LINGUIST List 15.1229

Fri Apr 16 2004

Review: Syntax:Deh�, Jackendoff, McIntyre & Urban(2002)

Editor for this issue: Naomi Ogasawara <naomilinguistlist.org>


What follows is a review or discussion note contributed to our Book Discussion Forum. We expect discussions to be informal and interactive; and the author of the book discussed is cordially invited to join in.

If you are interested in leading a book discussion, look for books announced on LINGUIST as "available for review." Then contact Sheila Dooley Collberg at collberglinguistlist.org.

Directory

  • Annie Zaenen, Verb-Particle Explorations

    Message 1: Verb-Particle Explorations

    Date: Fri, 16 Apr 2004 22:18:50 -0400 (EDT)
    From: Annie Zaenen <zaenenparc.com>
    Subject: Verb-Particle Explorations


    Deh�, Nicole, Ray Jackendoff, Andrew McIntyre and Silke Urban, ed. (2002) Verb-Particle Explorations. Mouton de Gruyter, Interface Explorations 1.

    Announced at http://linguistlist.org/issues/13/13-744.html

    Annie Zaenen, PARC & Stanford University.

    Verb-particle combinations fascinate syntacticians because they don't fall neatly within the distinctions one wants to draw between syntax and morphology and between idioms and compositional constructions. Given this, they bring out the most creative tendencies in linguists and have led to a great variety of treatments. This book contains 14 articles giving various perspectives on and discussing various aspects of verb-particle combination in Germanic languages. One important family of proposals en vogue during the 1990s, however, is not represented, the small clause analysis. The material is divided in two sections: syntactic, morphological and semantic perspectives and statistical and psycholinguistic ones. The language data examined are from English, Dutch, German, Swedish, Norwegian. Some contributions concentrate on a particular language, others contain comparative material.

    The editors define a particle as 'an accented element which is formally (and, often, semantically) related to a preposition, which does not assign case to a complement and which displays various syntactic and semantic symptoms of what may informally be called a close relationship with a verb, but without displaying the phonological unity with it typical of affixes.'(p.3) The discussion in the book shows that the formal relation to a preposition is not taken to be defining for everybody.

    Two articles give useful and well-written overviews of verb-particle constructions in one language and propose a 'constructional' view on the morphology/syntax debate: Separable complex verbs in Dutch: A case of periphrastic word formation by Geert Booij and Ray Jackendoff's English particle constructions, the lexicon, and the autonomy of syntax

    Booij's contribution summarizes the characteristics of Dutch particles and defends a commonsensical, intermediate solution to the morphology/syntax debate: V-P combinations are constructional idioms: they are phrases but they are created in the lexicon. Booij points to the existence of periphrastic forms in inflectional paradigms as a phenomenon of similar nature. His approach appeals to notions of Construction Grammar but is by and large presented in non-formal terms.

    Jackendoff's contribution distinguishes six different semantic classes of verb-particle combinations in English. The fact that they have the same syntax is construed as an argument for the autonomy of syntax. Jackendoff observes that the differences in argument structure and lexical status that, in some of the cases, could be used as arguments for binary structures do not generalize to all the cases and that, if one assumes syntactic uniformity, only a flat VP structure is compatible with all the data. This conclusion goes against much of the recent literature assuming small-clauses and Larsonian shells. In less than thirty pages the article manages to summarize most of the facts and arguments for several different analyses and to make a clear point: if you don't confuse syntax and semantics and take all the facts into account most recent proposals don't cut the mustard and both sides of the small clause/verb+particle complex controversy can be dismissed.

    Jochen Zeller, in Particle verbs are heads and phrases, takes, as his title indicates, the position that verb-particle combinations can be represented in two ways, as heads and as phrases. He aims at a unified approach for OV (Dutch and German) and VO (English and Norwegian) languages. The main ingredient of his proposal is a double syntactic analysis that is obtained through a restructuring of the VP dominating a PartP plus a V0 to a V0 dominating a Part0 and a V0. [The '0' in 'V0' and 'Part0' should be read as superscripts. --Ed.] The conditions on this restructuring and on the various movement rules that interact with the verb-particle combination should account for the facts discussed, mainly the appearance of verbal morphology on the verb, the movement of the verb only to V2 position, the non-scrambling of the particle and its restricted topicalization, the distribution of modifiers within the verb-particle construction. With respect to the latter, I have the impression that the Dutch facts described here are rather oversimplified.

    Contrary to the conciliatory proposals described above, Jaap van Marle's Dutch separable compound verbs: Words rather than phrases? attacks the problem of the morphological versus syntactic status of particle verbs head-on, concluding that they are words and comparing them to compound verbs such as 'mastklimmen' (poleclimb), which he terms DCV. The thesis of the author is that particle verbs and DCV's form one class and that much of the defective behavior of these verbs can be reduced to their inability to occur in verb second position. The discussion is rather sloppy but it brings out a couple of interesting points about verbs like 'mastklimmen'. The description given of their morphological distribution seems more adequate than the one assumed in Booij in this volume. But, in my opinion, these clarifications don't lead to a convincing argument in favor of the morphological view.

    A couple of papers check the constraints on particle preposing that have been proposed in the literature against the data from corpus analysis. And up it rises: Particle preposing in English by Bert Cappelle subjects the English data to this test and Stefan M�ller's Syntax and Morphology: German particle verbs looks at the German facts.

    Cappelle uses 478 sentences from the CobuildDirect corpus to show that the syntactic 'constraints' proposed (only literal meaning, no auxiliaries, only intransitives and only in main clauses) are only tendencies. As far as literal meaning goes, the requirement can be maintained but needs a reinterpretation of what literal meaning is. After observing that there is a tendency to equate literal meaning with spatio-directional meaning, Cappelle shows that it is rather difficult to maintain that all preposed particles have such a meaning. He proposes to define the meaning of a particle as literal ''if its meaning is constant across different verb-particle constructions''(p. 56). This seems reasonable but the paper doesn't discuss the way particle and verb meaning is combined in enough detail to ascertain whether it can indeed be maintained.

    The other constraints on preposing are discussed as consequences of the pragmatics of preposing. One function, bringing the particle into focus, has often been discussed. Cappelle claims there are two others: giving clause final focus to the subject or to the verb. These different functions account for different distributions: when particle- focusing is the main goal, the subject will often be a pronoun, whereas when subject focusing is intended it will be a heavy NP. Different particles seem to have a preference for one of the two possibilities. The no-auxiliary constraint is also not absolute and is supposed to follow as a tendency from the focusing functions of the construction. The analysis goes some ways to explaining the lack of direct objects in particle preposing because it would lead to two focused constituents in a row in the case of subject-focusing. When the subject is a pronoun, and hence not focused, it can co-occur with particle preposing. Full NPs are also shown to co-occur with preposed particles when they are highly salient.

    M�ller's contribution also relies exclusively on examples found in real texts. Using those he shows that most constraints on the preposing of particles in German don't hold water. The article concludes that most of the arguments that have been given for particles as morphological objects cannot be upheld in face of the data. For a formal treatment of German verb-particle constructions the article refers to M�ller's 2000 thesis on Complex Predicates.

    The meaning of particles is an important topic given the claims on particle fronting as we have seen; it is also important in the debate about idiomaticity as opposed to compositionality of meaning. Capelle tries to extend the class of compositional combinations by appealing to a meaning that is constant across verb-particle combinations. Andrew McIntyre also argues for construction-specific meanings in Idiosyncracy in particle verbs. For instance the particle down would have the same meaning in brush down, clean down, wash down, etc. It indicates that ''the action is performed on a substantial part of the entity appearing as an object''. If one accepts construction-specific meanings, many more verb-particle combinations are 'compositional' than under an approach such as the one proposed by Wurmbrand (2000) where only particles that can occur in a copular construction and that can be contrasted with another one are compositional. This means that for instance the topicalization test which is often used in German and Dutch to determine whether a verb-particle construction is idiomatic or not cannot be take at face value because it seems to be subject to conditions that are stronger than those on the notion of compositionality defended here. Some of the data in M�ller's paper reinforce this point.

    When one accepts MacIntyre's idea of construction-specific meanings for certain particles, restrictions on productivity need to be accounted for. The author discusses some potential solutions to this without coming to a crisp and general account. But in doing so he provides interesting examples of the similarities and differences between English and German particle verbs.

    Two articles are explicitly comparative.

    In Particle placement, Ad Neeleman attempts to account for the differences in particle placement between Dutch and English. In English the particle can show up in various positions depending on the form of the complement (DP or PP), in Dutch the position is next to the verb (in underlying structure) independent of the form of the verbal complement. Neeleman claims that the difference is due to a OV/VO parameter setting in the base. To a non-believer the assumptions that are needed to make this work are ad-hoc but they are mostly stated up- front (with only a couple tucked away in footnotes) and once they are accepted the mechanics hum pleasantly along to account for the basic contrasts in position given in the beginning of the paper.

    Fabrice Nicol in Extended VP-shells and the verb-particle construction starts from an analysis of the English facts and extends it to Scandinavian. His theoretical starting point is Larson's analysis of ditransitives, here recast in the Minimalist framework as the Extended VP-Shell Hypothesis (EVPS). Extra Wps and Xps are used to stuff verbs and particles in and to allow them to move picking up or shedding whatever is needed to get the facts that this paper focuses on: the constraint on wh-extraction out the NP in the VP-NP-Part structure, the difference in grammaticality in the different orders of ditransitive particle combinations, and the difference in particle modification between the V-NP-Particle and the V-particle-NP structure.

    The ratio of assumptions to observations in this paper is beyond what I consider explanatory but aficionados of Minimalist approaches might find it 'on the right track'. The paper is peppered with 'assumptions', 'suggestions', 'conjectures' and 'hypotheses' and reads as a parody.

    Scandinavian verb-particle combinations are also the topic of Ida Toivonen's Swedish particles and syntactic projection. Taking an LFG perspective, she tries to determine whether the idiosyncrasies of particles are due to their special status as a syntactic category or as a grammatical function. She argues that they do not form a special syntactic category but can belong to a number of them and that they are not in a one-to-one correspondence with a grammatical function either (particles can have a resultative, locative or aspectual function and resultatives and locatives can be expressed without particles). She proposes to treat them as non-projecting words that are head-adjoined to the verb. Whether a word is projecting or not is a lexical property. The proposal leads to the introduction of type of a phrase structure rule that is unique in Swedish but that is argued to fall within a typology of clitic-like elements.

    The remainder of the articles do not focus on the syntax of particles but rather on their psycholinguistic aspects. The first one, The influence of processing on syntactic variation: Particle placement in English by Stefan Gries claims that the long list of factors that have been invoked as influencing the choice between the Part-NP or NP-Part order in English can, when indeed relevant, be subsumed under a processing hypothesis according to which a message is formulated to communicate it with as little processing effort as possible. I find this a bit too general to be of much use and in my view the main value of the paper lies in the mono- and multi-factorial analysis that the author presents of 20 variables associated with the choice between the two orders. Unfortunately, it is presented in a very terse form and to my untrained eye not all the correlations seemed convincing, especially given the small set of data. Overall the General Linear Model used is a promising way to approach problems of variation but it might have been overapplied here.

    Dieter Hillert and Farrell Ackerman in Accessing and parsing phrasal predicates observe that the study of particle verbs can throw light on the psychological status of words. One can see the verb-particle combination as one unit from a content point of view and as two units from a formal one. They review some psycholinguistic evidence that seems to suggest that as far as early lexical access is concerned, verb-particles are accessed as semantic units even when they are presented discontinuously in the sentence. They speculate on the general form of access to the mental lexicon which might well be modular, accessing morphology, syntax and semantics separately.

    Lexical processing is also studied in German particle verbs and word formation by Anke L�deling and Nivja de Jong. The paper argues that the verb-particle combinations are not different from other phrasal entities such as resultatives and that the main difference is between opaque and transparent combinations. Their dictionary and corpus studies show that they are mainly opaque combinations that undergo word formation processes (mainly -ung nominalizations) but their psycholinguistic experiment shows that opaque and transparent verb- particle combinations behave the same way with respect to lexical priming: in both cases the reaction times correlated with the size of the morphological family. This points to the conclusion that even opaque combinations have composite entries in the mental lexicon. These results are different from those reported in the previous paper but the tasks are also different (the material presented in the previous article consisted of sentences; here individual words were presented) as is the type of complexity assumed: semantic versus morphological. As far as the main thesis proposed in this paper, the experiment seems not to have provided conclusive evidence.

    The last paper, Parsing verb particle constructions: An approach based on event-related potentials (ERP) by Silke Urban, uses a different method to shed light on the complexity of verb-particles and on the relation between simple verbs and verb-particle combinations. Event- related potentials are obtained by measuring the voltage changes on the human skull at a given moment (e.g. 400 milliseconds) after a triggering event has occurred. The reported experiments show that in a construction such as 'Er l�chelte den Arbeiter an' (He smiled at the worker) containing the verb-particle combination anl�cheln the verb is first interpreted in the same way as an intransitive verb (l�cheln: to smile) and later we get an effect when the particle appears that can be interpreted as an integration or reindexing effect. This paper is the worst edited of the whole volume with references to non-existing examples and typos of all sorts. It should also have been read by a native speaker of English.

    As this overview of the individual contributions indicates, this collection does not present the last word on verb-particle constructions. The collection is interesting because of the wide diversity of points of view: not only are different syntactic theories represented but the articles also cover a wide range of issues, from details such as particle preposing to general overviews, and methods. I appreciated the general overview of the construction(s) in English and Dutch and the papers criticizing a body of previously held assumptions; the use of corpus data here is a welcome addition to the methodology. I found that theoretical approaches, where the construction is more a touch stone for syntactic theories than an object of intrinsic interest worked the least well in a collection such as this one: the author can describe his/her view on (part of) the specific construction but doesn't have the space to show how it fits in the overall grammar of a language. The result is that the proposals tend to look ad hoc. The psycholinguistic papers give a glimpse of the intricate correspondences between measurements, language processing and theories about storage of lexical material in the brain. New methods such as ERP will most likely shed light on this but it seems to me that they are still in the process of being calibrated precisely.

    I am looking forward to more discussion about what counts as a particle and what can modify the different kinds, specially in Dutch because the data and the discussion here seems to indicate that it is not clear what types of modification are allowed and what the importance of their distribution is. The ideas about construction-bound meaning as falling between idiomatic and traditional compositional meaning are also worth pursuing and it is heartening to see that corpus data are becoming a tool in syntactic research on par with the traditional introspection.

    As is usual these days, proofreading seems to have been left solely to the responsibility of the different authors and some of them are better than others. In Zeller's article, a couple of Dutch examples contain misspellings: (3) should be: dat Jan zijn moeder wil opbellen, (6) dat Jan de tapijten ti wil verkopen, (21) omdat hij mij ti probeert op te bellen. To leave the proofreading to the authors also leads to typos that they cannot be held responsible for, because they don't see the end product. For instance somewhere (p.175) we find the string ??(ws) in an example. One would assume that one of the four editors of this book could have caught this.

    REFERENCES

    M�ller, S. (2000) Complex predicates: Verbal complexes, resultative constructions, and particle verbs in German. CSLI.

    Wurmbrand, Susi (2000) The Structure(s) of particle verbs, Ms. University of Montreal.

    ABOUT THE REVIEWER

    Annie Zaenen is Principal Scientist at PARC and Consulting Professor at Stanford University. Her main interests are syntax and linguistic analysis for computational NLT.