"Kissine offers a new theory of speech acts which is philosophically sophisticated and builds on work in cognitive science, formal semantics, and linguistic typology. This highly readable, brilliant essay is a major contribution to the field."
Date: Tue, 15 Jun 2004 13:00:39 +0100 From: Nicole Dehé <nicole@linguistics.ucl.ac.uk> Subject: Non-Projecting Words: A Case Study of Swedish Particles
Toivonen, Ida (2003) Non-Projecting Words: A Case Study of Swedish Particles, Kluwer Academic Publishers, Studies in Natural Language and Linguistic Theory 58.
Nicole Dehé, University College London
INTRODUCTION
Ida Toivonen's book contributes to the discussion of the verb particle combination (VPC, also referred to e.g. as phrasal verbs, particle verbs or separable complex verbs). It focuses primarily on Swedish, which sets it apart from most previous literature in the field. The book is divided into seven chapters plus an appendix. The chapters are well organised with an introduction/overview at the beginning and a summary or conclusion at the end of each chapter. In Swedish, as opposed to other Germanic languages, the particle must immediately follow the verb it occurs with, preceding the nominal object (e.g. Pia sparkade UPP bollen, 'Pia kicked up the ball'; *Pia sparkade bollen UPP). In her monograph, Toivonen argues that this is due to the phrase structural realisation of the Swedish VPC which differs from other languages such as English or Danish. Her analysis is motivated on empirical grounds and formulated within the theoretical framework of Lexical-Functional Grammar (LFG).
EVALUATION BY CHAPTER
Chapter 1 serves as an introduction to the phenomenon at hand: it provides examples of the Swedish VPC, establishes (traditional) criteria to distinguish between particles and related categories, and offers a short, useful introduction to LFG as the framework employed in the study. The Swedish verbal particle is defined as ''a non-projecting word which is adjoined to V°'' (p. 4). The particle is base-generated in its adjoined position. It is crucial to understand that the particle is not part of a morphologically complex V; rather, head adjunction occurs in syntax.
Chapter 2 is intended to further motivate and explain the proposed analysis which is based on two ''distributional facts'' (p. 16). Firstly, particles and full phrases cannot occur in identical positions, and secondly, particles cannot be modified. Toivonen also argues that there is no syntactic category 'particle', but rather that elements of all lexical categories can function as particles. The special status of particles is thus merely due to their phrase structure realization (p. 19). Note that other authors have argued in favour of a category 'particle', among them Olsen (1998) for English and German. Interestingly, Toivonen and Olsen arrive at opposite conclusions on the basis of similar arguments related to some distributional properties of particles. Toivonen's work on Swedish thus seems to confirm the special distributional status that has previously been attributed to particles in other languages. As opposed to previous literature, however, this status does not translate into a separate category, but is taken care of by the idea that particles are non-projecting words. The author then discusses alternative analyses previously suggested in the literature:
(1) She argues that the particle cannot be a (PP-) projecting head e.g. in terms of particle positions (particles, but not full PPs precede the object) and in terms of phrase structure (if particles were prepositions, there wouldn't be a structural difference between particles and true intransitive prepositions).
(2) A morphological approach to VPCs is rejected on the basis of two main arguments: (i) verbs and particles can be separated in c-structure and (ii) whereas morphological constructs are head-final in Swedish, VPCs are head initial (i.e. the verb precedes the particle).
More precisely, Toivonen argues that since compounds in Swedish are head- final, and particle verbs are head-initial, VPCs cannot form a morphological unit. However, the direction of headedness clearly does not provide conclusive evidence since the same difference between compounds and VPCs holds in English, a language for which Toivonen assumes the morphological analysis. Toivonen gives examples where Swedish VPCs enter into morphological processes such as passive formation, A-formation and N-formation, in which the particle ends up to the left of the verbal base and the derivative is non-separable in syntax, showing that the derivative, but not the underlying VPC forms a real morphological unit. Here the comparison with English becomes even more interesting. In English both the VPC and the derivative are left- headed (the looking up of the information/*the up-looking ...), while compounds are right- headed. However, English VPCs are analysed by Toivonen as complex heads. If VPCs are compared to compounds, and if exceptional left-headedness is allowed in English, then the direction of headedness in general cannot provide conclusive evidence regarding the complex head status of VPCs in Swedish. The author returns to this issue in Chapter 6. (3) The idea that particles might be clitics is rejected mainly on stress-related grounds: whereas clitics are phonologically weak, Swedish particles are accented. This observation has also been made in previous work on German (e.g. Wurmbrand 1998) and English (Fraser 1976, Nespor & Vogel 1986, Dehé 2001, 2002 among others).
Chapter 3 modifies and reviews the theory of c-structure in LFG and addresses phrase structure (X' theory) in order to accommodate non- projecting words such as Swedish particles. I'll be concerned mainly with one point related to phrase structure. In the framework employed here, each projecting category is realized in three levels: X-X'-XP, i.e. if an X is present it projects an X' which in turn projects an XP. On the other hand, the presence of an XP does not entail the presence of a head-X (i.e. we do get both structures like Y'-XP-X'-X and Y'-XP- X', but not Y'-X'-X). Structures that directly violate X-bar-theory are not allowed, which is why a structure such as VP[V'[V P]] is impossible (here the P fails to project a P'). It also follows from these constraints on phrase structure that the particle cannot accompany the V to I in V2 contexts. This would lead to a structure such as IP[I'[V[V Prt]]] where V wouldn't project V'. As regards adjunction, it follows that only non-projecting heads are allowed to head-adjoin. A structure like I[I V] is thus out, whereas V[V Prt] is allowed because a particle doesn't project. These assumptions about phrase structure are crucial for Toivonen's analysis. They account e.g. for the ungrammaticality of particle pied-piping in V2 contexts and for the ungrammaticality of the order where the particle follows the object. The latter is accounted for in terms of economy, such that the order V Obj Prt is less economical than the order V Prt Obj because it involves more structure. This idea is also further discussed in Chapter 4 (pp. 101-103) and in the Appendix.
At this stage, one point about phrase structure that deserves attention remains unaddressed. If projecting words must be ''immediately dominated'' by an X'(p. 66) and V° is a projecting head, then why is it possible for V° to occur in the adjunction structure V°[V° Prt], since clearly the lower V° is not immediately dominated by V', but only the higher V° is. I suppose this problem could be addressed and taken care of e.g. along the lines of the category-segment distinction of Chomsky (1986, 1995). As it stands, however, it seems to me that the lower V° violates the principles of X-bar-structure as employed here. In Chapter 4 Toivonen mentions that it ''is important that our theory of phrase structure permits recursive head-adjunction'' (p. 99), but unfortunately an explanation as to how this can be accounted for is not offered.
Chapter 4 further discusses and empirically supports the analysis. It also addresses apparent problems related to word order and provides suggestions as to how they can be accounted for in the present framework. Crucially, there is a Swedish-specific constraint that holds that verbal particles adjoin as non-projecting heads to V°, and V° only. Toivonen provides evidence for the adjunction structure from topicalization (only verb and particle together can be topicalized) and coordination (a VPC can be coordinated with another V°), (pp. 94-99). It seems to me that these facts do indeed support the assumption that the VPC forms a syntactic constituent of some kind at some point, but not necessarily the head-adjunction analysis.
In this chapter, Toivonen also draws the reader's attention to the fact that Swedish particles immediately followed by a PP can sometimes follow the nominal object (pp. 103-107). She suggests that in these cases the particle is in fact a projecting preposition which modifies the following PP, rather than a non-projecting particle. This is reminiscent of what has been argued for English elsewhere (Olsen 1998, 2000, Dehé 2002), namely that particles and prepositions are homophonous and that there often is a structural ambiguity between VPCs and adverbial structures which can be disambiguated along the lines of the scope of an additional modifier ('right' or 'straight').
Chapter 5 addresses the semantics of the Swedish VPC. In particular, three groups are identified: resultative constructions, aspectual constructions, and idiomatic verb- particle combinations. No structural difference is assumed between the different classes: differences are only related to meaning. There is also a section on argument structure which basically argues that if a verb is used in a VPC, the number and type of arguments it takes can be different from those it takes when it occurs on its own. As regards the semantic groups, a threefold classification of this kind has long been suggested in the literature for other languages (Emonds 1985 among others). Similarly, the influence of the particle on the argument structure of the verb has frequently been discussed at least for Dutch, English and German (Booij 1990, McIntyre 2001, 2003, Olsen 1998, Zeller 2001). As far as I can see, the results that Toivonen arrives at for Swedish fit into the overall pattern.
Chapter 6 offers a comparison between the VPC in Swedish and three other languages: Danish, German and English.
1) For Danish (pp. 160-162): Although the elements under question correspond closely to particles in Swedish in both meaning and form, Toivonen argues that Danish has no particles in the structural sense at all (i.e. under the definition where particles are non-projecting words that adjoin to the verb). Evidence comes from the observation that in Danish all particles must follow the object. They also appear to be modifiable and thus display the same behaviour as PPs.
2) For German (pp. 162-166), the same analysis as for Swedish is proposed except for word order. One piece of evidence comes from the fact that particles always attach outside a prefix-V combination, i.e. prefix and V do, but particle and V do not form a morphological unit.
3) The discussion of the English VPC (pp. 166-176) leads Toivonen to the conclusion that there is a crucial structural difference between the two languages: While a VPC in English is seen as a morphological unit, verb and particle are syntactically combined in Swedish (via head- adjunction). She explains that in English, but not Swedish, particles can optionally follow the direct object, and particles that appear verb adjacent in English seem to have a ''tighter connection'' (p. 169) to the verb than their Swedish counterparts (illustrated along the lines of coordination/gapping examples). Some possible counter- arguments, such as the frequently mentioned separability of verb and particle in the syntax and the fact that the particle can be modified when following the object, are taken care of by the assumption that the particle projects a PP of its own when following the object, but forms a morphological unit with the V when verb-adjacent. This seems to me to be the 'default analysis' which has been around for many years.
It is also worth mentioning that in English economy does not seem to have the same effect as in Swedish since in English it is possible for the particle to follow the object, thus violating economy under the present assumptions. This seems odd, in particular because in the Appendix Toivonen claims as a merit of her analysis that Swedish particles provide empirical evidence for the Economy principle (e.g. Bresnan 2001). Economy is argued to be powerful enough to rule out the V Obj Prt order in Swedish (''Without Economy, the ungrammaticality of [V Obj Prt] would be unexplained. ... Swedish particles show that ... an Economy principle is warranted in the grammar''; p. 103). If taken as evidence for economy in language in general, this seems neutralized by the fact that in English economy does not seem powerful enough to rule out an otherwise parallel construction. Under the assumptions about economy pursued here, it remains unclear to me why the V Obj Prt order is possible with unmodified particles in English. To put it differently: If ''[w]ithout Economy, the ungrammaticality of [V Obj Prt] would be unexplained'' in Swedish, then with economy, the grammaticality of [V Obj Prt] in English remains unexplained. It should at least be mentioned that in English (and presumably other languages such as Norwegian and Icelandic) there seem to be other factors at work which interact in such a way that they win over economy. This then is reminiscent of Grimshaw's (2001) hypothesis that economy is enforced by collective effects of general constraints on phrase structure, rather than by an economy principle.
Chapter 7 summarises the results of the previous chapters and serves as a conclusion. It is followed by the Appendix on Economy referred to above.
CONCLUDING REMARKS
Although this work was intended as a language specific study, it would have immensely benefited from more comparative remarks and more frequent references to related work. This holds in particular for the classification in Chapter 5 and for the empirical discussions in Chapters 2 and 4. In Chapter 6, Swedish is compared to other languages, but justice is not done to the huge amount of previous work on the VPC in these languages. Furthermore, it remains to be investigated if it is necessary to assume a language specific phrase structural realisation for particles or whether word order variation between languages follows from more general constraints on phrase structure. Unfortunately, there are also quite a few typing mistakes, errors in references, and misleading index entries.
Despite these drawbacks, Toivonen's book is a very welcome contribution to the ongoing discussion of the verb particle construction, in particular because it provides (at least to my knowledge) the first extensive study of the VPC in Swedish. Not least because of the wealth of data it provides, Toivonen's book will be an invaluable reference work for anyone doing specific or comparative work on the (Swedish) VPC.
REFERENCES
Booij, G. 1990. ''The boundary between morphology and syntax: Separable complex verbs in Dutch.'' In Yearbook of Morphology 1990: 45-63.
Bresnan, J. 2001. Lexical-Functional Syntax. Oxford: Blackwell.
Chomsky, N. 1986. Barriers. Cambridge, MA: MIT Press.
Chomsky, N. 1995. The Minimalist Program. Cambridge, MA: MIT Press.
Dehé, N. 2001. ''Intonation patterns of particle verb constructions in English.'' In NELS 31: 183-197.
Dehé, N. 2002. Particle Verbs in English: Syntax, Information Structure and Intonation (= Linguistik Aktuell/Linguistics Today 59). Amsterdam/Philadelphia: John Benjamins.
Emonds, J. E. 1985. A Unified Theory of Syntactic Categories. Dordrecht Foris.
Fraser, B. 1976. The Verb-Particle Combination in English. New York/ San Francisco/ London: Academic Press.
Grimshaw, J. 2001. ''Economy of structure in OT''. Ms., Rutgers University. Available as ROA 434-0601 at Rutgers Optimality Archive.
McIntyre, A. 2001. ''Argument blockages induced by verb particles in English and German: Event modification and secondary predication''. In N. Dehé & A. Wanner (eds.), Structural Aspects of Semantically Complex Verbs, pp. 131-164. Berlin/ Frankfurt/ New York: Peter Lang.
McIntyre, A. 2003. ''Preverbs, argument linking and verb semantics.'' In Yearbook of Morphology 2003: 119-144.
Nespor, M. & I. Vogel. 1986. Prosodic Phonology (= Studies in Generative Grammar 28). Dordrecht: Foris.
Olsen, S. 1998. ''Prädikative Argumente syntaktischer und lexikalischer Köpfe - Zum Status von Partikelverben im Englischen und Deutschen.'' In Folia Linguistica 16: 301-329.
Olsen, S. 2000. ''Against incorporation''. In J. Dölling & T. Pechmann (eds.), Linguistische Arbeitsberichte 74, pp. 149-172. University of Leipzig: Linguistics Department.
Wurmbrand, S. 1998. ''Heads or phrases? Particles in particular.'' In W. Kehrein & R. Wiese (eds.), Phonology and Morphology of the Germanic Languages. (= Linguistische Arbeiten 386), pp. 267-295. Tübingen: Niemeyer.
Zeller, J. 2001. ''How syntax restricts the lexicon: Particle verbs and internal arguments.'' In Linguistische Berichte 188: 461-494.
ABOUT THE REVIEWER:
ABOUT THE REVIEWER
Nicole Dehé is an Honorary Research Fellow at UCL, Dept of Phonetics
and Linguistics. She received her PhD in 2001 from the University of
Leipzig. She is the author of Particle Verbs in English: Syntax,
Information Structure and Intonation, 2002, Amsterdam: Benjamins, and
co-editor of a volume on particle verbs (Verb-Particle Explorations,
2002, Berlin/New York: Mouton de Gruyter; with Ray Jackendoff, Andrew
McIntyre and Silke Urban).
