LINGUIST List 24.4492

Mon Nov 11 2013

Review: Semantics; Syntax: Alexiadou & Schäfer (2013)

Editor for this issue: Joseph Salmons <>

Date: 12-Aug-2013
From: Laura Arman <>
Subject: Non-Canonical Passives
E-mail this message to a friend

Discuss this message

Book announced at

EDITOR: Artemis AlexiadouEDITOR: Florian SchäferTITLE: Non-Canonical PassivesSERIES TITLE: Linguistik Aktuell/Linguistics Today 205PUBLISHER: John BenjaminsYEAR: 2013

REVIEWER: Laura Arman, University of Manchester

INTRODUCTIONIn its own words, the work under review supports the claim by Huang (1999)that there is no real definition of canonical and non-canonical passives. Eachchapter presents a language-internal study of constructions labelled at somepoint in recent linguistic history as 'non-canonical passives', providingevidence that no clear distinction exists of what properties are exclusivelypassive. This means that no unifying theory of passives or account of theirstructure exists that accommodates all possible passives and the constructionor 'voice phenomenon' has been characterized as a cluster of features intypological work for decades (Siewierska 1984, Keenan and Dryer 2007). Thisbook makes no correction to our notion of passive as it stands, with mostchapters elaborating on cases where formal theories either dismiss or fail todifferentiate constructions. The above is an issue summarized and exemplifiedconcisely in the introductory chapter by the editors, taking into accountprevious well-used 'definitions' of passive.

SUMMARYArtemis Alexiadou and Florian Schäfer’s introduction outlines severalproperties considered to define or diagnose passives. Using Baker, Johnson &Roberts (1989) as a basis, the chapter exemplifies the kind of diagnosticswidely used for passives which are employed by all chapters of this collectedwork, disambiguating passives from superficially similar clauses: licensing ofby-phrases, control of PRO, agentive adverbs and disjoint reference inadjectival passives. The introduction also attempts to synthesize majorarguments from the last two decades on the analysis of non-canonical passives,including Alexiadou's own work -- the issue of implicit arguments raised hereis one which remains a constant for later analyses. Evidence is provided fromEnglish, German, Dutch, French and Japanese for the blurriness of what isanalysed as canonical and non-canonical within passives, to convince us of theneed for more detailed analyses as well as to outline some of the questionsthat come with such data:

- How is a get-passive different from a be-passive and why is get used overany other verb?- Why are only certain auxiliaries used with passives?- Why are there different types of adjectival passives?- Which restrictions on verb classes can be observed?

These data are revisited in greater detail in later chapters.

Andrew McIntyre's ''Adjectival passives and adjectival participles inEnglish'' posits several original ideas to the two titular forms, some at oddswith previous well-known analyses of the adjectival passive, such as Kratzer(2000). McIntyre proposes an analysis of subtypes of adjectival passivesincluding a type previously unidentified, or at least undistinguished:situation in-progress participles. A sentence such as 'that car seems badlydriven, so keep away from it' can only be uttered whilst the car is stillmoving, for example. Although there may also be a possible resultative readingof these participles, they clearly differ from resultative participles, suchas 'the car is scratched', and eventive-verb-related pure statives (previouslyknown simply as 'stative participles'), such as 'the bars are bent because thecraftsman moulded them that way', in that they are time-reference dependant ontheir verb (cf. Embick 2004). This contribution also concludes that adjectivalparticiples can have an implicit agent, or rather initiator (which covers anyexternal argument), as attested also for Hebrew (Meltzer-Asscher 2011) andGreek (Anagnostopoulou 2003), with examples such as 'The Picts paintedthemselves blue and stayed painted for several days' clearly containing aninitiator of the verbal participle's event. On the basis of such data,McIntyre suggests that transitive-based adjectival participles are trulypassive, whilst unaccusative-based participles are not. This analysisdisambiguates the previously problematic adjectival passives from adjectivalparticiples, necessary for solidifying the boundaries of passives in English.McIntyre concisely lays out formalisms for the semantics of his proposedparticiple types and proposes a syntactic analysis where the theme of theadjectival participle is merged externally to the participle's projection.This is not uncontroversial, but is supported by previous tests on Russian(Borer 2005), Hebrew (Meltzer-Asscher 2011) and Italian (Cinque 1990), as wellas work in lexicalist theories.

The second language-specific chapter is also on English, focusing on thechoice of verb for its periphrastic passives in Anja Wanner's corpus-basedstudy 'The get-passive at the intersection of get and the passive'. As mightbe expected, this paper additionally deals with register and style, unlikemost papers in the volume. She begins by reviewing English passives and claimsmade about get-passives and get-arguments in general, raising questionsechoing those in the introduction, such as whether the get-passive is a true,if quirky and restricted, passive or whether get is the regular lexical verbor some grammaticalization of it. Whilst the paper largely discusses previousanalyses of implicit arguments in get-passives, it also provides original dataon get-passives from the FROWN (ICAME) corpus. Wanner endorses Orfitelli’s(2011) notions of a responsibility get-passive and a non-responsibilityget-passive, to account for different get-passive types found in the 57get-passives of the corpus data. Whilst offering no strong conclusion on thepassive status of get-passives as a whole, the overall contribution here isthat get-passives are not necessarily strictly differentiated from theirbe-passive counterparts. Some constructions involving no secondary agentappear more easily with the get-passive, due to the causative nature of some'gets', but be-passives and get-passives still largely overlap -- 'he waspromoted' or 'he got promoted' for example. Instead, it is suggested that thestructure of 'get' is flexible enough that it allows for these differingcontextual interpretations, or we are left with the notion of two differentstructures for one surface phenomenon.

Alexandra N. Lenz's chapter 'Three ''competing'' auxiliaries of anon-canonical passive' is another interesting, data-driven take onget-passives. This time, the social variation under investigation was regionaland dialectal, as well as taking register into consideration. An enormousamount of data from various corpora were used to determine the sociolinguisticfactors involved in selecting one of three possible auxiliary verbs for theGerman get-passive, namely 'kriegen', 'bekommen' and 'erhalten' -- all withsome combination of the meaning ‘get/receive’. The meticulous study of Germandialect and register finds that the selection of passive auxiliary does indeedvary with both register and region, providing information which bothcontradicts and supplements what is assumed by literature on standard Germanand on German passives.

A more general analysis of 'Variations in non-canonical passives' is given inC.-T. James Huang's paper, with data drawn from Mandarin Chinese, English andGerman, although unlike other chapters it is mostly a discussion of thepossible theoretical analyses of a Mandarin construction. Despite unhelpfullycharacterizing the non-canonical passives of this dataset as 'chameleonic',Huang identifies two different analyses applicable to these sets ofnon-canonical passives, such as the English get-passive and the Chinesebei-passive -- they can be analysed as being control or raising constructions.In Huang's view, non-canonical passives can differ from other passives inselecting a semi-lexical verb to superimpose on the main predicate, which maychange the argument structure of the predicate as a whole. More detailed ishis analysis of the lexical selection and syntax of the Mandaringive-passives, which supports the assertion that auxiliaries with differingproperties along the causative-unaccusative continuum have an effect on theargument structure, and ultimately, the analysis of a construction as eitherraising or control.

'How much 'bekommen' is there in the German 'bekommen' passive?' presents bothexperimental and corpus data. Markus Bader and Jana Häussler's account of the'bekommen' passive auxiliary shows that the construction is sometimesacceptable with verbs whose dative object is not a recipient -- sometimes, butnot always. What they conclude from their study is that the 'bekommen passive'is indeed not a recipient passive and so only part of the semantics of'bekommen' is retained in its grammaticalized use as a passive auxiliary. Thesemantics of the auxiliary is discussed, but a more fine-grained analysis isleft for future work.

The next two contributions are also on German, this time rejecting theanalysis of non-canonical passive for the construction studied. In'Haben-statives in German: A syntactic analysis', haben-statives are analysedas a class of their own as Martin Bunslinger finds them to have the sameunderlying structure, rejecting the need for Kratzer’s (1993) distinction of'lexical' and 'phrasal' adjectivization. Unlike their English counterparts,German haben-statives consist of a main verb and an adjective, according tothis analysis, and this is supported in the next chapter, 'Another passivethat isn't one: on the semantics of German haben-passives'. Helga Gese'sanalysis captures the semantic overlap of participles in 'haben-passives' andGerman adjectival passive participles (with 'sein'). Gese claims that a zeroadjectiviztion affix can account for this overlap in the semantics of theparticiples of the two constructions, which are both main verb + predicativeadjective, but differ from their non-adjectivized counterparts in lacking theevent-kind meaning that comes with the deverbal adjectives.

James E. Lavine's 'Passives and near-passives in Balto-Slavic: on the survivalof the accusative' attempts to account for the structural accusative inpassives as found in Lithuanian, Polish and Ukrainian. He proposes thatv-Cause and v-Voice be 'bundled' together in Polish so that an externalargument has filled the specifier of the Voice node, resulting in thereanalysis of the Polish so-called impersonal construction as active. InLithuanian Inferential Evidentials, a similar analysis holds. However, in thiscase, the historical passive morphology, /-ma/-ta/, cognate with Polish/-no/-to/, has not been reanalyzed as active and thus still heads v-Voice. Asthe v-head is fused, accusative is not assigned to the Lithuanian evidentials,whereas it is hypothesized that Ukrainian has a split vP and so can stillassign accusative to its direct object.

The optimistically titled 'How do things get done: on non-canonical passivesin Finnish' offers not only another non-Indo-European passive puzzle, butproposes a new one. This contribution proposes that Finnish has agreeingpassives (as well as the previously recognized non-agreeing kind) whichsomehow resemble English get-passives and differ from the Finnish copularconstruction to which it may be compared. Fredrik Heinat and Satu Manninenclaim that the two passive types 'pattern alike' and differ significantly fromthe copular construction; they provide a syntactic analysis to this effect, onthe grounds that the agreeing passives are a copular construction whichcontains a passive. They see the copular construction differing from thepassives in its lack of vP, whereas the two passives are seen to select theirparticiples using a different head.

Marie Labelle's chapter 'Anticausativizing a causative verb: the passive 'sefaire' construction in French' brings an interesting case of a constructionwhich is semantically passive but formally causative. The chapter lays out theproblem of analysing the construction as causative, as the non-reflexivecounterpart would be. In this case, the subject of a 'se faire' constructionis not a causer and in fact Labelle analyses the construction as having noCAUSE in the syntactic structure. Instead, 'se faire' requires a missingobject in its embedded clause, which brings the non-canonical passive claim tothe fore. This paper sees an account similar to that of Embick (2004)'sresultatives being applied to the 'se faire' constructions, only that Voiceoccurs above v in this instance, for what are assumed to be resultanticausatives (Labelle & Doron 2010). This contribution provides a detailedsyntactic as well as semantic analysis of the construction in question,drawing on plenty of natural data, and dedicates space to discuss insatisfactory detail the issues arising from the analysis. Labelle concludesthat these constructions are decausativized and resemble result anticausativesin that, in both, a change of state event is headed by 'faire' and, in both,the external argument is not merged. The passive semantics is a result of thechange of state event of the higher clause.

Fatemah Nemati's provides an overview of the different passivizationstrategies available in Persian, 'On the syntax-semantics of passives inPersian'. Another Indo-European language using the become-passive, Persian'sother uses of 'šodæn', become, are contrasted with their use as a passiveauxiliary (in the form of a light verb construction), 'noun/adjective +become'. It seems that 'šodæn' will be interpreted as a copula unless theelement preceding it has a [+passive] feature, according to Nemati's analysis.Rather than assuming a different lexical representation for the passive use of‘šodæn’, it is the interaction of the lexical semantics of the constructionthat will determine the structure of the light verb.

Masanori Deguchi's brief chapter on 'Two indirect passive constructions inJapanese' deals with so-called ‘affective’ passives, in which the participantin the event is always somehow 'affected' -- negatively with the 'rare'passive and positively in the 'morau' passive. The key difference between thetwo seems to be that no causation is involved between the participants in the'rare', negative passive, whereas there is in the positive 'morau'. Thiscausation, it is argued, is the source of the beneficiary reading of this'affective passive'.

Another Indo-European light verb passive is found in Swedish 'Få and itspassive complement', and Eva Klingvall proposes that the Swedishget-constructions are structurally divided. Having outlined the previousresearch on and structure of the få constructions, Klingvall goes on toexemplify causer versus beneficiary subjects in these constructions in somedetail. Her structural analysis of beneficiary and causer få-constructionsconcludes that the two differ in their syntactic behaviour and thus havediffering structural representations. The passive complement of få is analyzedas being the past participle when its subject is a causer, whereas the subjectbeing a beneficiary will cause the complement of få to be the received DP.

Bjarne Ørsnes investigates 'The Danish reportative passive as a non-canonicalpassive' and finds them to be fully compositional passives. Theseconstructions would be deemed non-canonical only due to their treatment of thesubject of the passive verb -- an issue that arises due only to thetraditional definition of passive. Instead of being demoted as one mightexpect, these passives promote their subjects to subjects of the matrix verb.Interestingly, Ørsnes's HPSG analysis also proposes a lexical rule whichsuggests that reportative passives may be uniquely available tosubject-prominent SVO languages.

The final chapter, '(Non-)canonical passives and reflexives: deponents andtheir like', is the only one to offer a diachronic take on passive-likestructures, relating passive-form 'deponent' verbs in Latin (and Greek) to the'inherent reflexives' of Germanic and Romance languages (as well as others).Dalina Kallulli proposes that an actor-initiator feature is present in theseconstructions, without the presence of an external argument. These deponentverbs (cf. passive 'amor', I am loved vs. deponent 'mi:ror', I admire), whichKallulli considers to be canonical passives, are formed from nouns/adjectiveswhich, according to this analysis, makes them more likely to form psych verbs.The fact that the same analysis can be extended to inherent reflexives such asFrench 'le vase casse / le vase se casse', as well as evidence that bothconstruction types transitivize by means of overt morphological marking, isused to argue that both lack external arguments. Kallulli's analysis alsosupports Embick (1997) initial proposal of transitive deponent verbs as psychverbs.

EVALUATIONAs might be expected from a collection of conference papers, the contributionsvary in many ways. Whilst largely data-driven, the chapters vary in theirorientation towards theoretical implications. Certain chapters stand out asproviding a greater contribution towards an overall understanding of thesyntax and semantics of what are supposed to be non-canonical passives, whilstothers at the very least contribute language-internal analyses, giving aricher view of the phenomena under review. Several may shape future researchinto placing those constructions in linguistic theory, such as Wanner’s corpusstudy on get-passives and Lenz’s dialectal investigation into the Germanget-passive.

Andrew McIntyre's contribution stands out as one of the more theory-heavy, yetaccessible chapters, providing an original subdivision of much-discussedEnglish adjectival passives in both their semantics and their syntacticrepresentation. Similarly, a number of the chapters make strong conclusionsabout their data, such as Lavine on Balto-Slavic, Heinat and Manninen onFinnish, Labelle on French, Klingvall on Swedish and Ørsnes on Danish, and asa consequence will surely have implications for anyone working on similarphenomena either in the languages in question or in non-canonical passives ingeneral.

The scope of this volume is very broad, as non-canonical passives have beenshown to take many different verbal-forms and so will be relevant toresearchers in many areas of syntax, semantics or their interface. One onlyhas to turn to the final chapter to question how interrelated languages canshare syntactic structures among such constructions or processes asreflexivization, transitivization, and causation, from either a syntactic,semantic or even morphological point of view. The editors have very carefullyput together a volume in which the weaker contributions are few and farbetween. The book passes through related constructions, languages and notionswhilst keeping the original contributions varied in linguistic impact, movingbetween topics within the same language, then between languages on the sametopic. The empirical base remains constant whilst the theoretical persuasionis far from static, leaving the reader with a collected view -- although farfrom comprehensive of course -- of the problematic analyses of passives.However, the problems of passives are news to no one and volumes of previouslinguistic analyses have been dedicated to such studies, including thosetreating non-canonical passives, such as Perlmutter (1978), Siewierska (1984),Shibatani (1985), Keenan and Dryer (1987/2007), Anagnostopoulou (2003),Abraham & Lesiö (2006). This collection aims to answer some specificquestions, repeated from above.

- How is a get-passive different from a be-passive and why is get used overany other verb?- Why are only certain auxiliaries used with passives?- Why are there different types of adjectival passives?- Which restrictions on verb classes can be observed?

These questions have been answered in specific contributions, on the whole.However the biggest test for any volume on voice phenomena or non-canonicalconstructions is how they contribute to our understanding of the canon.

The editors ask, ''What are the properties of passivization?'' and ''Dopassives have the same properties across languages?'' (p.2). Unusually, forwork on passives, the individual contributions here do go some way toanswering at least the first question, proposing to reject some constructionsas non-canonical passives (haben-statives in German) and to include others asnewly considered passives (Finnish agreeing passives). Most chapters makesmall steps to advance the overall view of passives within linguistic theory,but it is always tempting to ask, as the contributors do, how we can discussnon-canonical passives with only a pre-theoretical and fluid idea of what acanonical passive should be. This work makes no claim to answer such aquestion, but perhaps its incremental approach to narrowing down the syntacticand semantic structures of what are valency-reducing or passive-like in alanguage are the most cautious way to proceed to an overall picture ofpassivization entails.

REFERENCESAbraham, W. & Lesiö, L. 2006. “Passivization and Typology: Form and Function.”[Typological Studies in Language 68]. Amsterdam: John Benjamins.

Anagnostopoulou, E. 2003. Participles and voice. In Artemis Alexiadou, MonikaRathert & Arnim von Stechow (eds.), “Perfect Explorations”. Pp. 1-36. Berlin:Mouton de Gruyter.

Baker, M., Johnson, K. & Roberts, I. 1989. Passive arguments raised.“Linguistic Inquiry” 20: 219-252.

Borer. H. 2005. “The Normal Course of Events”. Oxford: Oxford UniversityPress.

Cinque, G. 1990. Ergative adjectives and the lexicalist hypothesis. “NaturalLanguage and Linguistic Theory” 8: 1-39.

Embick, D. 1997. Voice Morphology, “Syntax and Inherent Specification”. PhDdissertation, University of Philadelphia.

Embick, D. 2004. On the structure of resultative participles in English.“Linguistic Inquiry” 35(3): 355-392.

The Freiburg-Brown Corpus (FROWN). ICAME Collection of English LanguageCorpora, Second Edition, 1999.

Huang, James C.-T. 1999. Chinese passives in comparative perspective. TsingHua Journal of Chinese Studies 29: 423-509.

Keenan, Edward L. and Matthew S. Dryer. 2007. Passive in the world’slanguages. In Timothy Shopen (ed.), “Clause structure, language typology andsyntactic description” (Vol. 1). Pp. 325-361. Second Edition. Cambridge:Cambridge University Press.

Kratzer, A. 1993. The event argument and the semantics of voice. Ms,University of Massachusetts at Amherst.

Kratzer, A. 2000. Building statives. “Berkeley Linguistic Society” 26:385-399.

Meltzer-Asscher, A. 2011. Adjectival passives in Hebrew: Evidence forparallelism between the adjectival and verbal systems. “Natural Language &Linguistic Theory” 29(3):815-855.

Doron, E., & Labelle, M. 2010. An ergative analysis of French valencyalternation. In Herschensohn, J. (ed.), “Romance Linguistics 2010: SelectedPapers from the 40th Linguistic Symposium on Romance Linguistics (LSRL),Seattle, Washington, March 2010” (Vol. 318). Pp. 137-154. Amsterdam: JohnBenjamins.

Orfitelli, R. 2011. Parsimony in passivization: Lexically defining the corecharacteristics of the get-passive. In “Workshop on Non-canonical passives,University of Göttingen”. Pp. 23-25.

Perlmutter, D. M. 1978. Impersonal passives and the unaccusative hypothesis.In “Proceedings of the annual meeting of the Berkeley Linguistics Society”(Vol. 4). Pp. 157-190.

Shibatani, M. 1985. Passives and related constructions: A prototype analysis.Language 61:821-848.

Siewierska, Anna. 1984. The passive: A comparative linguistic analysis.London: Routledge.

ABOUT THE REVIEWERLaura Arman is a PhD candidate in the department of Linguistics and EnglishLanguage at the University of Manchester. Her PhD research explores the WelshImpersonal construction, which has often been compared with the Welshget-passive. The focus of her PhD is the comparison of these two structuresand the argument structure of both.

Page Updated: 11-Nov-2013