From: Ludwig Fesenmeier <ludwig.fesenmeieruni-koeln.de>
Subject: Unaccusative Verbs in Romance Languages
Announced at http://linguistlist.org/issues/17/17-1594.html
Author: Mackenzie, Ian E.Title: Unaccusative Verbs in Romance LanguagesPublisher: Palgrave MacmillanYear: 2006
Ludwig Fesenmeier, Department of Romance Languages, University of Cologne
The purpose of the book is to question the Ergative Analysis, as discussedon the basis of Romance languages. One here can identify, amongintransitive verbs, a subtype named unaccusative or ergative verbs whosemembers share several syntactic features. These features are generallyassumed to be ''related to the fact that the apparent subjects of the verbsin question [...] are 'deep' objects, and in some cases surface objectsalso'' (p. 2). The author aims to show that a syntactic account of theseproperties is ''both theoretically unnecessary and empirically inadequate''and that ''alternative explanations are readily available'' (p. 5).
The book contains eight chapters, preceded by ''List of Tables'' (p. ix),''Acknowledgement'' (p. x) and ''Sources of Historical Examples'' (pp. xi/xii).The main text is followed by ''Notes'' (pp. 187-211), ''References'' (pp.212-219) - including the editions mentioned on pp. xi/xii - and ''Index''(pp. 220-230), containing both names and concepts.
The first chapter, ''The Ergative Analysis and the Unaccusative Hypothesis''(pp. 1-16) introduces the subject, providing the general background andbriefly presenting most of the syntactic features to be discussed in thefollowing six chapters. The final chapter, ''Conclusion'' (pp. 182-186),offers a short synthesis of the results achieved. As just mentioned,chapters 2 to 7 (ranging in length from 10 up to 58 pages) investigate insome detail the different properties of the phenomena at issue and proposerespective ''alternative explanations'', each chapter concluding with a briefsummary.
The property dealt with in chapter 2 (pp. 17-38) is ''Expletive Inversion'',since this construction ''has often come to be seen as an unaccusativediagnostic'', even if it ''is not the most widely cited'' (p. 18) as such. TheRomance examples are taken from French as a non-null subject language. Theauthor discusses two approaches in establishing ''a link between therelevant empirical data and the 'deep-object' analysis of unaccusativesubjects'' (p. 37): expletive inversion is incompatible with unergativeverbs since ''the expletive originates in the position in which unergativesubjects originate'' (pp. 37-8) or since such constructions ''call forpartitive/inherent case on the [expletive] associate'' (p. 38), which byhypothesis is excluded for unergative verbs. Adducing some counterexamples,he concludes that both analyses are unreliable and proposes an alternativeapproach, according to which it is the ''striking presentational capability''of most of the so-called unaccusative verbs, ''rather than their syntax,that explains their frequency in the expletive construction'' (p. 33). Moregenerally, the function of the expletive construction, which occurs notonly with unaccusative verbs (ex. 1) but also with passives (ex. 2),reflexive passives (ex. 3) and unergative verbs (ex. 4), would be that ofpreventing ''the verb from being accented and hence focused'' (p. 38):
1. IL EST ARRIVE quelque chose de très drôle. (ex. 1, p. 17)2. IL A ETE ELU beaucoup de femmes. (ex. 89, p. 36)3. IL SE CONSTRUIT beaucoup d'immeubles dans cette ville. (ex. 70, p. 31)4. IL A REGNE un silence de mort. (ex. 76, p. 33)
The topic of the third chapter is ''Partitive Cliticization'' (pp. 39-69;henceforth PC), which refers to ''the quantified noun pattern of extraction''(p. 40) as realised in the clitic pronouns NE and EN in Italian and Catalan:
5. Se NE sono perduti sette. (ex. 5, p. 39)6. Jo l'altre dia EN vaig veure un a la Rambla. (ex. 6, p. 39)
The argumentative relevance of PC resides in allowing systematicdistinction between unaccusative and unergative subjects. The author setsout with a critical evaluation of various formal-syntactic accounts andconcludes that none of them provide a compelling reason ''to expect that theoperation [= PC] will be impossible from unergative postverbal subjects''(p. 54). The author reasonably makes his point with, among others, the twofollowing examples containing unergative verbs:
7. Es poden inscriure 12 jugadors per equip i al ser de futbol 7, ENJUEGUEN 7. (ex. 44, p. 55)8. Al CNR lavorano 7.500 persone, mentre al CNRS NE LAVORANO 26.000. (ex.48, p. 55)
Subsequently, the author presents a different approach which links thelicensing of PC to two types of information structure: according to thisview, PC would only be possible with the subjects of unergative verbs inthe case of wide focus, while unaccusative verbs would license PC with bothwide and narrow focus. However, as is immediately pointed out, examplessuch as 8 and others (see pp. 57/58), have in fact narrow focus.Nevertheless, while stating the ''general absence'' of ''a systematicunaccusative-unergative asymmetry'' (p. 60) as far as PC is concerned,Mackenzie well acknowledges that unaccusative subjects turn out to be moreeasily compatible with PC than their unergative counterparts. Taking intoaccount not only Italian and Catalan data, but also some French sentenceswhich present expletive inversion, he suggests that the possibility of PCwith the subjects of intransitive verbs depends on the presentationalcapability or incapability of the latter. Such a view is confirmed,according to the author, by the PC-behaviour of adjectives (stage-level vs.individual-level predicates, see pp. 63-4) and of ''(indefinite) directobjects'' (p. 65).
Chapter 4 (pp. 70-102) is dedicated to ''Bare Subjects'' whose distributionin the Romance null subject languages is often assumed to correlate withthe unaccusative-unergative distinction (see p. 70), i.e. only unaccusativeverbs would be compatible with these types of subjects. The data are mainlytaken from Spanish, ''but the main conclusions would in broad terms beapplicable to Catalan, Portuguese and Italian'' (p. 70). The author firstoffers some sentences containing unergative verbs and postverbal baresubjects, concluding that ''the occurrence of bare subjects with unergatives[...] is a routine matter'' (p. 77). He then goes on to show that thepossibility of verbs to have bare subjects, as opposed to overtlyquantified subjects, depends largely on two semantico-pragmatic factors:bare subjects are given preference ''when, and only when, thequantificational meaning associated with the latter is incompatible withthe context or with the speaker's target assertion'' (see ex. 9 and 10below); bare subjects are also preferred when the 'verb + baresubject'-construction ''occur[s] in contexts in which an appropriatenon-subject argument can be determined'' (p. 101). Such would be therequirement for the assertion to be complete. This is due to the fact that''a complex consisting in a verb and a (non-generic) bare noun correspondsto a monadic predicate'' (p. 90; see ex. 11 below).
9. (a) De ese agujero ESTUVIERON SALIENDO HORMIGAS durante tres horas. (b) ??De ese agujero ESTUVIERON SALIENDO UNAS HORMIGAS durante treshoras. (ex. 53/54, p. 84)
10. (a) PARTICIPAN NIÑOS en muchas guerras africanas. (b) ??PARTICIPAN UNOS NIÑOS en muchas guerras africanas. (ex. 66/67,p. 87-8)
11. (a) POR AQUÍ pasan trenes. (b) ?Pasan trenes. (ex. 77a/b, p. 92)
Besides, the following observations also appear to be of some relevance:the distinction between individual-level and stage-level predicates (seepp. 95-99 and ex. 12 below) and the condition of the bare subject to formthe informational nucleus (see pp. 99-101 and ex. 13 below).
12. (a) Me GUSTAN las ostras. (ex. 88, p. 97; individual-level predicate) (b) Me APETECEN ostras. (ex. 89, p. 97; stage-level predicate)
13. (a) Hablaron expertos INTERNACIOALES. (ex. 113, p. 100) (b) ?Hablaron expertos. (ex. 112, p. 100)
''Perfect Auxiliary Selection'' is the topic of chapter 5 (pp. 103-161),which deals mainly with Italian, since ''the Ergative Analysis and theUnaccusative Hypothesis are inspired primarily by the Italian situation''(p. 103): it is commonly assumed that intransitive Italian verbs which takeESSERE as their perfect auxiliary are unaccusative, these verbs beingconsequently put in a group together with passives, raising predicates(PARERE and RISULTARE, among others) and reflexives. In the first part ofthe chapter, Mackenzie discusses (purely or, at least, primarily) syntax-and semantics-based approaches which are both committed to an analysis instrictly synchronic terms. It is precisely this perspective the authorrejects as inadequate, arguing, instead, in the second part that ''thelimited semantic regularities'' observed in auxiliary selection, especiallywhen one compares French and Italian, ''are a residue of an earlier moreunified semanticism'' (p. 130). The basic idea put forward is that in lateLatin/early Romance constructions containing a form derived from Latin ESSEand a perfect participle are to be interpreted as 'copula + adjectivedenoting a resultant state', the ESSE-form being considered a state verb(see p. 137). Mackenzie argues that the modern situation results from a''process of conventionalization, whereby specific auxiliary assignmentsbecame categorically associated with specific verbs or with specificstrands of meaning within the overall semanticism of given verbs'' (p. 161).Such would be the case with stative verbs which are systematicallyambiguous depending on their state and achievement readings (see ex. 14below). Another case would be verbs that formerly had an achievementreading and lost it (see ex. 15 below).
14. (a) AVANZA della pasta. (State) (ex. 138a, p. 147) (b) È AVANZATA della pasta. (Achievement) (ex. 138b, p. 147)
15. Per grande spazio BASTÒ il rovinìo delle pietre che cadevano giù. (ex.154, p. 151)
Chapter 6 (pp. 162-171) deals with ''Past Participle Agreement'' as apresumed diagnostic for unaccusativity. The author shows at first thecircumstances under which participle agreement can be observed in Frenchand Italian: intransitive (including reflexive) verbs which takeESSERE/ÊTRE as auxiliary, transitive reflexive verbs and direct objectclitics (ex. 9/10, p. 163: Giovanni LA ha accusatA, Je LES ai achetéS).According to some formal-syntactic approaches, these phenomena can be givena unified explanation and, thus, ''past participle agreement phenomena canbe seen as providing indirect support'' for the ''ergative analysis ofunaccusative verbs'' (p. 165). Nevertheless, given the assumptions in (atleast some of) the approaches discussed here, one would expect lack ofagreement in the passive (ex. 18, p. 166: Furono catturatI/*catturatOquattro presunti membri dell'ETA), while agreement ''might seem to be MORElikely'' (p. 166) in cases such as ''Ils ont FAIT/*FAITES des recherches''(ex. 21, p. 166). Thus, as far as past participle agreement is concerned,''no single principle can be expected to have general applicability'' and,therefore, such phenomena ''cease to count as evidence for the ergativeanalysis of unaccusative verbs'' (p. 167). Mackenzie suggests consideringthe modern synchronic situation as the result of a fragmentation process ofa ''genuinely unified pattern'' (p. 168) in Latin, i.e. ''the participleagrees with its argument'', as in ''[In ea provincia pecunias magnascollocatas] habent.'' (ex. 24, p. 168; ''[...]'' indicates a small clause).
Chapter 7 (pp. 172-181) is about ''Participial Absolutes'', more preciselyabout ''a subset of small clause-type constructions involving pastparticiple'', where only the participle of an unaccusative verb ''co-occursand agrees with a lexical subject in the same clause'' (p. 172), while suchconstructions would not be possible together with unergative verbs:
16. Salidos los padres jesuitas [...] (ex. 1, p. 172)17. *Jugados los niños [...] (ex. 2, p. 172)
It has been shown, though, that such a restriction holds only in the caseof an overt subject (''BUSSATO alla porta, Gianni entrò'', ex. 7, p. 173).Mackenzie instead argues that ''there is no compelling syntactic motive forassuming unergative subjects to be incapable of occurring in participalabsolutes'' (p. 179) and offers indeed some counterexamples from Italian andSpanish (see ex. 18 and 19 below). He nevertheless acknowledges the''undeniable unergative-unaccusative asymmetry in terms of overallproductivity'' of such constructions (p. 179).
18. Una volta BOLLITO il brodo, immergerci la zucca in pezzetti. (ex. 20,p. 178)19. CENADOS los niños, salimos al cine. (ex. 23, p. 178)
The explanation offered for this asymmetry resides in the fact that theparticipal clause always indicates a resultant state and that, therefore aterm capable of an achievement or accomplishment reading must be present.Since unaccusative verbs typically have precisely such meanings, whileunergative verbs more often than not are activity terms, the asymmetryobserved ''falls out straightforwardly from the aspectual composition of thetwo classes'' (p. 179).
The basic idea of the book under review is to show that the distinctionbetween unaccusative and unergative verbs within the overall group ofintransitive verbs, as encompassed in formal-syntactic approaches, raises alarge amount of problems both as far as theory-internal consistency anddescriptive adequacy are concerned. Furthermore, it turns out thatapparently contradictory empirical findings can be analysed coherently oncedifferent explanations are taken into account.
I think Mackenzie has well shown that it is far from clear how, from anmethodological point of view, the unaccusative/unergative-distinction andthe different phenomena discussed here are related, and that sometimes theempirical data poses some problems with regards to the theoretical claims.
Such problems, then, are theory-internal in nature and can thus only betreated and possibly resolved theory-internally, but, as Mackenzie rightlypoints out, it is equally legitimate to look for other explanations, e.g.within analytical paradigms which differ from the formal-syntactic one.
As far as expletive inversion (ch. 2), partitive cliticization (ch. 3), andbare subjects (ch. 4) are concerned, the author does actually propose analternative explanation in semantico-pragmatic terms. For an explanation tothe occurrence of intransitive verbs in absolute participle constructions(ch. 7), the author resolves to a purely semantic approach.
The problems that arise concerning perfect auxiliary selection (ch. 5) andpast participle agreement (ch. 6) are, by contrast, attributed to a purelysynchronic approach, a situation which according to Mackenzie, is theresult of the evolution of the various Romance languages, diverging from arather unified picture at an earlier point in time: such an analysis thusfalls short of taking into account the possibility of contemporaneousexistence of phenomena which actually have originated at different momentsin the history of the language.
The alternative ideas put forward by Mackenzie can thus be grouped togetherin two basic categories and one might have wished that this difference hadbeen made more explicit and/or would have also been reflected in theoverall organization of the book. Formally adopting his own distinction inthe overall structure of the book would have stressed the doubtful,commonly presumed interdependence between the various phenomena discussed.
A discussion of the problems which, according to Mackenzie, arise by theformal-syntactic analysis vis-à-vis the different matters seems beyond thescope of the present review. What can be said is that the approachesproposed by the author do appear rather promising and undoubtedly worthy offurther investigation. I however will equally not discuss them in greaterdetail, for the data offered for illustration turns out to be ratherheterogeneous:
- except for the historical examples (see in particular ch. 5, but also ch.6), many of the modern examples adduced do not seem to be authentic, butrather constructed (and sometimes unnatural, too, e.g. ''Unos amigosllegaron.'' (ex. 73b, p. 33); on p. 33 is given the example ''Il A RÉGNÉ unsilence de mort'' (ex. 76), an expression which usually appears in theimperfect ''Il RÉGNAIT un silence de mort'');
- undoubtedly, other examples are authentic, but the sources are indicatedonly very rarely (in most cases, they can be retrieved via Google);
- some examples are well attested, but have a quite particular status (seeex. 106 and 107, p. 140 (repeated on p. 174): ''O ci andate già MANGIATI[...]'' (ex. 107): such a construction seems to be exclusive to verbs suchas MANGIARE, PRANZARE, CENARE etc.);
- a few examples, which are supposed to be grammatical, are immediatelyjudged ungrammatical by native speakers (see ex. 41 and 42, p. 25: ''Il aété construit CETTE MAISON en 1890'' (ex. 42); in the mini-dialogue ''-Hotrovato noccioline. -Noccioline, LE ho trovate anche io'' (ex. 16, p. 75),quoted from an article on Spanish, according to native speakers the secondturn is ungrammatical);
Another problem concerns statements such as ''in French [...] passivesselect AVOIR in the perfect'' (p. 12; see also p. 117) which is definitelywrong, since they well select ÊTRE (it is the latter that takes AVOIR asits auxiliary); thus one cannot argue that French passives match up ''ratherconspicuously with unergatives'' (p. 12), as opposed to Italian (see p. 11).It is equally not the case that Italian CAPITOLARE selects ESSERE asperfect auxiliary, as is affirmed on p. 121 and p. 128 with regards to ex.42 and 65. Discussing the 'stop'-sense of (Old) Italian STARE, Mackenziequotes the example given in 20, taken from Dante Alighieri's DIVINACOMMEDIA, and comments that ''presumably SI STANNO is an intransitivereflexive and STEA is transitive'' (p. 152).
20. Se i piè SI STANNO, non STEA tuo sermone. 'If your feet are stayed, donot stay your speech.' (ex. 161, p. 152)
Under the entry STARE in the ENCICLOPEDIA DANTESCA (2. ed, Roma: 1984, vol.V, pp. 409-413), however, one reads that ''[i]n un numero assai ristretto diesempi, s. è impiegato assolutamente a indicare il persistere di una certacondizione o situazione'' (§ 1.1, p. 410), and among the illustrations givenone finds precisely the sentence in question followed by the comment ''(ilsecondo caso)''. Such an expert judgement casts considerable doubt on thetransitive interpretation proposed by Mackenzie.
One thus can state that sometimes the author's handling of the data israther casual, an impression which appears to be confirmed also on theformal level: there are some annoying typographical errors(''iNdentificarsi'' (p. 10, ex. 14), ''assEssino'' (p. 44, ex. 19/20), ''pocChi''(p. 83), ''Sietro'' (p. 137, ex. 99), ''CrestomaNzia'' (p. 212, Arese (ed.)1955)) and incorrect bibliographical references (the title of Schiaffini(ed.) 1926 is TESTI FIORENTINI DEL DUGENTO E DEI PRIMI DEL TRECENTO,without SECOLI (pp. xi and 218); on p. xi Cappelli's edition of IL LIBRODEI SETTE SAVI is stated as being 1865 (date of its first publication),but on p. 213 one reads 1986, while the correct year of the reprint is1968); in many of the historical examples omissions are not indicated (see,e.g., ex. 147, p. 149, where the first part is missing: ''TOUT CE, FET ELE,lessiez ester et [...]''); the Latin sentences are taken from a dictionary,but some of them are indeed authentic, yet slightly different (e.g., ''Aquaconclusa facile corrumpitur'' (ex. 174, p. 156) vs. ''Conclusa autem aquafacile corrumpitur'' (Cicero, DE NATURA DEORUM)).
To sum up, the views proposed in the book under review seem worthy of asystematic in-depth analysis which, however, should be based, whereverpossible, on data derived from authentic language corpora, both spoken andwritten, often easily accessible online (see, e.g., ''Frantext'' for French,the corpus used in the ''Opera del Vocabolario Italiano'', ''CREA'' and ''CORDE''provided by the Real Academia Española, and the ''Integrated ReferenceCorpora for Spoken Romance Languages'' (seehttp://lablita.dit.unifi.it/coralrom).
ABOUT THE REVIEWER
Ludwig Fesenmeier teaches Romance linguistics at the Department of RomanceLanguages, University of Cologne, and is currently working on hispost-doctoral thesis on lexical synonymy in the Romance languages.