Publishing Partner: Cambridge University Press CUP Extra Publisher Login
amazon logo
More Info

New from Oxford University Press!


Cognitive Literary Science

Edited by Michael Burke and Emily T. Troscianko

Cognitive Literary Science "Brings together researchers in cognitive-scientific fields and with literary backgrounds for a comprehensive look at cognition and literature."

New from Cambridge University Press!


Intonation and Prosodic Structure

By Caroline Féry

Intonation and Prosodic Structure "provides a state-of-the-art survey of intonation and prosodic structure."

Review of  Gradience, Gradualness and Grammaticalization

Reviewer: Lindsay Anne Morcom
Book Title: Gradience, Gradualness and Grammaticalization
Book Author: Elizabeth Closs Traugott Graeme Trousdale
Publisher: John Benjamins
Linguistic Field(s): Historical Linguistics
Issue Number: 22.617

Discuss this Review
Help on Posting
EDITORS: Traugott, Elizabeth Closs; Trousdale, Graeme
TITLE: Gradience, Gradualness and Grammaticalization
SERIES TITLE: Typological Studies in Language 90
PUBLISHER: John Benjamins Publishing Company
YEAR: 2010

Lindsay A. Morcom, Faculty of Linguistics, Philology, and Phonetics, University
of Oxford


This collection of papers, which resulted from the conference 'New Reflections
on Grammaticalization 4', held at the Katholike Universiteit Leuven in 2008, is
intended to bridge gaps in the linguistic discourse on the topics of gradience,
gradualness, and grammaticalization. It explores these topics from both formal
and functional viewpoints, and aims to encompass both synchronic and diachronic
data in doing so. Because the articles in the book cover such a wide range of
research concentrations and theoretical viewpoints, the editors have identified
three primary questions that all the papers seek to answer:

1. ''How are we to understand the intersection between synchronic gradience and
grammaticalization?'' (2)

With respect to this question, the authors consider variation to be both the
result of and reason for linguistic change. There are various ways to approach
this; in terms of generative theory, variation is taken as a function of
competing grammars and a matter of performance, and is therefore not central to
UG. In other theoretical frameworks, variation may be incorporated into the
linguistic model.

2. ''What insight does grammaticalization offer for assessing the validity of
Aarts' claims regarding synchronic gradience, specifically that there is a
significant difference between subsective and intersective gradience?'' (2)

In his work on the subject of gradience, Aarts (2004, 2007) explores linguistic
categories in terms of prototype theory. In this approach, categories may
display subsective, or intra-categorial gradience, with members being more or
less prototypical of the given category. They may also display intersective
gradience, with convergence occurring across categories. He examines this
primarily using data from English with a focus on parts-of-speech categories.
He concludes that these are different phenomena, and that although subsective
gradience is common, he does not believe that intersective gradience occurs, and
argues against unnecessary fuzziness in grammar. He states that ''the intuition
behind [his] proposals is that a particular formative may possess properties of
one or two categories to different degrees, resulting in gradience, but that the
categories in question can nevertheless be clearly delimited'' (242). However,
many authors, including several of those in this volume, such as Bisang, De
Smet, Denison, Hilpert, Rosenbach, and Schøsler, argue against this distinction
and against a lack of intersective gradience, especially with respect to
semantics and in cases where languages have little morphology.

3. ''What does the intersection between grammaticalization and synchronic
gradience tell us about the hypothesis of structural gradience, and about
whether work on grammaticalization needs reanalysis, analogy/extension, or some
other mechanism?'' (2)

Because of differing approaches across theoretical frameworks, varying
definitions of the terms in this question are necessary. The editors specify
that 'gradualness' refers to diachronic processes, while 'gradience' is
restricted to synchronic processes. The definition of 'grammaticalization' is
left to the discretion of the individual authors, although in all cases it is
treated as a process of extension. This question is based on the debate over
whether reanalysis or analogy is the dominant mechanism for linguistic change.

Although the papers included in this volume approach these questions from a
variety of theoretical standpoints, there is a general consensus regarding the
essential role of micro-changes in creating diachronic gradualness, as well as
synchronic gradience. Furthermore, the importance of semantics and pragmatics,
along with syntax and morphology, is emphasized throughout the volume; because
of this, corpora are a common source of data.

In the first article, ''Gradience, gradualness, and grammaticalization: How do
they intersect?'', Elizabeth Closs Traugott and Graeme Trousdale expand upon
their introduction and provide an in-depth discussion of the theoretical issues
that are the focus of this volume. They revisit the three questions that are
laid out in the introduction and give a summary of previous research on these
topics, comparing formalist and functionalist approaches. They ultimately
arrive at the view that a variationist/constructional approach best accounts for
both the semantic and morphosyntactic factors involved in grammaticalization.
They emphasize that although grammaticalization is gradual, the micro-steps
involved in it are cognitively abrupt for speakers, meaning that while
gradualness is diachronic, gradience is synchronic. They conclude that
grammaticalization is a constructional change that involves both form and
meaning and takes place through micro-changes across time, which may lead to
polysemy. It cannot be reduced to either analogy or reanalysis, which are seen
as related to grammaticalization but not as motivations for it.
In the second article, ''Grammaticalization, the clausal hierarchy and semantic
bleaching'', Ian Roberts approaches grammaticalization from a minimalist
perspective, taking the traditional view that all lexemes are members of one and
only one grammatical category. In his analysis, gradience and gradualness are
the result of a fine-grained system of categorization with distinctions that are
so small that they appear gradient. He commences his discussion by building on
Roberts and Roussou's (2003) idea that grammaticalization always involves
'upwards' movement to more abstract heads in a functional hierarchy. In this
view, reanalysis involves the suppression of movement, with an element that
formerly underwent movement being first-merged in a higher position. He adds to
this the concept of feature analysis, which allows for fine-grained category
distinctions. He then relates the work of Roberts and Roussou (2003) to
Cinque's (1999) modal hierarchy; he notes that the concept of the 'upward' path
of movement in reanalysis corresponds to Cinque's hierarchy. He examines
statistical data from a variety of languages that all show change from a lexical
category to a functional category or from a lower functional category to a
higher one in the hierarchy; in no case does he find a case of 'downward'
reanalysis. Therefore, these approaches are highly compatible in that Roberts
and Roussou (2003) establish the concept of 'upward' reanalysis, while Cinque
(1999) allows for the possibility of explaining this reanalysis in terms of
small, discrete changes in category membership that give the appearance of
diachronic gradualness and synchronic gradience. In this way, Roberts finds
that the minimalist approach to grammaticalization is compatible with the
functionalist approach discussed in the previous article by Trousdale and
Traugott, as these small changes correspond well to the micro-changes that they
describe. He then goes on to discuss semantic bleaching and its relationship to
grammaticalization. He arrives at the conclusion that the relationship of
semantic bleaching to grammaticalization is unclear since it is explanatorily
strong with respect to some instances of grammaticalization but fails to explain

The next article is ''Grammatical interference: Subject marker 'for' and the
phrasal verb particles 'out' and 'forth'', by Hendrik de Smet. De Smet examines
the ways in which grammatical elements that are undergoing grammaticalization
impact on the grammaticalization of other elements due to analogical thinking.
Using synchronic data drawn from a wide variety of English language corpora, he
examines the development of the subject marker 'for' in 'for…to-infinitive
constructions', as well as the phrasal verb particles 'out' and 'forth'. In the
case of the subject marker 'for', which appears in structures such as 'for me to
be here', he finds that this element developed from a focus marker 'for', as in
'for to help' and 'and now for something completely different', which is
different enough from the preposition 'for' based on both distribution and
semantics to consider the two to be homonyms. However, the preposition 'for'
influenced the development of the subject marker 'for' in that the structure of
prepositional phrases with 'for' is mirrored by 'for…to-infinitive' structures,
and because the two share certain distributional features. Similarly, phrasal
verbs with 'out' and 'forth' are often interchangeable, as in 'stretch
out/forth', although they are not interchangeable in all instances. Through an
examination of the historical development of these particles, the author
concludes that the development of 'out' copied that of 'forth'. Based on these
studies, de Smet concludes that grammaticalization is guided by connections
between various grammaticalizing elements, as well as other elements of grammar;
this occurs through analogical thinking and impacts structure, distribution, and
even persistence. From a constructionalist/connectionist stance, the author
states that connections between these elements form the basis for gradience;
grammatical elements intrude into each other's spaces on various grammatical and
semantic levels, leading to a lack of a one-to-one relationship between
different levels of symbolic organization and the sharing of some features but
not others. Gradience therefore extends beyond morphology and syntax to involve
semantics and lexically determined distribution.

Following this is the article ''Category change in English with and without
structural change'' by David Denison. In this article, Denison examines word
classes and category change in English. More specifically, he examines
instances of gradience between lexical categories, along with category change
involving structural change, such as the use of 'on behalf of', which acts as a
preposition as a unit while 'behalf' acts as a noun with a possessive marker.
He also looks at the introduction of new categories, such as English modals and
determiners, although he finds that category introduction is different from the
previously mentioned linguistic changes. Based on these studies, he draws
several conclusions. First, he finds that since category change is a process
that occurs in steps, Aarts' (2004, 2007) differentiation between subsective and
intersective gradience is unhelpful. He writes, ''the loss of prototypicality
within one category (subsective gradience) is not substantially different in
nature from the acquisition of an equal number of features of another category
(intersective gradience) and then onwards to full membership of the new category
(subsective gradience again)'' (113). He attributes gradience, as well as dual
inheritance, to analogy, which is commonplace in human cognition and therefore
to be expected in human language. He also states that that semantics coerces
syntax rather than the converse. Furthermore, he notes that in syntax-based
theories, certain rigid assumptions, such as the separation of semantics and
syntax and membership of lexemes in one and only one category, may be useful but
may not be the most informative way to examine issues of gradience, gradualness,
and grammaticalization. He supports the theory of Construction Grammar in
modelling grammaticalization, since it reduces underspecification, focuses on
whole constructions rather than individual parts, while not ignoring the parts
completely, and allows for dual or multiple inheritance.

Elly Van Gelderen's article 'Features in reanalysis and grammaticalization' is
largely a critique of the previous articles by Roberts, De Smet, and Denison.
She offers an alternative explanation of the data they discuss based on the
Feature Economy Principle, in which language variation is determined by feature
variation, and it is assumed that it is more economical to merge features in a
higher position than to do so in a lower position and then move. In this view,
change takes place in micro-shifts that occur in the course of child language
acquisition. She examines the data discussed by De Smet in terms of the
synchronic development of 'for', and states that the Feature Economy Principle
accounts for upward linguistic change, with feature loss and revaluation leading
children to reanalyze linguistic input. She points out that in some cases
different theories employ different definitions for the same terminology. This
can lead to radically different analyses; for example, the changes that Denison
discusses as being non-structural are structural in her account. She believes
that a discussion of gradualness and directionality is unnecessary, writing ''I
think we should stop worrying about whether change is gradual or not (it seems
to be agreed by most that it is) or directional or not'' (145).

In the article ''How synchronic gradience makes sense in the light of language
change (and vice versa)'', Annette Rosenbach takes a diachronic approach to
synchronic gradience, examining it in terms of a mismatch of syntax and
semantics. She commences her discussion by re-examining Aarts' (2004, 2007)
assumptions that there is a distinction between subsective and intersective
gradience and that category membership is determined by the application of
morphosyntactic criteria. She argues that intersective gradience is common if
semantic criteria are taken into account, and that a diachronic, corpus-based
examination of this phenomenon enables one to see how gradience develops and
drives language change. To demonstrate this, she examines the development of
determiner genitives like 'the woman's blue eyes' and noun modifiers like 'the
expensive theatre ticket', and the gradience that exists between them; although
the determiner genitive is generally referential and animate, while the noun
modifier tends to be inanimate and non-referential, expressions such as 'the
cheerful Obama supporters' show that this is not always the case. She concludes
that overlap between these constructions, which is relatively new in English,
occurs due to a mismatch of the mapping of semantic features to construction
type. Bridging constructions like s-less genitives and lexicalized expressions,
along with the process of expansion/extension of noun modifiers and genitives,
make this a possibility. Language change is a gradual process that occurs in
micro-steps and is based on analogy in which formal similarities result in the
sharing of semantic features, which in turn reinforces the similarities. She
points out that this cannot really be classified as reanalysis, since reanalysis
is generally defined in terms of formal change, while this change involves
semantic function; if semantics and morphosyntax are taken into account, the
line between reanalysis and analogy is blurred. Finally, she looks briefly at
typological data from a variety of languages, which indicates that a sharing of
semantic features between genitives and noun modifiers is not limited to English.

In the following article, ''What can synchronic gradience tell us about
reanalysis?: Verb-first conditionals in written German and Swedish'', Martin
Hilpert discusses various accounts of the evolution of verb-first conditional
structures in German and Swedish; in these languages, constructions such as 'had
I known this, I would have stayed at home' are far less restricted than they are
in English, with the Swedish construction being more grammaticalized than the
German. Using corpus data, he compares a dialogue-based account with an
analogical account by describing and evaluating the predictions that each makes.
The dialogual approach posits that verb-first conditionals developed from
didactic question and answer interaction. The analogical approach, on the other
hand, posits that they developed through analogy from complex clause patterns
that exist in these languages. He compares these approaches through an
examination of text frequency, the frequency of linking elements, the use of
first-person subjects, counterfactual verb-first conditionals, the collocational
overlap between conditionals and questions, and the displacement of the
subordinate clause (protasis). The first four of these are consistent with the
predictions of the dialogual approach, as is diachronic evidence, while the
latter two are not explained by it. He explores issues related to genre and
modality, as well as the analogical approach and an account involving two
juxtaposed declaratives as possible explanations for this. The author also
finds that these structures developed gradually, and that reanalyzed structures
retain aspects of the source structure for a long time. He also finds that
synchronic, cross-linguistic data is useful for the evaluation and comparison of
various theoretical approaches.

Lene Schøsler's article, ''A paradigmatic approach to language and language
change'' looks at synchronic gradience and language change, as well as reanalysis
and analogy, using construction paradigms and data from Danish and French. She
argues that the use of paradigms should be expanded to describe
syntactic/semantic phenomena. She discusses the development of the divalent
dative in French, the introduction of the verb 'brainstorm' into Danish, and the
use of verbs for electronic communication, such as 'skyper' in French. She
finds that paradigms, when viewed as 'packages' of content and expression, can
be extended to examine syntactic phenomena, with new verbs integrated into
existing grammatical structures. According to Schøsler, the extension of
paradigms from morphology to syntax is useful because it allows for consistent
analysis of various parts of grammar, as well as the interface between different
parts of grammar. She argues that grammaticalized constructions can be viewed
synchronically as a succession of paradigms that have undergone numerous
small-scale changes, and that this analysis facilitates the study of the
emergence and modifications that new constructions undergo.

Following this is the article ''Grammaticalization and the it-cleft
construction'', in which Amanda L. Patten examines development and extension in
English using a corpus-based constructional approach. First, she examines
non-NP focus It-clefts, such as 'It's in December that she's coming'. She then
goes on to discuss 'Informative-Presupposition It-Clefts', in which the relative
clause contains new information, such as 'It was Cicero who once said, ''Laws are
silent at times of war.'''. She finds that non-prototypical It-clefts, such as
non-NP focus It-clefts and Informative-Presupposition It-Clefts emerged via
coercion through the extension of the existing It-cleft structure; this
structure has undergone gradual, continual grammaticalization with increased
schematicity as a result of allowing a wider range of elements into the focal
position. Therefore, a diachronic examination of this structure using the
concept of grammaticalization aids in the understanding of its present
functional properties and distribution. As grammaticalization occurs,
prototypical associations between semantic meaning, syntactic category, and
pragmatic function are mismatched. The fact that NP-focus It-Clefts such as 'It
was the therapist that killed her' emerged first and are still the most common,
prototypical form of this construction, means that this construction exhibits
gradience, as do the categories of constructions that occur in the It-cleft.
Thus, extension from prototypical to less prototypical versions of a
construction over time leads to subsective constructional gradience, as well as
intersective gradience at the elemental level as members of different categories
come to share syntactic positions.

Walter Bisang's article, ''Grammaticalization in Chinese: A construction-based
account'' is the only article in the volume that focuses on a language that is
not Indo-European. Chinese is interesting in terms of grammaticalization
because Late Archaic Chinese is precategorial, meaning that lexical items are
able to fill a variety of syntactic slots; this is less the case in modern
Chinese. In his paper, Bisang first outlines his approach to construction
grammar with a focus on the concepts of scaffolding and coercion, in which
lexical items are slotted into syntactic structures, and syntactic structures
are able to coerce lexical items into particular functions. In the case of a
precategorial language, all lexemes can occur in all syntactic slots, and this
distribution specifies or highlights the function of a given lexeme in a given
construction; in Late Archaic Chinese, for example, even proper nouns are able
to go in the syntactic V slot. Such flexibility means that one sentence may
have numerous interpretations, and that the possibility of a lexical item being
reanalyzed and undergoing grammaticalization is increased because the number and
type of constructions in which it can appear is increased. This has resulted in
the development of structures such as the resultative, in which two verbs appear
together with the second indicating the result of the action conveyed by the
first; this did not occur in Late Archaic Chinese, but appears to have developed
due to the flexibility of its grammar. Based on this data, Bisang challenges
the assumption that words in all languages are members of one and only one
lexical category, and that word class membership is independent of meaning.
Furthermore, in terms of grammaticalization, Bisang shows that data from
languages like Chinese give an interesting insight because they show that
lexical items that are commonly used in one slot may be used in other slots as
well; when most or all lexemes can do this, it calls into question the division
between subsective and intersective gradience.

In the last article, ''Grammaticalization and models of language'' Nigel Vincent
and Kersti Börjars examine the role of theory in describing language change and
find that formal and functional approaches, rather than being mutually
exclusive, are both useful in the study of this phenomenon. They evaluate the
strengths and weaknesses of both approaches, and point out that the assumptions,
categories, and frameworks used by a given theory can change the outcome of a
study; for example, the categorical divisions that theoreticians in one
framework choose to employ may make grammaticalization appear gradual, while the
divisions in another theory make it appear abrupt. They compare the data and
conclusions of numerous other articles in this volume with each other and with
new data to demonstrate this. They then discuss the benefits of formal
theoretical approaches that take advantage of functional data, and they describe
the use of some of these approaches in detail. LFG and Stochastic OT have the
benefit of representing function and constituency independently, thus allowing
them to model both semantic and syntactic grammaticalization and to examine the
links between them. Dynamic Syntax models language based on the perspective of
the hearer/parser, which reverses the idea of grammaticalization as lexicon
becoming syntax; instead, syntax is seen as being incorporated into a lexical
entry. This allows for new ways of seeing grammaticalized structures.
Furthermore, the authors discuss the application of formal semantics to the
study of grammaticalization. They draw several conclusions based on their data
and the studies in this volume; first, they state that linguists must be willing
to step beyond construction grammar and minimalism as opposing views, and they
must examine other theoretical approaches. They also emphasize that formalism
and functionalism both have merit and are not mutually exclusive. Finally, they
find that form and function do not necessarily change together, and they argue
in favour of an approach that allows for their parallel study rather than
linking them intrinsically as in minimalism and construction grammar.


This volume is unique in that it examines the issues of gradience, gradualness,
and grammaticalization from a cross-theoretical standpoint. In producing such a
volume, the editors run the risk of it containing articles that are too attached
to a single theoretical viewpoint, and are therefore inaccessible to some
readers. However, in most cases this pitfall is avoided, and the majority of
the articles are written in an accessible way that outlines theoretical
assumptions and avoids excessive jargon. A volume that contains articles based
in different theories not only enables the reader to weigh the benefits and
limitations of various theoretical approaches, but also to compare the
conclusions of authors from other theoretical backgrounds with their own. It
also opens the reader up to new ways of examining data. As Vincent and Börjars
point out, ''Different theoretical approaches lead one to look for explanations
in particular places, so that awareness of a plurality of approaches means a
better appreciation of potential explanations, however these are ultimately
modelled'' (296).

However, the diversity of articles in this volume does present a weakness in
that the articles deal with a large range of data, make a variety of assumptions
that may not be laid out for the reader, and sometimes use terminology that is
defined differently across theories. Many of the authors acknowledge this, and
attempt to define their terminology adequately. However, due to space
constraints, and the fact that for many linguists who work solely in a single
theory, it is difficult to remember that not everyone makes the same
assumptions, in some articles terminology is not adequately defined and
assumptions are made not only about theoretical basics but also about what
constitutes common knowledge. The editors attempt to overcome this through the
use of the unifying questions summarized at the beginning of this review; in
many cases, this draws the articles together, although some authors do not refer
to the questions at all, and others refer to them only in passing.

A far more unifying feature of the book is that the authors have clearly read
each other's papers prior to publication, which has enabled them to comment on
other approaches and present alternatives. This is evident in many of the
articles, and in some cases is a real strength to the book, as it gives an
impression of a discourse between researchers from various backgrounds. Not
only does this clarify the issues presented in many of the articles, but it also
makes the volume much more enjoyable to read, since in many instances it has the
feeling of a lively debate. The arrangement of the articles in the volume
facilitates this. For example, Rosenbach compares the use of features in her
functional approach with Roberts' formalist approach and states her agreement
with de Smet on feature sharing and mismatch. Schøsler then refers to
Rosenbach's data on determiner genitives and builds on her approach to semantic
factors in gradience through the addition of new data. This culminates in the
article by Vincent and Börjars, who compare formalism and functionalism and
point out that although all the other articles in the book are either formal or
functional, other theoretical alternatives to these approaches exist. In some
cases, however, this backfires; for example, the majority of Van Gelderen's
paper critiques the articles of De Smet, Roberts, and Denison, while offering
very little in the way of empirical evidence and novel theoretical discussion.
This particular article reads more like a review of these authors than an
article in its own right, and does little to add to the overall academic
contribution of the volume.

This volume would be significantly strengthened if it included data from a
greater variety of languages. Roberts contains statistical data from several
languages, but almost no linguistic data, van Gelderen contains a chart of
macroparameters in various languages, and Rosenbach contains single
constructions from Georgian and Hebrew. In terms of in-depth analysis, all of
the authors but Bisang focus on Indo-European languages, with the vast majority
of data from Romance or Germanic languages. While data from these languages is
more accessible to English-speaking readers, any reader of this volume is likely
to be a fairly experienced linguist who would be able to cope with less familiar
linguistic structures. Bisang's excellent article demonstrates that data from
non-IE languages is extremely informative and indeed necessary for the
development of accurate typological theory and more broadly of linguistic theory
that adequately describes human language. Bisang's article discusses Late
Archaic Chinese as a pre-categorial language; the inclusion of data from such
languages is extremely important for a discussion of linguistic categories in
general. This article shows precategoriality in an isolating language; however,
the volume would benefit greatly from a discussion of this phenomenon in
languages like those of the Salish or Wakashan families, or one of the many
other language families that exhibit precategoriality in a system with complex
morphology. Numerous linguists have discussed gradience in non-IE languages;
certainly, therefore, sufficient synchronic and diachronic data exist to compose
an article with data from many of these languages. The inclusion of such
languages would make the conclusions drawn by the authors and editors of the
volume, and by the reader, better informed, more complete, and more reflective
of human language as a whole.


Aarts, Bas. 2004. Modelling Linguistic Gradience. Studies in Language 28, pp.

----. 2007. Syntactic Gradience: The Nature of Grammatical Indeterminacy.
(Oxford: OUP).

Cinque, G. 1999. Adverbs and Functional Heads: A Cross-Linguistic
Perspective. Oxford Studies in Comparative Syntax. (Oxford: OUP).

Roberts, I., and Roussou, A. 2003. Syntactic Change: A Minimalist Approach to
Grammaticalization. Cambridge Studies in Linguistics 100. (Cambridge: CUP).

Lindsay Morcom has recently completed a D.Phil in General Linguistics and Comparative Philology from the University of Oxford. Her research focuses on endangered and indigenous languages, especially those of Latin America and Canada. She has performed fieldwork on the Pokomchi' language of Guatemala, and has composed a grammar of this language in addition to publications and presentations on Pokomchi' text analysis and metaphorical systems. Her doctoral thesis is a comparative study of lexical categories in Salish and Wakashan languages and Michif; it examines data from these languages to explore gradience in parts of speech categories, and it compares grammatical categories with conceptual categories using prototype theory. She has also done research on coordination in Plains Cree. In addition to this, she has a passion for teaching and enjoys working with undergraduate students.