Date: Tue, 5 Jul 2005 16:39:28 +0200 From: Eugenia Romanova Subject: Structuring Sense, Volume 2: The Normal Course of Events
AUTHOR: Borer, Hagit TITLE: Structuring Sense, Volume 2 SUBTITLE: The Normal Course of Events PUBLISHER: Oxford University Press YEAR: 2005
Eugenia Romanova, Department of Linguistics, University of Tromso, Norway
This is the second volume of the Exo-Skeletal Trilogy by Hagit Borer, The Normal Course of Events. The other two volumes are: volume I, In Name Only, published in the same year by the same publisher, and volume III, Taking Form, in preparation. The book includes three parts with a total of 11 chapters.
Part I. Setting course. Chapter 1. Exo-Skeletal explanations. A Recap. In this chapter the author summarizes the system she laid out in Volume I and introduces the theoretical equipment she is going to use in Volume II. The generalization underlying this work is that properties of functional items define all aspects of the computation, whereby by functional items the author refers to both functional vocabulary (including all grammatical formatives and affixation) and to functional structure. Grammatical formatives can be independent ones (f- morphs), for example 'the' and 'will', and (phonologically abstract) head features such as , for past tense. Listemes, constituting the encyclopedia, a list of all arbitrary pairings of sound and meaning, are devoid of any syntactic properties. Functional heads are open values with a category label. Here the author (henceforth, HB) introduces a mechanism of range assignment, crucial for the system. Open values are assigned range by a variety of means: directly, by the Specifier- Head agreement (by f-morphs in the Spec), and indirectly, by abstract head features:
1) [DP every.d [#P every.# [dog]]]
In (1), the listeme 'dog' is merged with the functional structure that includes a quantificational phrase #P and the determiner phrase DP, and the quantifier 'every' assigns range to the open values within both functional projections. The theoretical outline above is a gist of the exo-skeletal approach: all the aspects of computation are defined listeme-externally, by the properties of functional items with which listemes merge. Conversely, in endo-skeletal approaches the aspects of computation are defined listeme-internally, by the inherent properties of lexical items. The detailed comparison of the two systems is given in Chapter 2. One more issue has to be added. It is not always the case that the functional structure underlies the interpretation of lexical items; the latter can have inherent properties and then they are idioms under the present views. However, even the stored information is structured:
2) TROUSERS = [pi3+div]
In (2) pi(Greek letter)3(subscript) stands for the phonological index /trauser/, and div(subscript, meaning 'divided') stands for a piece of functional structure which must be assigned range by the plural inflection (cf. vol. I). The degree of specification in idioms allows a gradation. In (2) an open value (div) to be projected and an obligatory range assigner (plural marker) to this open value are specified. In 'cross a bridge', div is specified, but range to be assigned to it, is not. In 'kick the bucket', not only is the open value specified for 'the bucket' (div), but so is an obligatory range assigner to it ('the'), which forces the projection of #(subscript) and d(subscript, meaning 'determiner'). The importance of idioms becomes obvious in Part III.
Chapter 2. Why Events? The chapter shows how and where the Exo-Skeletal approach has advantages over the Endo-Skeletal approaches. The issues touched upon in the course of the discussion are: variable behavior verbs (cross-linguistically), aktionsart, and UTAH (Baker (1988)). How is the present approach superior? HB argues that the verb 'drop', for example, can be embedded under any syntactic structure, where the functional structure associated with the arguments determines their interpretation, rather than any information associated with 'drop'. The proponents of Endo-Skeletal approaches would have to admit that 'drop' either has several lexical entries (transitive and intransitive) or is associated with different semantic roles it has to assign (UTAH). However, according to the author, the verb itself cannot define what arguments will be projected, either external or internal -- originator or subject-of-quantity, as she labels argument roles. In the following chapters she elaborates on the functional projections that host the arguments, for not only is the external argument severed from the verb (Kratzer (1994)), but so is the internal argument.
Part II. The Projection of Arguments. Chapter 3. Structuring Telicity. In this chapter HB translates telicity from purely semantic representations into syntactic structures. The tools needed for this operation are: a) internal arguments with some specific properties (Verkuyl (1993)), b) the notion of homogeneity and the notions of divisibility and cumulativity (Krifka (1992, 1998)). From the previous chapter we already know that telicity is not an inherent property of a verb, but depends on many factors, the presence of the internal argument being one, it is logical to expect that this chapter will present us with a structure responsible for the telic interpretation of the event. So it does. The telicity-inducing piece of (functional) structure is the aspectual projection, ASPq(subscript). The projection contains an open value, #, where '#' stands for 'quantity'. In English-like languages in most cases the range to this open value is assigned by the internal argument DP which, as we learnt in the previous chapter, is a subject-of-quantity. As the internal argument is projected only in two configurations, transitive and unaccusative, the ASPq is absent from unergatives and non-quantity transitives, for the open range # cannot be assigned range here. In addition, ASPq is a site where accusative case gets assigned to the subject-of-quantity. In unaccusatives, 'the case is not available by assumption, the s-o-q DP must move to receive nominative case, presumably in [Spec, TP]' (Specifier of Tense Phrase).
Chapter 4. (A)structuring Atelicity. Here the famous Finnish paradigm comes into the picture: accusative objects yielding a telic interpretation, partitive objects atelic interpretation of events. Having shown the wrongness of approaches trying to explain the case alternation by weakness vs. strongness of DPs (de Hoop (1992), van Hout (1992, 1996), Borer (1994)), the author offers an alternative. Atelic structures do not have a quantity projection, ASPq, the one hosting a subject-of-quantity and assigning accusative to it. Instead, they project a shell functional projection FP, in the specifier of which partitive is checked. Impersonal constructions in Italian, Spanish and Hebrew support this hypothesis.
Chapter 5. Interpreting Telicity. The chapter is mostly devoted to a critique of the semantics literature on telicity. The author clearly shows that a) events cannot be mapped onto objects ('an atelic event does not have quantifiable sub-events that are distinct from the whole'); b) Verkuyl's system is contradictory (the feature [+ADD TO] is on the lexical verb, whereas the feature [+SQA] (Specified quantity of A) is on the determiner, a functional element; c) aspectual ambiguity of 'push' ('push the cart' vs 'push a button') cannot be lexically encoded and the verb is aspectually interpreted according to the world knowledge; d) the interpretation of the verbs unspecified for boundedness in Finnish (Kiparsky (1998)) depends on the case of their object; e) scalar approaches (Hay, Kennedy and Levin (1999) and Kennedy and Levin (2000)) also compare disfavorably to the present theory, due to the high role of context in deciding upon the telicity of 'lengthen' and 'empty' (so, telicity is not encoded in the lexical meaning of the adjectives giving rise to the corresponding verbs).
Chapter 6. Direct Range Assignment and The Slavic Paradigm. This chapter introduces the Slavic paradigm of structuring telicity. It is achieved via prefixation on the verb: prefixes are the phonological spell-out of head features in the specifier of ASPq that directly assigns range to # within ASPq. However, prefixes, being quantificational in nature (Filip (1999, 2000)) perform a double role. As there are no determiners in most Slavic languages and as an object DP must be in the specifier of ASPq, whenever the latter is projected, the open value d inside the nominal domain is assigned range by the prefix as well. Thus, non-quantity DPs are barred in [Spec, ASPq]. This and the following chapters will receive my special attention in the critical evaluation part of the review.
Chapter 7. Direct Range Assignment: Telicity without Verkuyl's Generalization. The reason for rejecting Verkuyl's generalization for Slavic is the presence of intransitive perfectives in the languages, such as semelfactives, reflexive verbs and the verbs with superlexical prefixes (cf. Romanova (2003), (2004), Svenonius (2004)). As prefixes are direct ranges assigners to [ASPq #], the presence of the DP in ASPq for this purpose is unnecessary in Slavic. However, telicity without Verkuyl's generalization is also achievable in English, with adverbs of quantification ('twice', 'once') or particles and prepositions.
Chapter 8. How Fine-Grained? In this chapter Hagit Borer argues against event decomposition. She points out that a) expressions like 'float under the bridge' are not decomposable, they can be modified by both 'in an hour' and 'for an hour'; b) not all telic events have an end-point; c) there are no causal relations in resultative constructions; d) 'babies' is ok as the subject of the 'simple' state in 'babies are asleep', but disallowed as the subject of the resultant state in resultative constructions (*'he sang babies asleep'). Therefore she rejects the Small Clause analysis of resultative constructions (cf. Dehe et al. (2002) and references cited there) in favor of a Complex-Predicate approach (cf. Zeller (2001)): 'hammer- flat' in 'hammer the metal flat'. A big part of the chapter is devoted to the speculation about predicate modification (by 'for x time'/'in x time') and anti-telicity effects (reflexive dative in Hebrew and nominalizer '- ing' in English).
Part III. Locatives and Event Structure. Chapter 9. The Existential Road: Unergatives and Transitives. This chapter, like the other chapters in part III, deals with the topmost projection discussed in Borer's system: EP (Event Phrase) (not to be confused with Travis's EP (Travis (1994, 2000)). EP is projected above TP (Tense Phrase) for a number of reasons (the author demonstrates the presence of the event argument even in statives; in addition T never binds the event argument). There are several possibilities for binding the event argument represented by the open value E(subscript, meaning 'event') merging as the head of EP: the range to this open value can be assigned by a) a referential DP, having the originator argument role; b) an expletive; c) an existentially closed DP through the specifier-head agreement; d) by a locative. However, the account for the paradigm in Hebrew and Italian in examples (1)-(5), p. 255, is not yet clear. The paradigm presents the following problems: a) Why are weak postverbal subjects possible with unaccusatives? b) Why aren't strong postverbal subjects possible? c) Why doesn't the same hold of unergatives? And the author herself adds a question: why locatives? First, it is a well-known fact that locatives have an existential force (Freeze (1992)). Second, bare DPs are licit when they can be located in space (Dobrovie-Sorin and Laca (1996)). Third, in French existential preverbal subjects can occur only in the presence of locative expressions. As for the weak postverbal arguments in Hebrew, they are subject to the 'slavified' behavior of the locative clitic, which directly assigns range to the E and indirectly, via specifier-head agreement, to a relevant open value inside the DP (this DP comes into agreement with the lower copy of V-loc -- the verb with the locative clitic, which must raise to EP).
Chapter 10. Slavification and Unaccusatives. From this chapter we learn more about quasi-functional items, like locative clitics in Hebrew, and how they are related to Slavic prefixes. Hebrew locative clitics are really like Slavic prefixes, in that they can merge with [ASPq #] and assign range to it. Then they move with the verb to E and assign range also to E. If they merge with T, they fail to assign range to [ASPq #], but still can do so to E. However, Hebrew locative clitics have one distinction from Slavic prefixes: the former are not quantificational in nature, so they cannot assign range to [DP #]. Therefore, a DP in [Spec, ASPq] does not have to be strong. In unergative structures there is no functional specifier that could hold the postverbal DP and keep it from agreement with locative (=existential operator), therefore weak postverbal subjects in unergatives are banned. The question remains about achievements with non-quantity object DPs ('discovered gold' and 'found rare coins'). The answers are: a) a covert locative is present in the structure; b) achievements are idioms. The author develops the latter idea some more in the chapter. We find out that not all achievements allow their objects to be non-quantity -- which gives further support to the idiom hypothesis. At the same time, achievements mostly behave like accomplishments: as was said above, their object DPs must be quantity and they can be progressive. All in all, taking into account examples in (53)-(56), pp. 331-332, the author concludes that there is no separate event type 'achievements'. They are just cases in which the projection of ASPq is obligatory (unlike in regular accomplishments-activities). This already makes achievements idiomatic: it is specified what open value is going to project ('finish'). In 'discover' or 'notice' which can take non-quantity DP objects, the specification is even higher: it is specified what is going to assign range to [ASPq #], and this is a covert locative.
Chapter 11. Forward Oh! Some concluding remarks. The concluding chapter of this work makes the following points: a) there is no inter-language variation that couldn't be attested within the same language, the variation being attributed to formal properties of grammatical formatives; b) the exo-skeletal approach overgenerates, but this overgeneration can be curtailed by the existence of idioms, 'cases in which some grammatical formative is specified to occur with a phonological index, and meaning is assigned to the complex constituent as a whole.' Idioms are a concession, the author admits, however 'it is to be hoped that future research will shed some light on them' and then 'an exo-skeletal approach can be completely successful.'
The book is written in a clear language and has a very clear layout. Each chapter begins with a 'recap' of the conclusions made and the discussions conducted in the previous chapters. The questions asked inspire curiosity and a detective spirit in the reader, and the author confidently leads him/her to the answers, step by step, with no (big) digressions from the main path. The author also dares ask questions unanswerable at the moment, and has the courage to offer directions towards the solutions to 'future researchers'. The review and the critique of the relevant literature is fairly exhaustive. Not a single major work on the topic has escaped HB's attention. It also creates a feeling of one scientific continuum -- the present analysis hasn't sprung up from nowhere, it has grown from the well fertilized soil.
If it were a work of fiction, I would say that the metaphor chosen by the author is so apt that now it can break free of the author's will and start developing on its own. It will certainly influence generations of researchers. Importantly, the empirical data and the analysis offered in the work in most cases do not contradict each other. However, at this point I would like to switch over to some critical remarks, especially concerning the Slavic languages. I had a number of questions while reading the book, both big and small, technical and empirical. Here, for the lack of space, I will restrict myself to three broad questions.
1. Slavic paradigm. Some mistaken assumptions have been made about Slavic. I suspect the reason is that all the prefixes are treated in the same way (although the distinction between them is cited from Svenonius (2003) on page 195). Not all of them are quantificational and even quantificational prefixes are not a homogeneous group. For instance, na-, widely discussed in the work, always requires a partitive genitive on its mass and bare plural objects, which already hints at the absence of #P inside the DP, thus making it impossible for the prefix to assign range to [DP #]. According to the author, such structures must be excluded in the presence of a quantificational prefix. In addition, HB notes: 'in the presence of telic structures, partitive case can never occur'. But the Russian partitive does, not only with na- verbs, but also with the verbs with purely perfectivizing prefixes or with prefixless perfectives ('kupit' chaju' buy.P tea.PART'). The system proposed by HB also predicts the impossibility of strong quantifiers with perfectives, but they can co-occur: 'On s'jel vse jabloki.' he ate.P all apples.ACC' (the reason for this claim is that the prefix na- has been analyzed in the greatest detail and it does disallow strong quantifiers with the objects of the verb). In pursuing the reasoning in favor of atelicity as the lack of syntactic quantity structure, the author repeats the following point but never really explains it: 'the primary and the secondary imperfectives are by necessity not a uniform semantic class... the secondary imperfective inflection has a progressive function', so it is a species of outer aspect in the sense of Verkuyl (1972), whereas the primary imperfectives are species of the inner aspect. The empirical data do not support this view. Both the primary imperfective and the secondary imperfective have a) a progressive function; b) a habitual function.
2. Achievements. The question about achievements is their idiomatic nature. Intuitively, idioms are language-specific. Achievements that allow non-quantity DP objects in English share similar properties cross-linguistically. English: 3). He found gold.
Finnish: 4). Hän löytyi kultaa. he found gold.PART.
Russian: 5). On nashël zoloto. he found.P gold.ACC
According to HB's analysis, the possibility of a non-quantity object in (3) is accounted for by a covert locative that assigns range to the open value inside EP. The question about Finnish is: how can a covert locative help the partitive marked object avoid aspect-related effects? The reading of the VP is still bounded. The partitive here stands for 'some gold'. The question about Russian: assume a covert locative is also present in Russian. How can the head features of the prefix avoid range assigning to [DP #], for here there is no quantity interpretation of 'gold' -- this DP is vague between a weak and a strong reading.
3. What comes first? The question that has been with me throughout the whole process of reading is 'what comes first?'. However, I think the author herself is asking it -- also in the book. How do we know that some functional head does not project? How do we tell the difference between its not being projected and its not being assigned range? If we know that the shell FP projects instead of ASPq, we know that the nominal will be partitive. On the other hand, the presence of existential pro indicates that the structure cannot be unaccusative (p. 120). Does the verbal structure determine the nominal structure or does the nominal structure determine the verbal structure? Does the projection of ASPq forces the merger of the subject-of-quantity or does the s-o-q forces ASPq to project (p. 228)? Do we first know that it is an achievement and the ASPq is obligatory or does the presence of ASPq identify the event type? etc.
At the end, I will just point out the typos I noticed in the foreign examples: p. 210, ex.(52) -- missing umlauts in the Finnish examples, p. 138 ex.(24, d) -- a missing letter in the Finnish example (äiti), p. 173 ex.(25, a) -- a plural instead of singular marker in the Polish example (must be 'artykul'), p. 188 footnote 4 -- there is no such a stem as 'krac' in Russian ('krik'?), p. 189, ex.(15, a) -- must be 'a' instead of 'o' in 'vyprashivat''; (15, b) -- the form given as imperfective does not exist; the form given as perfective is actually imperfective, p.195 ex. (24, b) -- the infinitival marker is just a softness sign, not 'i'.
Baker, Mark. 1988. Incorporation. Chicago: University of Chicago Press
Borer, Hagit. 1994. The projection of arguments. University of Massachusetts occasional papers in linguistics 17, (eds.) Elena Benedicto and Jeff Runner. Amherst: GLSA, University of Massachusetts
Borer, Hagit. 2005. Structuring Sense, volume I. In Name Only. Oxford University Press
Dehe, Nicole, Ray Jackendoff, Andrew McIntyre and Silke Urban (2002, eds.): Verb-Particle Constructions. Berlin: Mouton de Gruyter
de Hoop, Helen. 1992. Case Configuration and Noun Phrase Interpretation, (Groningen Dissertations in Linguistics). Groningen: University of Groningen (Published by Garland, New York, 1997)
Dobrovie-Sorin, Carmen and Brenda Laca. 1996. Generic Bare NPs, ms., University Paris 7 and University of Strasburg
Filip, Hana. 1996. Domain restrictions on lexical quantifiers, ms., University of Illinois at Urbana-Champaign
Filip, Hana. 1999. Aspect, situation type and nominal reference. Garland, New York
Filip, Hana. 2000. The Quantization Puzzle In Tenny, Carol, Pustejovsky, James (eds.) Events as Grammatical Objects, pp. 39-96. Stanford: CSLI publications
Freeze, Ray. 1992. Existentials and other locatives, Language 68, pp. 553-95
Hay, Jennifer, Christopher Kennedy and Beth Levin. 1999. Scalar structure underlies telicity in Degree Achievements, in Matthews, Tanya and Devon L. Strolovich (eds.). The Proceedings of the Ninth Conference on Semantics and Linguistic Theory, pp. 127-44. Santa Cruz: CLC Publications
Kennedy, Christopher and Beth Levin. 2000. Telicity corresponds to degree of change, paper presented at Michigan State University, 30 Nov. 2000
Kiparsky, Paul. 1998. Partitive case and aspect, in Greuder, William and Miriam Butt (eds.) The Projection of Arguments, pp. 265-307. Stanford: CSLI
Kratzer, Angelika. 1996. Severing the external argument from the verb, in (eds.) Johan Rooryck and Laurie Zaring Phrase Structure and the Lexicon, pp. 109-37. Dordrecht: Kluwer
Krifka, Manfred. 1992. Thematic relations as links between nominal reference and temporal constitution, in (eds.) Ivan A. Sag and Anna Szabolsci Lexical Matters, pp. 29-53. Stanford, CA: Center for the Study of Language and Information
Krifka, Manfred. 1998. The origins of telicity, in (ed.) Rothstein, Susan, Events and Grammar, pp. 197-235. Dordrecht: Kluwer
Romanova, Eugenia. 2003. Prefixes and secondary imperfectives in Russian. A presentation given at the Syntax Reading Group 27.03.2003, University of Tromso
Romanova, Eugenia. 2004. Superlexical versus Lexical Prefixes. Nordlyd 32, No 2, pp. 255-278
Svenonius, Peter. 2003. The morphosyntax of Slavic prefixes, paper presented at the East European Generative Summer School, Lublin, Poland
Svenonius, Peter. 2004. Slavic Prefixes and Morphology : An Introduction to the Nordlyd volume. Nordlyd 32, No 2, pp. 177-204
Travis, Lisa. 1994. Event phrase and a theory of functional categories, Proceedings of the 1994 Annual Conference of the Canadian Linguistic Association, (Toronto Working Papers in Linguistics), (ed.) P.Koskinen, pp. 559-70. Toronto
Travis, Lisa. 2000. The L-syntax/S-syntax boundary: evidence from Austronesian, Formal Issues in Austronesian Syntax, (eds.) Ileana Paul, Vivianne Phillips, and Lisa deMena Travis (Studies in Natural Language and Linguistic Theory 49), pp. 167-94. Dordrecht: Kluwer
van Hout, Angeliek. 1992. Linking and projection based on event structure, ms., Tilburg University
van Hout, Angeliek. 1996. Event Semantics of Verb Frame Alternations PhD dissertation. Tilburg: Tilburg University
Verkuyl, Henk. 1972. On the Compositional Nature of the Aspect. Dordrecht: Reidel
Verkuyl, Henk. 1993. A Theory of Aspectuality. The Interaction between Temporal and Atemporal Structure. Cambridge: Cambridge University Press
Zeller, Jochen. 2001. Particle Verbs and Local Domains. Amsterdam/Philadelphia: John Benjamins
ABOUT THE REVIEWER:
ABOUT THE REVIEWER
Eugenia Romanova is in her last year of PhD studies at the University of Tromso, Norway. In her dissertation she connects prefixation and argument projection in Russian, specifically focusing on a) quantificational effects the former has on the latter, which can be seen from case alternation on direct objects (accusative vs partitive genitive); b) versatile behavior of motion verbs, where directed motion verbs represent typical unaccusative structures and non-directed motion verbs typical unergative structures. The two groups of motion verbs help to show the structural complexity of prefix-verb-object conglomerates and the relationships between events and nominals, in tackling the question "What gets quantified?" She developed an interest in particle verbs in Germanic as a by-product of the main work, whereas the interest in the Finnish direct object case system was the trigger for starting the dissertation.