Back to Book Reviews

Back to Cercles



Constructional Change in English

Developments in Allomorphy, Word Formation, and Syntax


Martin Hilpert


Studies in English Language Series

Cambridge: University Press, 2013

Hardcover. xiv + 233 pages. ISBN 978-1107013483. £65.00


Reviewed by Geneviève Girard-Gillet

Université Sorbonne Nouvelle – Paris III


Constructional change in English is a new book by Martin Hilpert, who is known to present quantitative approaches to linguistic phenomena, and who focuses this time on a corpus-based diachronic analysis of three case studies: in allomorphy the change by which the possessive determiners mine and thine respectively evolved into my and thy ; as regards word formation the –ment construction ; and in syntax the concessive parentheticals, though, although. The title Constructional Change suggests that, with these case studies, Hilpert aims at showing that a constructional approach to corpus-based data can bring new insights into the understanding of the evolution of English.

The book is divided into 6 chapters: an introduction, data and methodology, constructional change in allomorphy, construction change in word formation, constructional change in syntax, and a conclusion.

The introduction discusses well-known analyses of language evolution, then makes the point that constructional change is different from grammaticalisation. It does not subsume grammaticalisation, and it is not a “cover term” either, because some aspects of grammaticalisation do not instantiate constructional change. The question of frequency is particularly interesting in that regard. Hopper and Traugott [2003 : 126]  for instance, say that “textual frequency has long been recognized informally as a concomitant of grammaticalisation”. This means that, if a grammatical construction is less and less frequently used, it will not be considered as a case of grammaticalisation. But it is a constructional change. For Hilpert, construction change is more encompassing than grammaticalisation, which is commonly characterised by phonological reduction, reanalysis, univerbation, semantic bleaching  and pragmatic strengthening. He hypothesises instead that a grammatical change is the sum of many minor metamorphoses of specific units, and among these can be mentioned changes in form, function, frequency, and the distribution in the community of speakers. The theory is not interested in paradigmatic changes, and rejects the idea of “functional gaps”: if a construction falls out of use, it is not necessarily filled by another. Thus a constructional approach is thought to be more helpful to account for the emergence of constructions such as the way-construction (Israel 1996), let alone (Fillmore et al. 1988), or be like as a quotative (Buchstaller 2011), which do not belong to a paradigm. Hilpert gives the example of a change happening at the end of the Middle English period, and which concerns the replacement, for the third person singular, of the interdental suffix -(e) th, by the alveolar suffix - (e) s. The change only affected this grammatical form and did not motivate any change in adjectives or nouns: smooth and loath, broth and wealth kept their interdental suffix. But see chapter 3 for a generalisation of a paradigmatic subset.

In the conclusion to the introduction, Hilpert recognises that not all changes are constructional and he hopes that a discussion, about what grammaticalisation is, and what constructional change is, would be a welcome contribution to a better understanding of the evolution of English.

Chapter 2 presents the data and the methodology used. The sources of the data are varied, as is required by the usage-based approach and the diachronic perspective:


The Corpus Of Contemporary American English, COCA,  (Davies 2008)

The Corpus Of Historian American English, COHA (Davies 2010) which represents successive decades from the 1810s to the 2000s.

The Diachronic Corpus of Present-Day Spoken English, DCPSE. Half of the data are taken from the London-Lund Corpus, and the second half from the ICE-GB corpus.

The Old English Dictionary, OED

The Parsed Corpus of Early English Correspondence, PCEEC (Nurmi et al. 2006) which was compiled at the University of Helsinki. It contais about 5,000 personal letters written by more thatn 650 different authors between the 1410s and the 1680s.

The TIME magazine corpus, TIME (Davies 2007) with texts from 1923 to 2006.

The Penn Parsed Corpus of Middle  English, PPCME/PPCEME (Kroch & Taylor 2000) and the Penn Parsed Corpus of Early Modern English (Kroch et al. 2004), which contain prose texts produced between 1150 and 1710.

These corpora complement each other by offering different sources.

As constructions are multifaceted linguistic units, they may display simultaneous changes at different levels of description: frequency, structure, meaning and distribution, and thus it is necessary to understand how changes at different levels relate to one another over time. The methodology used tries to take these factors into consideration. To this end, Hilpert resorts to several techniques he already used in previous work, with Gries ( Gries & Hilpert 2008, 2010), (Hilpert & Gries (2009) or alone (Hilpert 2011), developed from page 32 to page 74, with examples: the procedure called variability-based neighbour clustering, which operates on the variability between neighbour, the binary logistic regression, the Hierarchical Configural Frequency Analysis, HCFA (Kraut & Lienert 1973, Von Eye 1990, 2002) and multidimensional scaling (LDS). These methods are used to analyse the three case studies mentioned above, and are carried out on the basis of even-sized sequential periods.

Chapter 3 deals with a constructional change in allomorphy, namely the evolution from mine, thine, to my, thy. Hilpert first refers to the results obtained in previous corpus-based analyses (Schendl 1997, Busse 2002, Nevalainen and Raumolin-Brunberg 2003). All these studies found that a crucial factor that determined the writer’s choice was the collocational context: frequent right-side collocates such as eye and own lead to the use of mine and thine. But other parameters come into play at some point over time: whether the writer is a man or a woman, the role of the social class, of the region concerned, of the age of the writer. Because of the existence of these many factors infuencing the evolution it is necessary to apply a multivariate methodology to the data. The variables studied were: the phonological environment, stress pattern, priming, formality, gender, and grammatical person. The results confirm that the choice between mine and my depends greatly on the phonological nature of the following segment, and whether it bears initial stress, but the relative frequency varies over time, as do other factors [82-94] ; but the geography factor was not assessed. Another question is  whether the first and the second person behave similarly. If differences were observed it would mean that the first person forms and the second person forms  represent different constructions, and the answer is thought to matter for the theoretical issue. The conclusion of the analyses is that we are dealing with one single constructional change, and not two changes. This shows that speakers have formed a generalisation over a subset of the paradigm of possessive determiners. This illustrates the notion that constructional generalisation can be formed at all levels of schematicity (Langacker 1991). Another important conclusion is that the writers’ social characteristics must be taken into account, along with language-internal factors.

Chapter 4 investigates a constructional change in word formation, and more particularly in the -ment construction. Hilpert presents a long summary of previous work, starting with Gadde (1910), who collected 1450 items from the O.E.D, then Anshen and Aronoff (1999), and Bauer (2001). This construction is worth analysing, as it was very productive till around the year 1650, then showed a sharp decrease, whose reason is not very clear. Another puzzle is that it ceased to be productive in spite of high type frequency. Hilpert resorts to the O.E.D. too, which is the most comprehensive resource for the construction under study, even it is not a regular diachronic corpus. He applies the VNC algorithm to determine the stages of the productivity of the construction over time: from 1250 to 2000, 5 periods are considered. But as the history of the construction is not only a change in productivity, Hilpert tries to determine the changes that the construction exhibits as regards to several properties: the etymological source (is the form borrowed from French, Latin, Spanish, or is it derived ?), the stem type (not all items combine a verbal stem with the suffix –ment, the branching structure (prefixation and right branching for [non-[attach-ment]], or left branching for [[be-devil]ment], transitivity (transitive and intransitive types), semantic types of the whole structure (shatterment is an act, settlement is a result, and so on). He applies the HCFA to the data for a multivariate approach that considers all the variables at the same time. The results are that the story of the V-ment construction is not one of increasing generality, nor is it one of increasing specificity, and that the changes observed are not unidirectional. Hilpert concludes that the evolution of the construction is not simply a case of grammaticalisation, or a case of lexicalisation, and that the notion of constructional change is more appropriate to account for the simultaneous and interrelated changes in its form, function and frequency.  

Chapter 5 concerns syntactic phenomena. Contrary to the previous phenomena presented in chapters 3 and 4, syntactic changes have recently received more attention from a constructional perspective  (de Vogelaer 2008, Fried 2008, Hilpert and Koops 2008, Traugott 2008, Tousdale 2008, Patten 2010, among many others). Hilpert focuses here on a family of concessive parentheticals exemplified in: power, although important, is not everything /The results showed a negative although non-significant correlation, among other instances, and proposes to investigate on how they emerged and how they developed. He uses the Time corpus to understand whether concessive parentheticals are the product of syntactic reduction or the product of analogy, and concludes from the data collected that the reduction hypothesis is more robust. Another question is whether these constructions, which display similarities and differences, can be grouped together under a common label or not. The data over the past 150 yeard are taken from the COHA and  divided into fifteen sequential decades, each of which is analysed through metric multidimensional scaling (MDS). Hilpert concludes in favour of the construction family hypothesis rather than of the macro-construction hypothesis, since the diachronic tendencies  do not show any gradual convergence  in their syntactic behaviour. He then wonders whether the evolution of the concessive parentheticals is better explained by resorting to the constructional theory, or not, since their emergence is a well-argued case of grammaticalisation, and their usage by speakers is highly subjectified. His answer is that it is, as the data collected and analysed show diversification in their syntactic and semantic profiles, instead of structural convergence, and suggest formations of local generalisations at different levels of abstraction.

In chapter 6, the conclusion, Hilpert sums up the main discussions previously developed and insists on the importance of the notion of constructional change to account for phenomena that only affect well-defined parts of the linguistic system. Another important point present all through the book is the question of the levels of abstraction at which a construction should be posited, and how speakers and hearers form the generalisations. The book also shows how new quantitative techniques of corpus analyses are essential to better understand the varied factors at work over time in the evolution of English.

Constructional Change in English by Martin Hilpert is a well-researched, well-written and easy-to-read book, thanks to the recurrent question of whether we are dealing with cases of grammaticalisation or not in all three case studies. Even if its aim is to present what a constructional approach can do, the analyses are meant as a complement to existing approaches to language change. It is brimming with data, reflections on previous work, and debated issues, which should help the readers to follow the argumentation, and to form their own conclusions. The quantitative corpus methods are well-described and contribute to the understanding of what the approach could bring to less-researched phenomena of evolution. The most interesting point could be, after all, that Martin Hilpert hopes, in his own term  that “if a meaningful discussion unfolds around constructional change, this would probably justify introducing the term”.


Cercles © 2014

All rights are reserved and no reproduction from this site for whatever purpose is permitted without the permission of the copyright owner.

Please contact us before using any material on this website.