John Benjamins Publishing Company This is a contribution from Structure Preserved. Studies in syntax for Jan Koster. Edited by Jan-Wouter Zwart and Mark de Vries. This electronic file may not be altered in any way. Permission is granted by the publishers to post this file on the author s webpage and/or institutional server, to distribute it through mailing lists and to use it in any other way that serves the promotion of the publication. Tables of Contents, abstracts and guidelines are available at <http://www.benjamins.com/>www.benjamins.com
Depictives and the word orders of English and Dutch Herman Heringa University of Groningen Depictives are secondary predicates, usually adjectives that are predicated over a noun that functions as an argument of the main predicate. The depictive describes the state of affairs of that argument. Some examples in Dutch are given in (1): (1) a. dat Jan dronken thuiskwam that Jan drunk home.came that John arrived home drunk b. dat Marie het vlees rauw at that Marie the meat raw ate that Marie ate the meat raw For an extensive overview of the properties of depictives, I refer to the typological study by Schultze-Berndt & Himmelmann (2004: 77 78). Here, I will just sketch some of the most important characteristics and focus on the differences in word order between English and Dutch with respect to this construction. In order to explain these differences, I will build my analysis on Koster s (1999, 2000) theory on one of the main differences between English and Dutch: the OV/VO order. I will combine Koster s ideas with new insights on the differences between Germanic languages from Broekhuis (2008). The first word order difference between the two languages can be illustrated with the examples in (1). In Dutch, the depictive occurs in the middle field, to the left of the main verb, whereas in English, it occurs in the right periphery of the sentence, as the English translation shows. The second word order difference can be revealed if we combine two depictives in one sentence. As the examples in (1) show, depictives can be related both to the subject (1a) and to the object (1b) of the main predicate. A PP complement, on the other hand, cannot function as the controller for a depictive, as shown in (2) for English and in (3) for Dutch. Note that the English sentences in (2) are fine if the depictives are related to the subject. The Dutch depictives in (3) cannot be controlled by the subject, because Dutch subject-oriented depictives have to precede the PP complement.
158 Herman Heringa (2) a. *Ivan looked at Boris naked. b. *John came to Mary angry. (3) a. *Ivan keek naar Boris naakt. Ivan looked at Boris naked b. *Jan kwam naar Marie boos. Jan came to Marie angry. The literature suggests that the same is true for indirect objects (4a b), but the Dutch examples in (4c d) show that they can be related to a depictive at least in some cases (Jan-Wouter Zwart, p.c.): (4) a. *Mary gave John the book drunk. b. *Pete told Bill the story tired. (5) a. De kachel geeft mij naakt minder warmte dan gekleed. the heater gives me naked less warmth than dressed The heater gives me less warmth when I am naked than when I am dressed. b. De kou is mij naakt pas opgevallen. the cold is me naked only struck I didn t notice the cold until I was naked. Combining a subject-oriented depictive and an object-oriented depictive in one sentence reveals the second difference between English and Dutch. In English, subject-oriented depictives follow object-oriented depictives (6). In Dutch, it is exactly the other way around (7). In both of the grammatical options in (7b-c), the subjectoriented depictive precedes the object-oriented one. (6) a. She ate the meat raw naked. b. *She ate the meat naked raw. (7) a. *dat zij het vlees rauw naakt at that she the meat raw naked ate b. dat zij het vlees naakt rauw at that she the meat naked raw ate c. dat zij naakt het vlees rauw at that she naked the meat raw ate These word order differences provide an argument against an important type of analyses for depictives in the literature, namely adjunction of the depictive to a verbal projection. Analyses of this type focus on the fact that depictives are always optional. Notice that this distinguishes them from small clause complements, where
Depictives and the word orders of English and Dutch 159 the predicate is obligatorily present. This is illustrated with the examples in (8), from Richardson (2007: 113) and their Dutch equivalents in (9): (8) a. Ivan returned home (healthy). b. I consider Ivan *(intelligent). (9) a. Ivan kwam (gezond) thuis. Ivan came healthy home b. Ik vind Ivan *(intelligent). I find Ivan intelligent This optionality in combination with the fact that depictives always occur at the right side of the sentence in English, has lead several people to analyses in terms of right adjunction. In order to relate the depictive either to the subject or to the object, it would be adjoined to different layers of the (extended) verb phrase. A recent account in this direction comes from Richardson (2007). She proposes to adjoin subject-oriented depictives to the right of big VP, whereas object-oriented ones should be adjoined to little. Note that most of the data Richardson uses comes from Russian, which has the same order as English with respect to depictives. She gives the following example (Richardson 2007: 138): (10) Vona zjila mjaso syrym pjanoju. she ate meat raw drunk AP NP vona v pjanoju drunk she v VP V zjila eat VP NP mjaso meat AP syrym raw Whereas an analysis like this might seem to work for English and Russian, it does not explain the word order in Dutch. The only way out would be to assume that, whereas English and Russian use right adjunction, Dutch uses left adjunction in this case: (11) dat zij (het vlees) naakt (het vlees) rauw at that she the meat naked the meat raw ate
160 Herman Heringa CP C dat that NP i zij she TP T T NP k (het vlees) AP the meatnaakt naked t i v v NP k VP (het vlees) AP the meat rauw raw V at ate VP t k There seems to be no independent reason to assume that a parameter on the direction of adjunction between languages would exist, however. Therefore, I would like to propose an analysis in which the differences between English and Dutch follow from other principles. Also, note that an adjunction analysis as the one above needs to assume short object shift, moving the object to a position between little and big VP, optionally followed by scrambling the object to a higher position, only for Dutch. In the analysis I propose here, I follow Broekhuis (2008) in saying that both English and Dutch involve short object shift. In order to account for both the Dutch and the English/Russian order, I start out from Koster s (1999, 2000) theory to explain a more prominent word order difference between Dutch and English: the OV/VO order. He formulated his theory in terms of a pied-piping parameter and argued that it could explain a whole range of differences between the two languages, including Dutch V2 and the mirror symmetry of PP placement, which are absent in English. His main idea is as follows. Both English and Dutch have an underlying VO order. Also, both languages have to move their objects to SpecAccP (Koster splits the better known AgrOP in an AccP and a DatP)
Depictives and the word orders of English and Dutch 161 in order to check their case. The languages differ in that Dutch moves only the object itself, whereas English moves the complete VP containing the object. Therefore, the English surface order remains VO, whereas Dutch changes the underlying VO order to surface OV: (12) a. English: [ AccP [ VP V O][ Acc Acc t VP ]] b. Dutch: [ AccP O [ Acc Acc[ VP V t O ]]] This idea helps to explain the first word order difference: the occurrence of the depictive in the Dutch middlefield versus the English right periphery. In order to let this work, the depictives cannot be adjuncts, but have to be generated in functional projections higher up in the clause, like adverbs in the universal hierarchy proposed by Cinque (1999). In English, then, the whole VP moves to the left of the depictive, whereas in Dutch only the object moves and the verb is left behind: (13) a. English: [ AccP [ VP V O][ Acc Acc [ FP Dep [ F F t VP ]]]] b. Dutch: [ AccP O [ Acc Acc [ FP Dep [ F F [ VP V t O ]]]]] This leaves us with the second word order difference: the inverse order of subjectoriented and object-oriented depictives. This mirror word order of English and Dutch might suggest an analysis in terms of roll-up movement for English. In English, the VP first moves over the lower depictive and then the higher category, including the lower depictive, moves on over the higher depictive: rolling up (14a). As a result, the two depictives are swapped around. In Dutch, on the other hand, only the object moves over the lower depictive (14b) and optionally also over the higher one (14c), leaving the order of the depictives in both cases as it is. An analysis of depictives along these lines can be schematically represented as follows: (14) a. ate the meat raw naked b. naakt het vlees rauw at naked the meat raw ate c. het vlees naakt rauw at the meat naked raw ate
162 Herman Heringa naked naakt raw ate VP the meat rauw at NP het vlees An analysis like this clearly explains the word order difference between English and Dutch. However, both the adjunct analysis and the analysis sketched here so far need something extra. In both analyses, depictives are basically analysed as adverbial elements, either generated as adjuncts to verbal projections, or in functional projections higher in the clause. However, there is an essential difference between depicitives and adverbials. As stated before, depictives are related to a participant of the main predicate. Adverbials, on the other hand, are related to events. The difference, in English morphologically marked by the -ly suffix for adverbials, is illustrated in (8), slightly adapted from Rothstein (2006: 210) (see also Geuder 2002): (15) a. John left the room angry, # although he was actually faking his displeasure. b. John left the room angrily, although he was actually faking his displeasure. As (15a) shows, a depictive describes a state of its controller, just like a nominal predicate describes a state of its subject. In contrast, the adverbial in (15b) only modifies the predicate. It does not describe a state of the agent, but the way in which the agent carries out the event. A closer relation between the controller and the depictive is also confirmed by agreement between the two in other languages. The French example below is from Legendre (1997: 52): (16) Marie donne ses conférences assise. Marie gives her talks sitting:fem.sg Marie gives her talks sitting down. Phenomena like these suggest that the relation between the depictive and its antedecent should be represented as a predication relation. Thus, a sentence involving a depictive contains two predicative elements: the main predicate and the depictive. The time of the state described by the depictive and that of the eventuality expressed by the main predicate coincide. In (15a), John was angry at the time that he left the room. The idea of a predication relation between a depictive and its controller can be accounted for by
Depictives and the word orders of English and Dutch 163 analysing the depictive as a small clause, consisting of a PRO subject, controlled by the antecedent, plus the adjective itself (see Chomsky 1981; Hornstein & Lightfoot 1987, and Legendre 1997; among others). This might seem a straightforward adaptation, but it also has consequences for the rest of the analysis. Since the PRO subject in such an analysis has to be linked to its antecedent, it is an obligatorily controlled PRO. This means that we would expect the antecedent at hand to c-command PRO inside the depictive. If we want to maintain that English moves the complete VP, it is not clear how the object could c-command an object-oriented depictive. Since the object remains embedded in the VP in this case, it cannot c-command out of it. This appears to be a problem at first sight, but I will show that it vanishes as soon as we take into account Broekhuis (2008) independently developed ideas concerning short object shift. According to Broekhuis, the difference between English and Dutch is not in moving the entire VP versus the object only. Instead, he argues, on the basis of binding phenomena and NPI licensing, that both languages involve short object shift to a position between the big VP and the little. After that, English moves the verb to little v, whereas Dutch does not. If this is indeed the case, the object is no longer embedded in the VP after short object shift, which means that it is in a position from where it is able to c-command an object-oriented depictive. Implementing Broekhuis (2008) idea in the analysis I proposed so far results in the following representation for object-oriented depictives in English and Dutch: she v zij v v v PRO raw VP PRO rauw VP V ate the meat V at het vlees The object-oriented depictives are small clauses with a PRO subject, generated in a functional projection of the big VP. First, the object moves over the depictive small clause in both languages and c-commands the PRO inside. After that, the verb moves to little v in English, but not in Dutch. Thus, the surface order is derived. The question now remains how we can use an analysis involving short object shift in order to explain the inverse word order of the two types of depictives in English. Of course, the subject moves to SpecTP in the end. From that position, it will be able to c-command the PRO in the subject-oriented depictive small clause. Now, we combine
164 Herman Heringa Broekhuis idea with the roll-up movement suggested before. The question then arises which higher category can be moved over the higher depictive after applying both short object shift and v-to-v movement and where this complex should go. The most obvious option is to move little to a position somewhere between the functional projection containing the subject-oriented depictive and T. Note that little in the representation for English below is derived by short object shift and v-to-v movement as illustrated above. In Dutch, the object optionally scrambles further to the specifier of a functional projection between the higher depictive and TP: TP TP T T T T i she v ate the meat raw PRO naked t i PRO naakt zij v het vlees rauw at Note that in English the subject, which is generated in Spec, has to be extracted from after the latter has moved. This violates the raising principle (Wexler & Culicover 1980), which states that a raised node is frozen, meaning that nodes below it cannot be extracted. However, the same violation was also present in Koster s original proposal. He took the subject to be generated within the VP and after moving the VP extracted the subject to SpecAgrS (see Koster 1999: 35). Also, people (eg. Di Scullio & Isac 2008) have argued that the raising principle is not that rigid. Therefore, I conclude that the differences in word order between English and Dutch with respect to depictives can be derived as described above and leave the raising principle violation for future research. In sum, I proposed an analysis for depictives in English and Dutch that can explain the differences in word order in this respect between the two languages, based on ideas from Koster (1999, 2000) and Broekhuis (2008) on the word orders in the Germanic languages. As a basis, I assume an underlying VO order for both languages. In order to account for the predicate relation between depictives and their controllers, I analyse depictives as small clauses with a PRO subject and the depictive as the predicate. These small clauses are positioned in functional projections above VP for object-oriented depictives and above for the subject-oriented ones. In order to c-command, and thus control, PRO inside the small clause, the object in both languages moves out of the VP over the lower depictive: short object shift. In English, unlike in Dutch, this
Depictives and the word orders of English and Dutch 165 shift is followed by V-to-v movement (over the object), explaining both the VO/OV difference between the two languages and the fact that English depictives surface in the right periphery of the sentence, whereas Dutch depictives appear in the middle field, to the left of the verb. After this, English moves the whole little, containing the lower depictive, over the higher depictive, explaining the mirror order of subject-oriented and object-oriented depictives between the surface and the underlying structure in this language. Dutch, on the other hand, only may scramble the object further over the higher depictive, keeping the underlying order of the two types of depictives intact, but swapping around the order of the object and the higher depictive. The optionality of this last step in Dutch explains that two word orders are possible in this language. Thus, the differences in word orders between English and Dutch with respect to depictives are derived from the same underlying order. References Broekhuis, Hans. 2008. Derivations and evaluations: Object Shift in the Germanic languages. Berlin: Mouton de Gruyter. Chomsky, Noam. 1981. Lectures on government and binding. Dordrecht: Foris. Cinque, Guglielmo. 1999. Adverbs and functional heads. Oxford: OUP. Di Scullio, Anna Maria & Isac, Daniela. 2008. Movement chains at the interfaces. Canadian Journal of Linguistics 53: 181 217. Geuder, Wilhelm. 2002. Oriented adverbs: Issues in the lexical semantics of event adverbs. Ph.D. dissertation, Universität Tübingen. Hornstein, Norbert & Lightfoot, David. 1987. Predication and PRO. Language 63: 23 52. Koster, Jan. 1999. The word orders of English and Dutch: Collective vs. individual checking. Groninger Arbeiten zur germanistischen Linguistik 43: 1 42. Koster, Jan. 2000. Pied piping and the word orders of English and Dutch. Proceedings of NELS 30: 415 426. Legendre, Geraldine. 1997. Secondary predication and functional projections in French. Natural Language and Linguistic Theory 15: 43 87. Richardson, Kylie. 2007. Case and aspect in Slavic. Oxford: OUP. Rothstein, Susan. 2006. Secondary predication. In Martin Everaert, Henk van Riemsdijk, Rob Goedemans & Bart Hollebrandse (eds.), The Blackwell companion to syntax, Vol. 4, 209 233. Oxford: Blackwell. Schultze-Berndt, Eva & Himmelmann, Nikolaus. 2004. Depictive secondary predicates in crosslinguistic perspective. Linguistic Typology 8: 59 131. Wexler, Ken & Culicover, Peter W. 1980. Formal principles of language acquisition. Cambridge MA: The MIT Press.