier in its other contexts. Her answer depends
on two arguments. One is that negation occurred (and
occurs) at two different points in the phrase structure
tree, one above MoodP, the other below it (e.g. 'He didn't
not eat his greens'). The other is that the phrase
structure should recognise both a MoodP and an AspP. This
allows Han to divide V-to-I movement into two: V-to-Asp and
Mood-to-Tense. Loss of (the higher) Mood-to-Tense movement
occurred first, affecting interrogatives and negative
declaratives, but not negative imperatives, which do not
project a TenseP. Hence _do_-support developed in
interrogatives and negative declaratives but not in negative
imperatives. Then loss of V-to-Asp movement occurred early
in the seventeenth century. This did affect negative
imperatives, resulting in the development of _do_-support in
This paper is not only clearly written, but it presents an
interesting challenge to people working in other frameworks.
Han shows that a particular change (the development of
_do_-support) occurred in different structural contexts at
different times, and offers an answer to the question 'why?'
which is based on differences between these contexts. Her
answer is strongly framework-dependent, as it appeals to
Minimalist-style phrase structure.
Like Batllori and Roca, Han appeals to Kroch's model of
change as representing competition between grammars.
Indeed, she suggests that at the end of the sixteenth
century, three grammars were in competition: a grammar with
both Mood-to-Tense and V-to-Asp movement, one without
Mood-to-Tense but with V-to-Asp movement, and a third with
**Thorbj�rg Hr�arsd�ttir, 'Interacting movements in the
history of Icelandic'
The paper investigates the diachronic shift from verb-final
to verb-medial order in Icelandic. Although the author
talks about OV and VO order, she makes it clear that the
shift entailed not only objects, but other complements of
the verb--prepositional phrases, adverbs and adjectives,
verbal particles (i.e. the 'separable particles' of German
traditional grammar) and, where V is an auxiliary,
non-finite main verbs. She needs to explain why the
pre-verbal frequency of all these complement types declined
at the same rate over time. Her explanation of this change
is perforce quite different from Delsing's for Norwegian
(see above), since she tackles the shift in relation to all
complement types as a unitary phenomenon.
Hr�arsd�ttir accepts the Minimalist assumption that
underlying order is universally VO. Where there is a finite
auxiliary and a non-finite main verb, this results in the
order Aux Main Obj. She proposes three main transformations
that operate on this order. First, in Older Icelandic, the
embedded VP was optionally extracted from the matrix VP and
fronted to Spec,PredP, giving the order Main Obj Aux.
Secondly, at all stages of Icelandic, the direct object
moved leftward to Spec,AgrOP, giving OV order. In Older
Icelandic this results in Obj Main Aux, in Modern Icelandic
in Obj Aux Main. And finally, the remnant VP containing the
(finite) auxiliary was preposed, giving the order Aux Obj
Main (= OV) in Older Icelandic and Aux Main Obj (= VO) in
Modern Icelandic. The difference between Older and Modern
Icelandic, then, is that the first transformation, PredP
fronting, has ceased to apply in Modern Icelandic. The
change in constituent order over time receives a unified
explanation in the gradual loss of PredP fronting.
**David Willis, 'Verb movement in Slavonic conditionals'
The topic of this paper is the grammaticalisation of the
Russian conditional marker _by_, which was originally a
conditional form of the auxiliary 'be'.
Willis examines conditional usage in Old Church Slavonic and
Old Russian texts, where the conditional was formed from the
conditional of auxiliary 'be' and the active (_-l_
participle). Using the negator and clitic pronouns, he
establishes the movements which the auxiliary and the
participle undergo, and shows that the auxiliary moved from
T to to be right-adjoined to C. However, the environment for
this movement was wider in Old Russian than in Old Church
Slavonic. Crucially, the second/third person singular form
of the auxiliary was uninflected _by_ in Old Russian and
always underwent movement to C. As a result, Willis
hypothesises, _by_ came to be base-generated by learners as
a conditional marker at C (rather than as an auxiliary at T
undergoing movement to C) by the early fourteenth century.
This left a sentence with no finite verb, a configuration
which would have been rejected during acquisition, but for
the fact that in Old Russian the third person singular
perfect auxiliary was normally null, i.e. a sentence-type
without a finite verb already occurred. Willis takes us
through the steps which led, in the fifteenth century, to
the complete loss of the conditional auxiliary and the
generalisation of _by_ as conditional marker.
This story is theoretically interesting, as Willis points
out. Not only is it an example of reanalysis and
grammaticalisation, but, more specifically, it is a case
where movement is eliminated by reanalysing the derived
position as basic (unlike other well studied cases of
movement elimination, where movement has simply stopped
This chapter is one of the best written in the volume: the
account is lucid in that each step is carefully spelled out
and supported by clear tree diagrams.
Some comments on individual contributions are included in
the description above.
The title of this book suggests a wide sweep of subject
matter, but this expectation is disappointed in two
respects. First, almost all the papers are in one or other
version of a single framework, P&P/Minimalism (the
exceptions are Vincent [OT/LFG] and Briscoe, whose focus is
on community-based patterns of change). Second, the range
of languages from which data are drawn is very narrow. Only
Whitman's contribution takes us outside Europe, and of the
remaining twelve chapters seven are on Germanic, three on
Romance, one on Slavonic and one (Briscoe's) has no language
data. The second limitation follows in some measure from
the first, as it is a necessary condition of the P&P
approach that diachronic analysis is a comparison of a set
of synchronic states (as the editors [p16] and van Kemenade
[p53] both point out), and synchronic states from the past
must of necessity be based on the analysis of text corpora.
Hopefully, diachronic text corpora covering a wider range of
languages will gradually become available, although for most
languages and many language families this will never be
possible. Pre-modern texts are available, for example, for
very few languages of the Austronesian language family.
Diachronic syntax here entails syntactic reconstruction,
which falls outside the boundaries of diachronic syntax as
currently practised in the P&P framework.
Setting aside the expectations of the previous paragraph,
there is much of interest in this book. The strongly
text-oriented approach of most of the contributors
guarantees an empirical basis and challenges practitioners
of other frameworks to offer their own accounts of the
changes described in the book. The editors' contribution
provides a good introduction to issues facing diachronic
syntacticians in the P&P/Minimalist framework, and, broadly
speaking, these are not different from those facing
historical linguists working in other frameworks. They
include the interpretation and evaluation of written
sources, the mechanisms of change from both individual and
community perspectives (especially the interpretation of
variation), and the place of grammaticalisation within a
wider theory of morphosyntactic change.
Defining the domain of grammaticalisation within the
Minimalist Program is a topic that surfaces at various
points in the book. The editors point to Roberts and
Roussou (1999) as a significant contribution to its study.
Willis provides a well documented example of
grammaticalisation as defined by Roberts and Roussou: a
lexical head which moves is reanalysed as an in situ
functional head. Whitman adds another dimension of
grammaticalisation, discussing cases where there is no major
syntactic restructuring but a change in the categorial
feature of the head. Van Kemenade deals with the reanalysis
of a functional specifier as a functional head, but her main
concern is to show that grammaticalisation is
morphosyntactically (not semantically) driven (Willis'
chapter supports her in this). Vincent suggests that a
derivational framework is not the best way to tackle the
analysis of grammaticalisation.
There are some significant differences between the
enterprise pursued by the authors in this book and that
pursued by, e.g., Harris and Campbell (1995). The latter
are pursuing diachronic morphosyntax, i.e. they study what
happens to particular forms over time in a given structural
context. The writers in this book pursue 'the new
comparative syntax', without necessarily attending to form.
An example of this occurs in Martins' contribution (p211),
where we find French _Marie n'a achet� aucun livre_ 'Mary
didn't buy a book'. Here _aucun_ is the negative
indefinite, but it is cognate with the Spanish positive
indefinite _alg�n_ (and its cognates in other languages).
That is, the Old Romance positive indefinite has become the
French negative indefinite (the same is true of _personne_).
This morphosyntactic process is obviously relevant to the
history of these items in Romance, but it receives no
attention here because it is a matter of morphological
continuity, not of syntax.
Related to this is the labelling of the Old Spanish
construction with _todo_ as a 'reflex' of Latin _omnia res_
by Batllori and Roca (p250). For me, at least, a reflex is
a formal descendant. Here it is the semantically
There are occasional reminders that dependence on a single
paradigm can limit one's insights. For example, Delsing
finds "the division of objects into these two types of noun
phrases... surprising" (p262). Since one of his two types
consists almost entirely of pronouns, I was surprised by his
The book is beautifully presented and has been well
copyedited and proofread, so much so that it would be
churlish to enumerate the very few errors I found.
Harris, Alice C. and Lyle Campbell, 1995. Historical syntax
in cross-linguistic perspective. Cambridge: University
Hopper, Paul J. and Elizabeth Closs Traugott, 1993.
Grammaticalization. Cambridge: Cambridge University Press.
Keller, Rudi, 1990. Sprachwandel. T�bingen: Francke.
Kroch, Anthony, 1989a. Reflexes of grammar in patterns of
language change. Language Variation and Change, 1: 199-244.
Kroch, Anthony, 1989b. Function and grammar in the history
of English: Periphrastic do. In Ralph W. Fasold and Deborah
Schiffrin, ed., Language change and variation, 133-172.
Amsterdam: John Benjamins.
Kroch, Anthony, 1994. Morphosyntactic variation. In Papers
from the 30th meeting of the Chicago Linguistics Society, 2,
180-201. Chicago: Chicago Linguistics Society.
Le Page, Robert B. and Andr�e Tabouret-Keller, 1985. Acts
of identity: creole-based approaches to language and
ethnicity. Cambridge: Cambridge University Press.
Lightfoot, David, 1979. Principles of diachronic syntax.
Cambridge: Cambridge University Press.
Milroy, James and Lesley Milroy, 1985. Linguistic change,
social network and speaker innovation. Journal of
Linguistics, 21: 339-384.
Milroy, Lesley, 1987. Language and social networks (2nd
ed.). Oxford: Blackwell.
Roberts, Ian and Anna Roussou, 1999. A formal approach to
"grammaticalization". Linguistics, 37: 1011-1041.
Ross, Malcolm D., 1997. Social networks and kinds of
speech-community event. In Roger M. Blench and Matthew
Spriggs, ed., Archaeology and language 1: Theoretical and
methodological orientations, 209-261. London: Routledge.
Santorini, Beatrice, 1992. Variation and change in Yiddish
subordinmate clause word order. Natural Language and
Linguistic Theory, 10: 596-640.
Taylor, Ann, 1994. The change from SOV to SVO in Ancient
Greek. Language Variation and Change, 6: 1-37.
Wurff, Wim van der, 1997. Deriving object-verb order in
late Middle English. Journal of Linguistics, 33: 485-509.
Malcolm Ross is a Senior Fellow in the Research School of
Pacific and Asian Studies at the Australian National
University. He works on Austronesian and Papuan languages,
focussing particularly on New Guinea and northwest Melanesia
and on the reconstruction of the linguistic and culture
history of this region. He is particularly interested in
diachronic morphosyntax, as well as in the effects of