LINGUIST List 26.5482

Wed Dec 09 2015

Review: Historical Ling; Ling Theories; Semantics; Syntax; Typology: Gianollo, Jäger, Penka (2014)

Editor for this issue: Sara Couture <>

Date: 06-Aug-2015
From: Gerhard Schaden <>
Subject: Language Change at the Syntax-Semantics Interface
E-mail this message to a friend

Discuss this message

Book announced at

EDITOR: Chiara Gianollo
EDITOR: Agnes Jäger
EDITOR: Doris Penka
TITLE: Language Change at the Syntax-Semantics Interface
SERIES TITLE: Trends in Linguistics. Studies and Monographs [TiLSM] 278
PUBLISHER: De Gruyter Mouton
YEAR: 2014

REVIEWER: Gerhard Schaden, Université Lille - Nord de France

Reviews Editor: Sara Couture


This book grew out of a workshop at the meeting of the German Society for Linguistics in 2012. It consists of 11 chapters (of which the first is an overview and a presentation), a preface and an index, in vi + 359 pages. As its title indicates, it constitutes an exploration of grammatical change at the syntax-semantics interface. Most contributions use an explicitly generative and formalizing perspective (be that minimalist or cartographic syntax, or formal semantics).


The first chapter, written by the editors Chiara Gianollo, Agnes Jäger and Doris Penka, provides an overview of current research on grammatical change from a formal, interface-based perspective, and offers a presentation of the chapters contained in the book.

The second chapter, ''Semantic and Formal Features: Feature Economy in Language Change'' by Elly van Gelderen, investigates regular and directed language change from the vantage point of the feature theory of the Minimalist Program (MP). Drawing on examples from a wide variety of languages, she argues that changes in the grammatical status of linguistic items can be seen as feature changes, where what she calls ''semantic'' features are reanalyzed first as interpretable phi-features (that is, grammatical features that contribute to meaning), and finally, as uninterpretable phi-features (that is, purely formal features). The focus of this chapter is not so much an in-depth investigation of any specific kind(s) of linguistic change, but rather to take advantage of internal, regular and cyclic changes in order to investigate features and the language faculty from the perspective of MP.

In the third chapter, ''Linking Syntax and Semantics of adnominal possession in the History of German'', Simon Kasper studies the different possibilities of expressing possession in complex noun phrases in German - from different kinds of localizing/directional strategies (notably ablative ''von'', Engl. ''of''), to different kinds of genitive complements, and finally the notorious adnominal possessive dative (''dem Harald sein Hund'', lit.: the.DAT Harald.DAT his dog, that is, ''Harald’s dog''). The study starts with data from Old High German, and also takes into account contemporary non-standard German dialects, where genitive case has often been entirely lost. Kasper shows that there are three types of change: the replacement of the genitive case by other strategies; the displacement of genitives from a prenominal to a postnominal position; and the expansion of alternative strategies of expressing possession, which initially were not used for core-possession. While identifying many correlations between syntax and semantics, Kasper remains non-committal to the existence of a single cause for the observed changes.

The fourth chapter, ''MOST historically'' by Remus Gergel, traces the history of the quantifier MOST in English from a basic meaning ''biggest'' or ''largest'' to its current signification. His diachronic study is informed by recent formal semantic treatments of MOST (namely Hackl, 2009), which argue that MOST should be treated as a superlative expression, and not as a generalized quantifier. Gergel shows that in the oldest stages of Old English, MOST was syntactically an adjective. Subsequently, MOST became akin to a quantificational head by incorporating something like ''part of'' into its meaning, such that from an original structure MOST [part of [N]], one moved to MOST [N]. However, Gergel dismisses - based on Hackl - the idea that MOST has simply become a head (as would be implicated by van Gelderen's Head Preference, see van Gelderen 2004), since this would not be appropriate for the contemporary meaning of MOST. He also argues against a development by avoidance of movement (movement is assumed to be more costly than simple merge). Instead, he suggests that the presence of co-occurring nouns like PART-nouns (rising in frequency from Old English to the earliest period of Modern English, where it peaks and subsequently becomes rarer) acted as a catalyst for the development of the current meaning. In his conclusion, Gergel argues that the change undergone by MOST does not illustrate a straightforward case of grammaticalization, but rather what is at stake corresponds to what he calls ''ontological semantic functionalization'', that is, a case of incorporation of (at least partly) lexical material.

In ''The 'indefinite article' from cardinal to operator to expletive'', Paola Crisma studies the historical development of A(N) in English, comparing some of its diachronically attested stages with the status of indefinite articles in several contemporary languages. She argues that in Old English, AN was simultaneously a cardinal and an existential operator, but not yet a ''true'' indefinite article, which she takes to be an expletive. Her study - based on a syntactically annotated corpus of Old English - investigates nouns preceded by AN, and also ''bare'' nouns, which might serve as an alternative to AN. Crisma shows that in Old English, AN never appears with generics, and rarely with nouns under the scope of other operators. However, with what she calls ''existentials'' (nominals not being in a scope relation with other operators), the AN-variant is globally more frequent than the bare version, and especially so in a subpart of the corpus (e.g., Ælfric's Lives of Saints). Crisma proposes that the grammar of Bede corresponds to modern Icelandic, where AN does not occupy a D°, but rather a lower quantity projection. However, in Ælfric, she argues that AN is an existential operator used to mark specificity, similar to modern Hebrew XIT, and located in the D-field. Finally, after this stage, Crisma considers how AN grammaticalized in subsequent periods, with clear instances of phonological reduction and semantic bleaching, making it a standard case of grammaticalization (though for a period after Old English).

The sixth chapter by Nikolaos Lavidas, ''The Greek Septuagint and language change at the syntax-semantics interface: from null to 'pleonastic' pronouns'', examines the loss of the ellipsis of referential object-pronouns in Greek between Classical and post-Koine Greek. The author's aim is to test the hypothesis that this phenomenon is due to a transfer from Biblical Hebrew, where pronominal suffixes had to be used on the verb, and ellipsis was not available. As Lavidas notes, the specific Hebrew interference in the translated Septuagint became a general characteristic of later
Greek. Lavidas shows that there is a marked difference between the frequency of third person pronouns (encoding no interpretable features), and the frequency of first and second person pronouns in the accusative; whereas the former show a clear increase between the Classical and post-classical period, the latter remain stable. The author makes the connection with transfer and attrition in language learning, citing the ''Interpretability Hypothesis'': features that have semantically interpretable content are easier to access than purely formal features for language learners in a context of L2 acquisition or L1 attrition. This hypothesis is invoked because the introduction of these pronouns into Greek is not what one would expect under van Gelderen's Feature Economy Principle (where the direction goes from interpretable features to uninterpretable features); the historically attested change in Greek seems to go temporarily into the opposite direction. Lavidas argues that - if one integrates the perspective of language contact - the ''normal'' directionality assumed by van Gelderen can be upheld.

In the seventh chapter, ''The agreement of collective nouns in the history of Ancient Greek and German'',Magnus Breder Birkenes and Florian Sommer investigate the role of Corbett's Agreement Hierarchy in diachronic change. Their (slightly modified) agreement hierarchy is the following: attributive < copredicative < predicative < relative pronoun < personal pronoun. Collective nouns show an opposition between their ''semantic number'' - plural - which contrasts with their ''formal'' number - singular. Breder Birkenes and Sommer show that in Ancient Greek, agreement mismatches in collective nouns are more frequent with pronouns, whereas they are completely absent in attributive constructions, and show by and large intermediate results for intermediate positions on the agreement hierarchy. They also show that the average linear distance (that is, the number of words intervening between target and controller) is lower with the (matching) singular than with the (mismatched) plural. The same general picture holds for German, even though many structures are not directly comparable. However, they point out that cases of mismatches strongly decrease from the 18th century on - a factor that they attribute to the emerging standardization of the German language. Breder Birkenes and Sommer claim that direct instances of predicate mismatch (that is, of type ''the committee have decided'') are extremely rare in both German and Greek, and arguably ungrammatical in German (instances sometimes being direct calques from Hebrew and/or Septuaginta Greek). They also discuss whether there is a systematic syntactic difference between Ancient Greek and German in their agreement morphology. The main difference they identify is that Ancient Greek is a standard pro-drop language, whereas German - while having occasional cases of topic-drop - is not. Since pro-drop increases the linear distance between target and (overt) controller, this might be a factor.

In the eighth chapter, ''Vedic local particles at the syntax-semantics interface'', Antje Casaretto and Carolin Schneider study morphemes that are used as prepositions and preverbs (e.g., German AUF). In Ancient Indo European languages like Vedic Sanskrit, these morphemes were syntactically very independent, and basically adverbials. Casaretto & Schneider examine how far these adverbs advance in their change to adposititions and preverbs in Vedic. Since syntactical criteria cannot be relied on in a language with very few restrictions on word order, they investigate other, more semantic and morphological criteria. They show that - though Vedic does not have full-fledged adpositions, and the essential locative meaning relies on the case-system - there are instances where cases lose their meaning, and the local adverbial/pre-adposition comes close to being obligatory. With respect to the adverbials becoming preverbs, the linguistic system of Vedic has advanced further. There are clear cases of antonymization of the verbal base by a preverb (e.g., AVA, down + SĀ, bind => RELEASE, not TIE DOWN), and also cases of change in argument structure and telicity. Yet, complete lexicalization of the compound is argued to be still relatively rare in Vedic. Casaretto & Schneider also discuss whether - in cases where a local particle is reanalyzed towards both adposition and preverbal status - there is an anteriority of adpositional status, which has been argued for Latin and Ancient Greek. However, they find no evidence for this process in Vedic. Their general conclusion is that one can link the rather incipient development of adpositions to the fact that the Vedic case system (with locative, ablative and accusative) provides most of the local meanings speakers would need, whereas in the verbal domain, verbal bases need to cover a more complex meaning space, such that local particles are recruited in order to saturate this space.

The ninth chapter, ''Aspect Shifts in Indo-Aryan and trajectories of semantic change'' by Cleo Condoravdi and Ashwini Deo, remains in the domain of Old Indic, but also includes a study of Middle-Indic varieties. They propose a case-study of the development of a form expressing a resultative-stative in Ṛgvedic, a full-fledged perfect in Late Vedic, and a perfective tense in Middle Indo-Aryan (specifically, Mahārāṣṭri). The authors provide full formal semantic analyses for each stage, and argue that the underlying diachronic change is an instance of meaning generalization (where the meaning of the resultative entails the meaning of the perfect, and the perfect entails the meaning of the perfective). Condoravdi and Deo argue against the idea of grammaticalization involving the recruitment of pragmatic inferences into the semantics of the form. They hypothesize that the breakdown of the past-tense oppositions of the Vedic tense-system is caused by the fact of native speakers of Prakrit mapping the Prakrit-categories onto (Classical) Sanskrit.

Anne Breitbarth's chapter, ''The development of conditional SHOULD in English'', analyzes grammatical change as upward movement in the traditions of Roberts and Roussou (2003), set in a cartographic framework. The article aims to explain the tense-mismatch in SHOULD, which - while being morphologically past - is used most often with non-past time reference (instead of SHALL). Breitbarth shows that from the period of Middle English onward, SHOULD becomes rarer and rarer in counterfactual conditionals, and has entirely disappeared from this context in Modern British English. Realis conditionals were rarer in older varieties than in contemporary British English, whereas hypothetical uses have increased. She argues that SHOULD may be on the way of becoming a conditional marker (and occupying Irrealis-MoodP), which may eventually enter into competition with IF (citing an example from Trousdale 2012: ''Should Obama get_S_ the nomination...''), and thereby lose its verbal status. Breitbarth argues that in this particular case, semantic changes are ipso facto syntactic changes, and can be analyzed as upward reanalysis in the different mood-categories as provided in Cinque (1999). This is based partly on the idea that different types of modal effects can be linked to differing scopes of modal operators with respect to other elements in a proposition (event-variables, speech-act operators, etc.).

The eleventh and last chapter, ''The Greek Jespersen's cycle: Renewal, stability and structural microelevation'' by Katerina Chatzopoulou, investigates the evolution of the two Greek negations from Homer to Modern Greek. Greek features two different types of negators, the first being referred to as NEG1 (based on Indo European *NE) – the ''standard negation - and the second as NEG2 (based on Indo European *ME) - a type of negation restricted to non-veridical contexts. The peculiarity of Greek negation is that there is historically no attested stage of a discontinuous negation (cf. French NE verb PAS), although there clearly is renewal of the expression of negation, and syntactically, the originally post-verbal negator is now expressed as a preverbal negator. Furthermore, the Ancient Greek NEG1-morpheme has been reinforced by a word glossed as 'thing'. In order to account for the change observed in NEG1, Chatzopoulou proposes a reconceptualization of Jespersen's cycle, passing from an intensified predicate negation to a plain propositional negation. She traces the evolution of NEG2 from a general Negative Polarity Item to a speech act modifier. The diachronic development of NEG1 is analysed as a change from a phrase to a head (following van Gelderen's Spec-to-Head Principle), whereas the change of NEG2 is argued to correspond to an upward reanalysis in a cartographic syntactic framework.


The contributions in this book are united by their investigations of language change as acting at and on the syntax-semantics interface. This volume should be of interest to all researchers interested in grammaticalization, and especially to those interested in the interplay of formal linguistic theories (mostly in different flavors of generativism) and linguistic change. Everybody interested in grammaticalization (even those opposed to formalist approaches) will benefit from reading at least the introductory chapter. The quality of the editing is excellent.

Since I am far from knowing all the empirical issues and theoretical frameworks in the same manner, instead of focussing on specific contributions, I will rather present very general issues and observations that arise in this book, and that complement the presentation offered by the editors in their introductory chapter.

While many papers couch their analysis explicitly in a given framework (different flavors of generative syntax, Role and Reference Grammar, formal semantics), some of the articles do not commit to such a theoretical position (a stance that Haspelmath (2010) has explicitly advocated). I do not take the latter position to be defective – nor the former, by the way. However, I would like to point out an issue that has not been lost on at least some authors - especially in the realm of semantics, less so in connection with syntax: are formal linguistic methods appropriate for studying historical linguistics? The problem (brought up by Gergel, this volume) is that some crucial diagnostics simply cannot be applied, because they would require access to native speaker judgment. While formal models are extremely powerful (probably the most powerful at our disposal), their mere power may be useless if the resolution we get from accessible data is insufficient. However, this is probably not a general problem, and likely to be more problematic for semantics than for syntax. In any case, a move that allows to supplement historical data (or their absence) is the comparison with contemporary varieties showing the same kind of linguistic phenomenon (see, e.g., Crisma, this volume).

So, if there is a benefit of applying formal theories to diachronic phenomena, let us consider the other direction: is the study of diachrony relevant for formal linguistic theorizing? There are at least two issues here. First, can diachronic studies or diachronic data be used to decide between rival formal theories of a phenomenon, and second, do formal theories (which have primarily been intended to cover synchronic facts) need to accommodate diachrony - and does the fact whether they are able to do so tell us anything about their worth? The latter question has recently received a book-length treatment in Kinsella (2009), where the assumptions of the Minimalist Program have been argued to be implausible from an evolutionary point of view. It seems to me that it is perfectly legitimate to answer both questions in the negative - although I would not assume any of the authors in this volume to agree on this. Everything else being equal, it is of course preferable to have a theory that is synchronically and diachronically appropriate. Yet, if one assumes that a formal description of a language at a time t is as precise a description of the competence of a native speaker (at time t) as possible, without any claim to psychological validity, there is nothing inconsistent in denying diachrony any importance. But then, the price of abandoning claims to psychological validity will probably appear too costly to most linguists (but see Geurts & Rubio-Fernández (to appear) for stimulating discussion).

A second point I would like to discuss here is the question of the constitution of the corpus, and indications of frequency. In some of the papers in this volume, these questions are explicitly discussed, and some, though not all of the papers, report frequencies. I assume that the authors who did not report frequencies did so for a reason, and there are several plausible issues that can be mentioned. First of all, the available sources may not be representative for the spoken language as a whole. Such is arguably the case for Ṛgvedic Sanskrit, whose metric verse is probably in many respects not representative of the spoken language. Having a fully representative, and ''ecologically faithful'' corpus of Human linguistic behavior is out of the question for most periods of the past (and even for the present, it is difficult to build if you are not the NSA). Yet, as Pierre Briant is fond of saying, ''historians cannot choose their sources'', and skewed evidence is better than no evidence at all. A second issue is the general utility of frequencies. If one is concerned solely with the question of the competence of native speakers (which is the mainstay of generative linguistics), frequency - which can be argued to be a performance issue - does, in principle, not really matter. Even though it is easier to argue that a linguistic phenomenon is not a performance error when it appears hundreds of times, a single attestation would be enough to show that a given phenomenon is possible given a (native) speaker's competence (although confidence increases with frequency).

I would like to close with the question of the general causes of linguistic change. Generative approaches tend to focus on learning as the source of change, whereas functionalist approaches tend to rely on production biases and speaker usage as the source of change. Ultimately, it is very likely that both learning and production leave their marks on language change. The question is: can we attribute specific instances of change to either learning or production biases, and how do these factors interact?


Pierre Briant (2002): From Cyrus to Alexander: A History of the Persian Empire. Eisenbrauns.

Bart Geurts & Paula Rubio-Fernández (to appear): ''Pragmatics and Processing''. Ratio. Available at

Martin Hackl (2009): ''On the Grammar and Processing of Proportional Quantifiers: 'Most' versus 'More Than Half'''. Natural Language Semantics 17, 63-98.

Martin Haspelmath (2010): ''Framework-free grammatical theory''. In B. Heine & H. Narrog (eds.): The Oxford Handbook of Grammatical Analysis. Oxford: OUP, 341-365.

Anna R. Kinsella (2009): Language Evolution and Syntactic Theory. Cambridge: CUP.
Ian Roberts & Anna Roussou (2003): Syntactic Change. A Minimalist Approach to Grammaticalization. Cambridge: CUP.

Graeme Trousdale (2012): ''Grammaticalization, Constructions and the Grammaticalization of Constructions''. In: K. Davidse, T. Breban, L. Brems & T. Moertelmans (eds): Grammaticalization and Language Change. New Reflections. John Benjamins, pp. 167-198.

Elly van Gelderen (2004): Grammaticalization as Economy. John Benjamins.


Gerhard Schaden is a lecturer at the University of Lille, and a member of the CNRS UMR 8163 ''Savoirs, Textes, Langage''. His research interests include syntax, semantics, pragmatics, and language change.

Page Updated: 09-Dec-2015