* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Päike sulatas suure jääpurika ära.
Polish grammar wikipedia , lookup
Old Irish grammar wikipedia , lookup
American Sign Language grammar wikipedia , lookup
Swedish grammar wikipedia , lookup
English clause syntax wikipedia , lookup
Germanic weak verb wikipedia , lookup
Kannada grammar wikipedia , lookup
Germanic strong verb wikipedia , lookup
Portuguese grammar wikipedia , lookup
Ancient Greek grammar wikipedia , lookup
Macedonian grammar wikipedia , lookup
Old English grammar wikipedia , lookup
Navajo grammar wikipedia , lookup
Russian grammar wikipedia , lookup
Sotho verbs wikipedia , lookup
Chinese grammar wikipedia , lookup
Japanese grammar wikipedia , lookup
Turkish grammar wikipedia , lookup
Yiddish grammar wikipedia , lookup
Icelandic grammar wikipedia , lookup
Serbo-Croatian grammar wikipedia , lookup
Modern Hebrew grammar wikipedia , lookup
Kagoshima verb conjugations wikipedia , lookup
Latin syntax wikipedia , lookup
Georgian grammar wikipedia , lookup
Hungarian verbs wikipedia , lookup
Lexical semantics wikipedia , lookup
Delimitedness, telicity, direct objects and obliques Anne Tamm 0. Introduction Many mapping or linking theories have tried to capture the relation between aspect and syntactic objects. Tenny (1994) and Van Hout (2000), for instance, claim that universal principles of mapping between the lexicon and syntactic argument structure are governed by aspectual properties. More specifically, they posit a link between the presence of a direct object and the expression of certain aspectual properties, such as ‘delimitedness’, ‘measuring out of events’, or ‘telicity’. However, the following discussion provides examples of Estonian sentences that suggest a revision of this strong hypothesis. First, there are sentences without any direct object that, contrary to expectations, are compatible with Tenny’s tests of delimitedness and measuring out and with Van Hout’s tests for telicity. Second, the relations between delimitedness, object cases and verbs with particles present a wider array of data than Tenny’s theory can capture. This article will demonstrate that some problems in Van Hout (2000) are solved by clarifying the partial overlaps in the terminology used in current literature on aspect. Also, the article shows that aspectual properties govern Estonian overt syntax even more than Tenny (1994) envisaged. However, some predicates show that the link between delimitedness, telicity, and measuring out on the one hand, and direct objects on the other, is a tenuous one. Furthermore, the particle verb data lead to an account in terms of two types of measuring out scales: volitional, and object or path related scales. 1. Thematic structure and syntactic argument structure Several syntacticians have hypothesized that aspectual properties govern the mapping or linking between thematic structure and syntactic argument structure. The occurrence of a direct object in a sentence is related to the expression of aspectual properties such as ‘delimitedness’, the ‘measuring out of events’ (Tenny 1994), or ‘telicity’ (Van Hout 2000). This section revises the claims put forward by these authors. 1.1. Aspectual properties govern syntactic argument structure One of the strongest hypotheses about the relation between aspect, thematic structure, and argument structure is put forward by Tenny (1994:2). Her Aspectual Interface Hypothesis (AIH) is formulated as follows: The universal principles of mapping between thematic structure and syntactic argument structure are governed by aspectual properties. Constraints on the aspectual properties associated with direct internal arguments, indirect internal arguments, and external arguments in syntactic structure constrains the kinds of event participants that occupy these positions. Only the aspectual part of thematic structure is visible to the universal linking principles. (Tenny 1994:2) 1 This hypothesis assumes a link between, on the one hand, certain aspectual properties that are termed ‘measuring out’ and ‘delimitedness’ and, on the other hand, arguments, such as direct objects. In order to see how the constraints on aspectual properties work, and how the aspectual properties are linked to argument types, it is important to study some of the constraints in more detail. Since direct objects are an essential part of the following discussion, the passage below presents Tenny’s formulation of the relation between ‘measuring out’ and direct objects: (i) (ii) (iii) Measuring out Constraint on Direct Internal Arguments The direct internal argument of a simple verb is constrained so that it undergoes no necessary internal motion or change, unless it is motion or change which ‘measures out the event’ over time (where ‘measuring out’ entails that the direct argument plays a particular role in delimiting the event). Direct internal arguments are the only overt arguments which can ‘measure out the event’. There can be no more than one measuring out for any event described by a verb. (Tenny 1994:11) This constraint states that the object can undergo internal change and motion only if the change and motion are connected to ‘measuring out’ the event. Only objects can perform ‘measuring out’; a verb can describe only one ‘measuring out’ of an event. Tenny’s Measuring Out Constraint does not say anything about verbs without direct internal arguments. Van Hout (2000), on the other hand, studies the event type properties of both transitive and intranstitive verbs. She constrains the aspect-syntax mapping more tightly. More specifically, “event type properties play a role in the lexicon-syntax interface where verbs get mapped onto syntactic verb frames (Van Hout 2000:239).” Van Hout locates the information about event types in the lexicon. As she writes, “the lexical properties that the mapping system must ‘see’ in a verb [do] not only include the number of arguments, but also its event type” (Van Hout 2000:239). A certain aspectual interpretation of ‘telicity’ is impossible without a direct object: “Telicity…requires checking of an aspectual feature by the direct object” (Van Hout 2000:239). Van Hout, therefore, assumes a correspondence between a verb frame and the verb’s event type properties. For instance, telicity is only possible if the verb frame contains an object. On the other hand, objectless sentences are necessarily atelic in Van Hout’s theory. She compares the transitive and intransitive verb frame alternations of Dutch verbs, such as drinken ‘to drink’ and schrijven ‘to write’. She concludes that “one finds that only the transitive frame can yield telicity for these verbs; a verb frame without an object or with an oblique object cannot” (Van Hout 2000:243). In sum, both theories see objects to be the condition for ‘telicity’, ‘delimitedness’, or ‘measuring out’. Obliques, on the other hand, are not considered to be related to the abovementioned aspectual properties (Van Hout 2000), or not considered (Tenny 1994). However, the way the relations between syntax and lexicon are handled differs in these two approaches. While Tenny exploits the notion of aspectual thematic roles, Van Hout has objections to making reference to those roles: Standard models of the lexicon-syntax interface are defined in terms of the semantics of a verb’s arguments which are lexically listed for each verb in terms of a set of θ-roles, as variables in lexical-conceptual structures or as argument structures. In these models there is no room for event-semantic notions such as 2 telicity, and so… there is no way they can incorporate the telicity/object association. (Van Hout 2000:239) Tenny has the same basic problem with the thematic roles, which do not capture those facets of lexical information that are relevant for syntax. However, differently from Van Hout, she introduces special aspectual roles in order to mediate between the lexicon and syntax. One of these roles is termed as the MEASURE role, and it is “assigned to an argument of the verb, which (in the event as described by the verb) either undergoes some internal change or motion, along a single parameter; or provides a scale or parameter without undergoing change or motion; that measures out and defines the temporal extent of the event” (Tenny 1994:95). This role determines the possibility of measuring out of the event, establishing the link between the temporal progress and the change in the entities that are processed. A verb may assign the MEASURE role to its object, and a noun phrase argument may bear it in a sentence. In Tenny’s accout, MEASURE can also be composed of the other two aspectual roles PATH and TERMINUS, and PATH is seen as a defective MEASURE role. Arguments with aspectual roles are aspectually licensed, that is, licensed by the temporal structure of measuring out (Tenny 1994:96). In sum, while Tenny sees a way to associate aspect and objects by making reference to lexicon-related aspectual roles, Van Hout rejects the idea of lexicon-based roles and establishes an interface of event types. As a consequence of this distinction, the thematic role composition of the objects that are considered in these theories is different. Tenny considers themes, patients and paths that change, or that are processed in the course of the event; Van Hout does not constrain the the thematic role composition of the verbs that participate in verb frame alternations. 1.2. Oppositions of telicity and delimitedness have similar diagnostics Apart from the exact details of the aspectual interface, Tenny and Van Hout seem to capture similar generalizations. However, the terminology they use reveals further conceptual differences in their work. Tenny’s ‘measuring out’ refers to the role played by the argument in marking the temporal terminus of the event (Tenny 1994:10-11). For example, the role of apple measures out the event in the following sentence (1): (1) Thomas ate an apple up. The complete consumption of the apple marks the end of the eating event. Measuring out is a composite concept that in its turn comprises two concepts: a measuring scale associated with its argument and a temporal bound or delimitedness (Tenny 1994:15). Whether a sentence is ‘delimited’ is tested by the sentence’s compatibility with the temporal durative and time frame adverbial tests. For example, the adverbials for a day/in a day or for an hour/in an hour test the presence of delimitedness in the following sentences, (2) and (3): 3 (2) Bill pushed the cart (for an hour/*in an hour). The sentence (2), incompatible with the time frame adverbial and compatible with the durative adverbial, is non-delimited and has a non-measuring argument, the object cart. The sentence (3), compatible with the time frame adverbial and incompatible with the durative adverbial, is delimited; it has a measuring argument, the object house. (3) Mary built a house (*for a day/in a day). Succintly, if the sentence is compatible with the time frame adverbial test (for instance, in an hour/in a day), the sentence is classified as delimited. The verb’s argument is considered to measure out the event if it undergoes an internal motion or change that corresponds to the temporal progression of the event. Otherwise, the sentence is classified as non-delimited, and its argument is not considered to measure out the event. Further tests for measuring out are grouped according to the three types of predicates that have arguments that measure out: incremental theme verbs (build a house, eat an apple), change-of state verbs (crack the glass), and pathobject ‘route verbs’ (climb the ladder, walk the Appalachian Trail). However, the object of push the cart to the store would not be a measuring argument, since it undergoes no internal change or motion. A concept that is similar to delimitedness is studied by Van Hout (2000) in relation to the event structural property ‘telicity’. Tenny, on the contrary, consciously avoids this term “because it implies a focus on the goal-oriented nature of certain events. It has engendered some confusion among readers, since the ‘goal’ of a delimited event can be arrival at a certain state as well as a location” (Tenny 1994:125). Van Hout still considers the culmination-oriented nature of some events an essential property in the lexicon-syntax mapping. This intuition has led the author to incorporate this aspectual concept as a central one, defined as follows: “TELICITY refers to the internal temporal make-up of an eventuality and describes whether it is homogeneous and cumulative (atelic) or not (telic), or, in more intuitive terms, whether or not it has a natural moment at which the event culminates” (Van Hout 2000:241). This definition attempts to unite concepts of sentence aspectual properties: homogeneity, culmination, and telicity. Van Hout uses practically the same diagnostics for establishing telicity as Tenny does for establishing delimitedness. A sentence is considered to be telic if it is compatible with the time frame adverbials and incompatible with the durative adverbials; it is atelic if it is compatible with the durative adverbials and incompatible with the time frame adverbials. Sentences with a definite concrete noun object and a simple (incremental-theme) verb (4) are telic by these criteria (Van Hout 2000:242): (4) Judy heeft *urenlang/in 5 minuten haar kopje koffie gedronken. J. has *hours-long/in 5 minutes her cup of coffee drunk ‘Judy drank her cup of coffee *for hours/in five minutes.’ Sentences with a definite concerete noun object and a particled incremental-theme verb (5) are also telic by Van Hout’s diagnostics (Van Hout 2000:244): 4 (5) Judy heeft *urenlang/in 5 minuten haar kopje koffie opgedronken. J. has *hours-long/in 5 minutes her cup of coffee up- drunk ‘Judy drank her cup of coffee up *for hours/in five minutes.’ Sentences without an object and with a simple (incremental-theme) verb (6) are atelic by these criteria (Van Hout 2000:243): (6) Judy heeft urenlang/*in 5 minuten gedronken. J. has hours-long/*in 5 minutes drunk ‘Judy drank for hours/*in five minutes.’ Also, sentences with an oblique argument and a particled incremental-theme verb (7) are atelic by these tests (Van Hout 2000:243): (7) Judy heeft urenlang/*in 5 minuten van haar kopje koffie gedronken J. has hours-long/*in 5 minutes of her cup coffee drunk. ‘Judy drank “of” her cup of coffee for hours/*in five minutes.’ In sum, the notions delimitedness and telicity are tested by identical tests. Tenny’s measuring out does not play a role in Van Hout’s account. Instead of the mapping from aspectual roles as in Tenny’s account, Van Hout builds on a typology of eventualities where the concept of telicity plays the most important role. Therefore, more data will be accounted for by aspectual properties by Van Hout. As an example of an important difference, Van Hout’s theory requires the existence of the object in push the cart to the store since the described event is telic. The object cart would not be considered by Tenny, since it is not a measuring argument according to her constraints – the object does not undergo any internal change or motion. 2. Arguments for a different approach Despite the fact that the described hypotheses explain much of the cross-linguistic data, there are reasons to argue for a more finegrained approach to the syntax-lexicon interface. The following discussion provides examples of Estonian sentences that suggest a revision of the strong hypotheses of Tenny and Van Hout. The first section (2.1.) determines the exact nature of the terminology that Van Hout uses and its relation to some Estonian verb frame alternations. The second issue, broached in section 2.2., is the existence of sentences without any direct objects. Contrary to expectations, those sentences display compatibility with Tenny’s tests of delimitedness and measuring out as well as with Van Hout’s diagnostics for telicity. The third, rather complicated issue pertains to the relations between telicity, delimitedness, Estonian object cases, and verbs with particles. That issue is discussed in section 2.3. In sum, this section demonstrates that the Estonian language presents a wider array of data than Tenny’s and Van Hout’s theories explain in their formulations from (1994) resp (2000). 5 2.1. Van Hout’s ‘atelic’ is ‘homogeneous’ and does not determine verb frame alternations This subsection shows that some parts of Van Hout’s terminology are problematic. The aim is to find out to what extent the terms used by Tenny and Van Hout differ from each other and, more importantly, how they relate to the situation that we find in Estonian sentences. For this end, the subsection takes a closer look at transitive sentences and the Estonian object case. An analysis of Estonian data shows that in several cases the (a)telicity referred to by Van Hout does not correlate with the alternation of Estonian verb frames. Consider the following examples: (8) Mari ehitas masinat tundide Mari.nom build.3sgpast machine.part hour.plgen ‘Mary was building a machine for hours.’ viisi. wise (9) Mari ehitas masina ühe Mari.nom build.3sgpast machine.gen one.gen ‘Mary built a machine in an hour.’ tunniga. hour.comit The difference between (8) and (9) emerges only in the object’s case alternation, not in verb frame alternation. There is no verb frame alternation, even if the diagnostics classify sentence (8) as atelic and sentence (9) as telic. This fact makes it possible that the definition of (a)telicity as used in Van Hout might contain two definitions and relate to two conceptually different phenomena. Further discussion demonstrates that on the one hand, Van Hout’s definition contains the concept of (non)homogeneity (cf. Tenny (1994:25), Van Hout (2000:243), Kiparsky (1998)). On the other hand, it contains the concept of (the lack of) culmination or ‘telos’ orientedness – that is, the intuitive, situation (a)telicity ‘proper’ (cf. Comrie (1976:44), Depraetere (1995), Dahl (1985:90-95)). The first part of Van Hout’s definition employs thus the notion of homogeneity: “TELICITY refers to the internal temporal make-up of an eventuality and describes whether it is homogeneous and cumulative (atelic) or not (telic)” (Van Hout 2000:241). The notion of homogeneous reference is often used in frameworks that study the parallels between spatial properties of entities and the temporal properties of eventualities. The reference is homogeneous if it is both cumulative and distributive. This subsection demonstrates that the data classified as atelic are more correctly analysed as homogeneous and not necessarily ‘atelic proper’ (cf. Comrie (1976), Depraetere (1995), Dahl (1985)). The homogeneous reference, or homogeneity, of the sentence (8) with the partitive object can be proved and tested by its cumulativity (a) and distributivity (b) as follows: (a) the sum of two events, Mary is constructing a machine and Mary is constructing a machine is still an event of Mary is constructing a machine; (b) a part of the event of Mary is constructing a machine is still an event of Mary is constructing a machine. Therefore, the sentence in (8) is homogeneous, since it has cumulative and distributive reference. Nonhomogeneity is termed thus ‘atelic’ in Van Hout’s terminology. The nonhomogeneity of the sentence (9) with the genitive object can be tested as follows: (a) the sum of two events, Mary constructed a machine and Mary constructed a machine, is an event of Mary constructed TWO machines; 6 (b) a part of the event of Mary constructed a machine is not an event of Mary constructed a machine, but something else. This can be, for instance, Mary was constructing a machine, or Mary constructed a part of a machine. The reference type is here neither cumulative nor distributive; therefore, the sentence in (9) is nonhomogeneous, which is ‘telic’ in Van Hout’s terminology. This paper will use further the term ‘homogeneity’ to avoid ambiguity. Now, according to Van Hout’s predictions about the verb frame alternations, a contrast in homogeneity would be expected to yield different argument frames for a verb. But in the case at hand, the contrast in homogeneity does not necessarily yield different verb frames; this contrast yields object case alternation in Estonian. While the first part of the definition does not always predict the existence of verb frame alternations, the second one does. More specifically, Van Hout’s account of the correlation between (a)telicity and the verb frame alternations explains Estonian verb frame alternations when the second, more intuitive part of her definition of telicity is applied. As Van Hout states, “telicity refers to … whether or not [an eventuality] has a natural moment at which the event culminates” (Van Hout 2000:241). This formulation approximates the view that authors such as Comrie (1976:44), Depraetere (1995), or Dahl (1985:90) hold about the phenomenon. In their view, the telicity and atelicity of situations are different with regard to the situations’ internal structure. A situation that has a built-in terminal end point or that can reach such an end point is called telic. A situation that has no such terminal point or a possibility to reach one is called atelic. Comrie (1976:44) describes the contrast between the atelic sing, as in John is singing, and the telic make a chair, as in John is making a chair. In John is making a chair, in Comrie’s words, “there comes eventually a point at which John completes the action of making a chair, the chair is ready… until this point is reached, the situation described by make a chair cannot come to an end, but can only be broken off part way through” (Comrie 1976:44). On the other hand, in John is singing, “John can stop singing at any point, and it will still be true that he has sung, even if he has not completed his song or songs he set out to sing” (Comrie 1976:44). Comrie proposes the following test to classify the situation as atelic versus telic: “if a sentence referring to this situation in a form with imperfective meaning (such as the English Progressive) implies the sentence referring to the same situation in a form with perfect meaning (such as the English Perfect), then the situation is atelic; otherwise it is telic” (Comrie 1976:44-45). Therefore, John is singing is atelic, since one can deduce John has sung from it; John is making a chair is telic, since one cannot deduce John has made a chair from it. This kind of telicity will be referred to further as ‘culmination-orientedness’. Applying the second part of the definition of telicity, the sentences with an object would be predicted to be culmination-oriented, and sentences with an oblique would be not culmination-oriented. This is indeed the prediction for the verb frames of the verb lõhkuma ‘to break’. Consider the following data in sentences (10) – (13): (10) Töömees lõhkus aiavärava. Repairman.(nom) break.3sgpast gate.gen ‘The repairman broke/destroyed the gate.’ (11) Töömees lõhkus aiavärava Repairman.(nom) break.3sgpast gate.gen ‘The repairman broke/took down the gate.’ (maha). down 7 The transitive sentences (10) and (11) depict an eventuality where there is a natural moment at which the eventuality culminates. This is the moment at which the gate is broken (10) or the moment at which a gate is taken or broken down (11). (12) Töömees lõhkus aiaväravat (maha). Repairman.(nom) break.3sgpast gate.part down ‘The repairman was breaking/taking down the gate.’ The sentence with a partitive object (12) depicts a culmination-oriented eventuality, and the sentence describes the eventuality as not having reached its natural culmination moment of the gate being taken or broken down. By Comrie’s test, the sentence (12) describes a telic situation, since the imperfective (12) does not imply the sentence referring to the same situation in a form with perfect meaning, that is, sentence (11). Example (13), on the contrary, describes an atelic situation, since it implies the sentence referring to the same situation in a form with perfect meaning1. (13) Töömees lõhkus aiavärava kallal. Repairman.(nom) break.3sgpast gate.gen at ‘The repairman was breaking ‘at’ the gate (almost achieved breaking the gate).’ The intransitive sentence (13) with an oblique does not depict an eventuality where there is a natural moment at which it could possibly culminate – there is no culmination implied in carrying out successive acts of almost breaking the gate or breaking ‘at’ the gate. The contrast between (10)/(11) and (13) is also that of homogeneity; also, this would indeed follow from Van Hout’s account. The tests for sentences (10) and (11) show that these sentences are nonhomogeneous: The repairman broke (down) a gate and the repairman broke (down) a gate is the repairman broke (down) two gates. A part of the denotation of the repairman broke (down) a gate is not the repairman broke (down) a gate. Therefore, the sentences (10) and (11) are nonhomogeneous. The sentence (13) is not: The repairman was breaking ‘at’ the gate and The repairman was breaking ‘at’ the gate is The repairman was breaking ‘at’ the gate. A part of the denotation of The repairman was breaking ‘at’ the gate is still The repairman was breaking ‘at’ the gate. Therefore, the sentence (13) is homogeneous. However, the contrast between (12) and (13) is not that of homogeneity; it is a contrast of culmination-orientedness. The sentences (13) and (12) are equally homogeneous: the repairman was taking/breaking down a gate and the repairman was taking/breaking down a gate is the repairman was taking/breaking down a gate. A part of the denotation of the repairman was taking/breaking down a gate is still the repairman was taking/breaking down a gate. In conclusion, the concept that Van Hout terms atelicity is more precisely covered by the term ‘homogeneity’. Figure 1 below classifies the discussed sentences according to their position 1 Sentence (13) does not have a good equivalent in English. Native English-speakers might recall a similar situation from the film Shining. Jack Nicholson was trying to enter the house, but the door was locked; so, he started to break the door with an axe. Before he managed to break it, the situation can be described with the sentence (13): he was breaking ‘at’ the door with his axe for long seconds. Many thanks to Jon Readey for recalling this situation. 8 in the two binary oppositions of culmination-orientedness and homogeneity. Also, Figure 1 indicates the argument type (O – object; Obl – oblique) and object case type (Ot – total object cases; Op – partitive object). nonhomogeneous homogeneous culmination-oriented not culmination-oriented (9) Ot (8) Op (13) Obl (10) Ot (11) Ot (12) Op Figure 1. (A)telicity in terms of homogeneity and culmination-orientedness. The figure demonstrates that the verb frame alternations for which Van Hout was seeking an aspectual explanation do not correspond strictly to contrasts in homogeneity. These alternations can correspond to contrasts in culmination-orientedness. The object case alternation in Estonian, on the other hand, is a contrast of homogeneity. Therefore, what Tenny terms delimited is closer to what Van Hout discusses under the examples that she terms telic (nonhomogeneous). However, Van Hout describes delimitedness without measuring out or path, an option raised, but not exhaustively elaborated in Tenny (1994:15-16). Delimitedness without measuring out will be one point discussed in the further subsections under (2.2.). Particled verbs with obliques and objects will be studied further in subsection (2.3.). The immediately following discussion will clarify the notion of objects in Tenny’s and Van Hout’s theory. 2.2. Sentences without direct object The most challenging issue is the classification of intransitive sentences that are compatible with Tenny’s tests of delimitedness and Van Hout’s diagnostics for telicity. Sentences without direct objects will be discussed in two subsections. The motivation for dividing the discussion along this distinction is the different status of those sentences for the above reviewed theories. The theoretical framework of those accounts regards some surface subjects as underlying objects (cf. Levin and Rappaport 1995). For reasons of structural Case assignment, phrases that are originally in the object’s configurational position are moved to the subject’s configurational position. Unaccusative sentences are those intransitive sentences that have such derived subjects. Predicates that occur in unaccusative sentences are, for instance, fall, freeze, exist, disappear, roll, glow, cease, or survive. The subjects of the so-called unergative intransitive sentences, on the contrary, have subject phrases that never occupy the position of an object in any stage of derivation. Predicates occurring in unergative sentences are, for example, work, speak, shout, bark, or cough. Not all diagnostics that are valid for testing the unaccusative/unergative nature of the predicates in English or other languages are applicable to Estonian. Also, perhaps the nature of the phenomena the Estonian tests are sensitive to is different from what has been understood under unaccusativity. Whatever is the case, there are diagnostics that show that the Estonian intransitive verbs are not a homogeneous class either syntactically or semantically, and there is a correlation between those verbs’ semantic and syntactic properties (Tamm 1998). Further 9 discussion of some tests can be found in Tamm (1998). Here follows a brief presentation of what might be considered the the most reliable among them – the test based on the resultative construction. The test of resultative construction assumes the validity of the Direct Object Restriction. The Direct Object Restriction refers to the ability of any verb to have no more than one direct object, of which the resultative phrase can be predicated. This direct object can be overt or underlying, which is the case with unaccusative sentences. The idea of the test is that if a sentence has one underlying object, then no other object can be added. For instance, passive or impersonal verbs have an underlying object and do not allow the addition of any more objects, compare (14) and (15): (14) Leib lõigati viiludeks. The bread was cut slices.transl 'The bread was cut into slices.' (15) *Leib lõigati end/teda The bread was cut self/it. part ('The bread was cut it into slices.') viiludeks. slices.transl The unaccusative verbs are similar to impersonals and they are supposed to behave similarly, as can be demonstrated in (16) and (17): (16) Raamat kukkus laua book.nom fall.3sgpast table.gen ‘The book fell under the table.’ alla. under-directional (17) *Raamat kukkus end laua book.nom fall.3sgpast self table.gen (‘The book fell itself under the table.’) alla. under-directional Unergative intransitives, on the other hand, can assign object case to the phrase in the object position. They can have a nonsubcategorized object argument, for instance, in the form of a fake reflexive in resultative constructions. In those constructions, the resultative phrase is predicated of the fake reflexive, as illustrated in (18). (18) Tudeng köhis end hingetuks. student.nom cough.past3sg self breathless.transl ‘The student coughed himself breathless; the student coughed until he was short of breath.’ 10 The resultative phrase cannot be predicated of the subject with unergatives: (19) *Tudeng köhis hingetuks. student.nom cough.past3sg breathless.transl (‘The student coughed breathless.’) Not only fake reflexives, but also other nonsubcategorized, cognate objects can occur with the resultative phrase with the unergative type verbs (20): (20) Tudeng köhis oma hääle kähedaks. student.nom cough.3sgpast own voice.gen coarse ‘A student coughed his voice hoarse; the student coughed until his voice was hoarse.’ The resultative phrase is predicated of the surface subject with unaccusative verbs; this is demonstrated by example (21): (21) Tuhanded inimesed külmusid jääpurikaks. thousand.nompl people.nompl freeze.3plpast icicle.transl ‘Thousands of people froze (got frozen) into icicles.’ The insertion of a cognate object is not possible with unaccusatives, as in sentence (22): (22) *Tuhanded inimesed külmusid oma käed thousand.nompl people.nompl freeze.3plpast own-hand.3plnom (‘Thousands of people got-frozen their hands into icicles.’) jääpurikaks. icicle.transl These examples let us assume that there is an evident syntactic distinction between two types of Estonian intransitive verbs with regard to their behavior in resultative constructions. Therefore, this syntactic environment will further be employed to find out whether the subjects under discussion are underlying objects in Tenny’s theory. The organization of the following argumentation thus reflects the distinction between the clearly unaccusative intransitive sentences without oblique arguments (2.2.1.) and other intransitive sentences with oblique arguments (2.2.2.). Whether the intransitive sentences with obliques are unergative is a contentious issue that will be put aside for the time being. While unaccusative sentences do not necessarily pose theory-internal problems to the two accounts considered, the oblique-argument sentences certainly do so. This consideration has motivated the division of the presentation of the following arguments along those lines. Also, possible nonderivational solutions must consider the two different facets of a derivational solution in the field of objects in modeling the mapping between syntax and aspect. 11 2.2.1. Unaccusative intransitive sentences There are unaccusative sentences, a subtype of sentences lacking an overt direct object, that are compatible with Tenny’s tests of delimitedness. The verb sulama ‘melt’ is unaccusative; it can have an underlying object, since no object can be inserted into the sentence, and the resultative phrase can be predicated of the surface object. This fact is demonstrated by the resultative construction test in (23): (23) Jää sulas (*oma kihid) Ice.nom melt. 3sgpast its layer.plnom ‘Ice melted into water.’ (‘The ice melted its layers into water.’) veeks. water.transl The following sentence (24) shows that the sentence is compatible with the time frame adverbial ‘in a week’, and therefore it is delimited: (24) Jää sulas ühe Ice.nom melt. 3sgpast one.gen ‘The ice melted in a week.’ nädalaga. week.comit This sentence is unaccusative; it has an underlying object. Its argument is a measuring argument, and the sentence is delimited. Since the behavior of this sentence type correctly follows from Tenny’s and Van Hout’s predictions, the given verb class will not be further studied. The discussion will proceed to more problematic intransitive sentences. 2.2.2. Intransitive sentences with oblique complements According to Tenny’s and Van Hout’s mapping accounts, sentences without any (underlying) objects cannot express telicity or delimitedness and measuring out. This section shows that they can. More specifically, there are examples of Estonian intransitive sentences that cannot be classified as unergative by the resultative diagnostics that are identified above. Nevertheless they can express telicity or delimitedness and measuring out, but there are difficulties in establishing the exact measuring or delimiting argument. These sentences are based on verbs such as loobuma2 ‘give up, decline’, tutvuma ‘become acquainted with something or someone’, süvenema ‘concentrate on something’, pühenduma ‘become devoted to something’, rahulduma ‘become satisfied with something’, or joobuma ‘become elated, happy about something, become drunk with joy’. These verbs are special, on the one hand, since they can only form (surface) intransitive sentences as the unaccusative verbs do. They cannot be combined syntactically with a fake or a cognate object. But what speaks against considering these sentences as unaccusative is the fact that the resultative phrase cannot be predicated of the surface subject either. Also, their subjects are not uniformly theme type subjects, which is considered a typical type of unaccusative subjects. Many of these sentences resemble unergatives in that their subjects are agentive and volitional. Another characteristic of these verbs is that they 2 Cf the pattern 289, 30.0 of Rätsep (1978 : 193; 130). 12 are more abstract than the usual verbs studied in connection with aspectuality; in other words, their behavior is not dependent on cognitively perceivable or verifiable facts. This factor makes these verbs an interesting subject of study. Also, being more abstract entails for these verbs that the change is more abstract; the increments are more abstract; or the path of movement of the object is more abstract. Contrary to Tenny’s and Van Hout’s predictions, these verbs and sentences are able to express delimitedness and telicity. However, they demonstrate various degrees of compatibility with the time frame adverbial tests. The sentences are grouped from the least compatible to the most compatible according to the author’s judgements. Group 1: This group of verbs describing emotional reactions is the least compatible one with the time frame adverbials: (25) ?Mari joobus õnnest paugupealt M.nom get drunk happiness.elat immediately ‘Mary got happy/drunk with joy immediately.’ (26) ?Mari rahuldus piskuga paugupealt. M.nom was satisfied little.comit at once ‘Mary was/got immediately pleased (Mary was immediately satisfied) with very little.’ These sentences are ambiguous between the stative and the change of state interpretation. Intuitively, the change of state reading is possible in the sentences with those verbs, but it is not the prominent one, as indicated by the question mark reflecting the native speaker’s hesitation. Properties of stative verbs are discussed by Smith (1990:43), who demonstrates that statives are not compatible with expressions of dynamism, agency, or completion. These properties are indeed absent from the sentences’ stative readings. Those sentences are, however, compatible with expressions of simple duration and punctuality, an option for stative sentences, as shown by Smith (1990:43). In case of the change of state interpretation, the change is not proceeding in increments; this is demonstrated in the following examples (27) and (28). (27) *Mari joobus õnnest üha aeglasemalt/aeglaselt. M.nom get drunk happiness.elat (more and more) slowly ‘(More and more ) slowly, Mary got happy/drunk with joy.’ (28) *Mari rahuldus piskuga aeglaselt/üha aeglasemalt. M.nom was satisfied little.comit (more and more) slowly (‘(More and more ) slowly, Mary was pleased/satisfied with very little.’) As mentioned, the change of state reading is not the prominent one in those sentences: it is the stative reading that prevails. Also, the prominence of the stative interpretation would explain the 13 curious fact that these verbs are compatible with durative adverbials denoting a longer span of time than a mere moment. Compare sentences (29)/(30) and the above examples (25)/(26): (29) Mari joobus õnnest tund aega. M.nom get drunk happiness.elat for an hour ‘Mary was happy/drunk with joy for an hour.’ (30) Mari rahuldus piskuga tund aega. M.nom was satisfied little.comit for an hour ‘Mary was satisfied with very little for an hour.’ Here the durative adverbial indicates the boundaries of the state. The compatibility with durative adverbials does not reflect the fact that the event is proceeding in increments. Whether these verbs and sentences are agentive is tested by the modification with the adverbial ‘deliberately’ in (31) and (32): (31) *Mari joobus õnnest meelega. M.nom get drunk happiness.elat deliberately (‘Mary got deliberately happy/drunk with joy.’) (32) ??Mari rahuldus piskuga meelega. M.nom was satisfied little.comit deliberately (‘Mary was/got deliberately satisfied/pleased with very little.’) These verbs and sentences are not agentive, unless the verb rahulduma ‘become satisfied’ is used as a verb expressing an event of someone uttering that he is satisfied with very little. For Tenny, these verbs do not have measuring out arguments; the sentences based on these verbs are, however, delimited and nonhomogeneous. Group 2: This group of verbs of mental effort is rather similar to the previous one: its verbs describe either a state or a change of state. Contrary to the previous group, perhaps the change of state interpretation is pragmatically the more dominant one. The change can proceed in increments, even if this interpretation is not the most common one; the verbs are more agentive; these facts are demonstrated by the positive results of the combined tests in sentences (33) and (34): 14 (33) Mari M.nom pühendus devote.3sgpast paugupealt immediately /üha aeglasemalt /more and more slowly /tund aega /for an hour /meelega /deliberately /tund aega /for an hour /meelega /deliberately tööle work.allat ‘Mary devoted herself to her work.’ (34) Mari Mari.nom süvenes immerse.3sgpast paugupealt immediately /üha aeglasemalt /more and more slowly mõtte(i)sse thought.(pl)illat ‘Mary became/got immersed in thought.’ The test with a durative adverbial tund aega ‘for an hour’ pertains to the modification of a gradual change and to delimiting the duration of a state. These are telic, homogeneous (stative) and nonhomogeneous (change of state) sentences. It is difficult to pinpoint the argument that delimits and measures out the event -- the whole situation or state of affairs, within it most of all participants, the subject referent, undergoes a change. Group 3 : The verb tutvuma ‘to get acquainted with someone or something’ describes a change of state (35). Contrary to the examples of the previous verb groups, this verb cannot describe a state. Incrementality and measuring out is possible if the sentence is understood agentively: (35) Mari M.nom tutvus get acquainted.3sgpast Katiga Kate.comit paugupealt /??üha aeglasemalt immediately /more and more slowly ‘Mari got/became acquainted with Kati.’ /tund aega /for an hour /meelega /deliberately This verb is also telic, occurring in homogeneous (process or activity) and nonhomogeneous (change of state) sentences; it is the whole situation and within it, most of all the subject referent that undergoes a change. The sentence (36) is another objectless sentence with the same verb. 15 Here, the event could be measured out by the comitative phrase. The verb can more felicitously be combined with rate adverbials and durative adverbials if the referent of the comitative phrase is a mass noun: (36) Mari M.nom ??paugupealt immediately tutvus get acquainted.3sgpast materjaliga material.comit /üha aeglasemalt /more and more slowly /tund aega /for an hour /meelega /deliberately ‘Mary became/got acquainted with the material.’ The difference between sentences (35) and (36) is the quantification of the comitative phrase. Sentence (36) allows the comitative phrase to be interpreted as a mass noun; therefore it can be more felicitously modified by the durative adverbial and the rate adverbials. The progression through the material reflects the progression through the event. However, the material is not a direct object, although the end of the event coincides with the moment when the material is totally processed by Mary. Also, the sentences differ in the role of Mary in the situation. In (35), Mary does not necessarily make any effort to become acquainted with Kate. In (36), Mary necessarily makes an effort to become acquainted with the material. Group 4: The verb loobuma ‘decline’ describes a change of state (37). There is no incrementality or possible measuring out present in this agentive sentence: (37) Haigekassa nõukogu ‘The Council of Public Health paugupealt immediately loobus give-up.3sgpast /*üha aeglasemalt /more and more slowly /*tund aega /for an hour /meelega /deliberately otsustamisest. deciding.elat ‘The Council of Public Health declined to decide; the Council declined to make the decision.’ If any of these objectless verbs can be seen as unaccusative, then they are predicted to be able to measure out by Tenny’s theory. If not, then there is a problem. Since the essence of the phenomenon of unaccusativity has not been adequately studied for Estonian, this article will leave this topic to be exhausted in the future. The conclusions are drawn considering that none of the analysed verbs are included in lists of unaccusative verbs, and these verbs do not allow either an additional object in a sentence or the resultative phrase predicated of the subject. 16 This subsection has shown that there are many intransitive predicates that nevertheless are compatible with Tenny’s tests. The sentences are delimited even though they have predicates without any overt object that could be ‘held responsible’ for the aspectual properties emerging in the sentence. Since the tests that Tenny uses for delimitedness are identical with Van Hout’s tests for (a)telicity, the presented Estonian data are problematic for Van Hout’s analysis as well. The next section will elaborate on particled verbs and their relation to objects, obliques and aspectual oppositions. 2.3. Oblique arguments can co-occur with perfective verbal particles We have seen that Van Hout’s and Tenny’s theories can account for several aspectual phenomena in intransitive sentences if they use a derivational framework where objects appear in underlying syntactic structures. This section takes a look at intransitive nonhomogeneous or delimited sentences with oblique arguments and particles. Typically, resultative phrases or verbal particles and (especially genitive/nominative) objects come in pairs in Estonian. However, there is a difference between the relation of measuring out and objects in the following two sentences with the same verbal particle ära, taken from Metslang (2001: 449, 453-454): (38) Päike sulatas suure jääpurika ära. sun big.gen icicle.gen particle melt.3spast ‘The sun melted the big icicle.’ In sentence (38), the object is affected totally, the icicle has disappeared. In sentence (39), the object is not affected; there is no concrete clown being affected by playing. (39) Mind kutsuti reklaamipäevale I.part invite.imperspast advertising_day.allat clown.part Mängisin ära klouni ja sain klouni mängima. play.mainf sada krooni play1sgpast clown.gen particleand get.1sgpast a hundred.nom kroon.part ‘I was invited to play a clown on the advertising day. So I played the clown and got 100 kroons.’ A type of acting, clown-acting, is performed, and this performance is carried out up to its final point; the final point coincides with the fulfillment of a set of acts, the fulfillment requirements set by the person paying a hundred crowns. The particle ära has become more frequent in Estonian; it expresses the reaching of a result, and perfectivity. The following examines the range of possibilities with the particle and the given verb from Metslang (2001). First, it is possible to ‘drop’ the object ‘clown’ and use the following sentence (40): (40) (Pidin 10 minutit klouni mängima. I had to play the clown for ten minutes.) 17 Mängisin ära ja sain play.1sgpast particleand get.1sgpast ‘I played and got a hundred kroons.’ sada krooni. hundred.nom crown.sgpart A construction with dropped object klaverit ‘the piano’is also acceptable3: (41) Pidin 10 minutit klaverit mängima. I had to play the piano for ten minutes. Mängisin ära ja sain play.1sgpast particleand get.1sgpast ‘I played and got a hundred kroons.’ sada krooni. hundred.nom crown.sgpart The previous examples have shown that the objects can be unaffected in the course of the action and even be missing from the sentence with the particle ära. An object, therefore, stands in a different relation to nonhomogeneity ( or telicity, as used by Van Hout). Also, an object need not be totally affected in order to be marked with the genitive/nominative case. The object can semantically serve as the specification for the type of action that is carried out: the clown-playing type of playing. Achieving the intended end of the action can qualify as a result that is sufficient to case-mark the object with genitive. However, the object can be understood as underlying in these sentences. There are intransitive particle verb sentences where there are, however, no underlying objects. The third example, sentence (42), contains the same verb and a comitative complement. Here follows the description of the third situation: (42) Lapsed pidid järgemööda mängima autodega, ladudes neid värvi järgi ritta, siis klotsidega ja seejärel nukkudega. ‘The children had a task to play with cars, arranging them according to their color; then they had to play with blocks, and then with dolls.’ Mari Mari.nom mängis play.3sgpast autodega car.pl.comit ära (ja siis asus klotsidega mängima). particle (and then settled down to play with the blocks). ‘Mary played with the cars (and finished this part of the task), Mary finished playing with the cars and then started to play with the blocks.’ However, my intuition would find the following object-verb combination ‘to complete playing the piano’ stranger: Pidin 10 minutit klaverit mängima. I had to play the piano for ten minutes. ??Mängisin klaveri ära ja sain sada krooni. play.1sgpast piano.gen particle and get.1sgpast hundred.nom crown.sgpart ‘I played the piano and got a hundred kroons.’ 3 18 Contrary to expectations, the sentence (42) is compatible with Van Hout’s (2000) tests for telicity and also with Tenny’s tests of delimitedness: (43) Mari Mari.nom mängis play.3sgpast viie minutiga in five minutes autodega car.pl.comit ära particle ja siis asus klotsidega mängima. and then settled down to play with the blocks. ‘Mary played with the cars (and finished it), Mary finished playing with the cars in five minutes and then started to play with the blocks.’ If there is a specific task related to the situation described by the verb with a comitative argument, and the task is fulfilled, then the delimiting particle ära can occur in the sentence. There is no object present in the sentence. The following three examples with particles, expressing delimitedness, have no objects either: (44) Ta hüppas tulest ühe sekundiga s/he.3sg jump.3sgpast fire.elat one.gen second.comit ‘He jumped over the fire in a second.’ üle. over (45) Ta sõi ühe minutiga ära. s/he.3sg eat.3sgpast one.gen minute.comit particle ‘She finished eating in one minute.’ (46) Ta sai vastasest nädalaga s/he.3sg get.3sgpast enemy.elat week.comit ‘S/he defeated/overcame the enemy in a week.’ võitu/jagu. won Here we see that culmination-orientedness exists without any contribution from subcategorized objects. The same conclusion can be drawn about delimitedness as defined by Tenny. However, these particle verb sentences with obliques are much less felicitous with durative adverbials. Iterativity, a possible interpretation of sentence (47), is out of the scope of this study. Disregarding the possible iterativity, compatibility with durative adverbials would presume a very conscious choice of the agent to attain the goal, or even misjudgement of the situation by the agent. If these conditions are not present, the sentence (47) is not compatible with a durative adverbial: (47) *Ta sai vastasest nädal aega s/he.3sg get.3sgpast enemy.elat week.nom time.part ‘S/he was defeating/overcoming the enemy for a week.’ võitu/jagu. won 19 In sum, there is ample evidence that delimitedness and nonhomogeneity (telicity) do not require an overt object. Thus, nonhomogeneity (telicity) and delimitedness do not have as strong linkage to syntactic frames as envisaged in these theories. These aspectual properties, however, have a very distinct morphological marker in Estonian if the sentence has an object – the genitive/nominative object case. However, achievement verbs are less suitable for the expression of these oppositions and are more likely than accomplishment verbs to appear in verb frames without objects. As a conclusion to section 2, it can be admitted that objects typically occur in Estonian sentences expressing culmination-orientedness or measuring out. However, the correlation between these aspectual properties and objects is a tendency and not an absolute rule. Several intransitive sentences are found to display the aspectual properties of telicity, culminationorientedness and measuring out. This article examines intransitive sentences with obliques that contain verbs such as loobuma ‘give up, decline’, tutvuma ‘become acquainted with something or someone’, süvenema ‘concentrate on something’, pühenduma ‘become devoted to something’, rahulduma ‘become satisfied with something’, or joobuma ‘become drunk with joy’ (cf. the examples (25)-(37) in section 2.2.2). In order to further evaluate Tenny’s and Van Hout’s analyses, a theory on Estonian underlying objects is essential. However, the data includes cases where measuring out and culmination-orientedness are possible without any type of objects (tutvuma ‘to get or become acquainted with’), which is problematic for both accounts surveyed. Some of the above listed verbs may receive an explanation in one of the surveyed theories, but more often, they are problematic. In a nutshell, in the following lines I sketch some of those points where a different approach might be more successful than the reviewed theories. First, it turns out that Tenny’s account does not cover many aspectual verbs with oblique arguments (cf. rahulduma ‘become satisfied with something’, joobuma ‘become happy about something, become drunk with joy’). Second, Tenny (1994) does not explain the occurrence of objects in sentences where the implicit path measures out the event. Therefore, while Tenny’s account of the aspect-object correlations is better constrained than that of Van Hout, its scope is narrow for establishing the tendencies observable in the Estonian data. In addition, the syntactic differences that arise in the case of delimitedness without measuring out are not treated in detail. The lack or the reduced importance of measuring out is the typical case of achievement verbs, which form a natural class in many languages. However, while Van Hout’s scope is considerably wider, her account can be improved by clarifying the relation between the terminology and the Estonian data. The aspectual oppositions that occur with verbs with non-alternating frames and resultative constructions are not considered; also, for instance, the intransitive verbs süvenema ‘concentrate on something’, pühenduma ‘become devoted to something’ are not predicted to form aspectually opposite sentences. The next section offers an overview of the tendencies in Estonian syntax and aspect, followed by some tentative pointers to the solutions. 3. Tendencies in the relations between Estonian aspect and objects Three tendencies can be established in the relations between aspect and syntax in Estonian. One, the presence of an object tends to depend on culmination-orientedness and measuring out in the sentence, and thus on the presence of verbs and constructions that express culmination-orientedness and measuring out. The reverse side of this tendency is that purely culminational and delimited sentences, typically containing achievement verbs, display more 20 exceptions to this tendency of having objects. Two, Estonian objects, as opposed to obliques, tend to occur when the sentence measures out even if the measuring out is not related to overt paths or objects. Three, volition is the constraint for enabling measuring out in many sentences with achievement verbs if the measuring out in a sentence is not related to (implicit) paths and objects. The passages below explicate each of these tendencies in turn. 3.1. Objects occur in culmination-oriented sentences First, the presence of objects tends to depend on the presence of culmination-orientedness in the sentence – in other words, on the possibility of measuring out. The tendency in Estonian is, therefore, that it is not nonhomogeneity, or delimitedness alone, that would bear a relation to the occurrence of syntactic objects. This tendency is more specific than the account of Van Hout and confirms Tenny’s theory. The discussion in section 2.1. has pointed out that Van Hout’s approach builds on partly overlapping terminology. Clarifying the terminology has helped to view the Estonian data in a different light. Most importantly, a division of labor between culmination-orientedness and homogeneity has been pronounced relative to the Estonian data. Assuming two terminological distinctions, two related but different syntactic phenomena are explained. First, the aspectual distinction between culmination-oriented and not culminationoriented sentences predicts some regularities of verb frame alternations (cf. the examples (10)(13) with the verb lõhkuma ‘to break’). Culmination-oriented sentences have typically casealternating objects. Examples of this type are abundant; some verbs that typically occur in culmination-oriented sentences are listed here: alistama ‘subjugate’, ehitama ‘build’, istutama ‘plant’, kirjutama ‘write’, minetama ‘forfeit, lose’, omandama ‘acquire’, rajama ‘create, establish’, trükkima ‘print’, and äratama ‘rouse, wake up’. Sentences describing eventualities without culmination-orientedness can – but need not – have objects, and the objects can only be case marked with the morphological partitive case. Here follow some examples with verbs that occur in sentences that have only partitive objects: võrdlema ‘compare’, abistama ‘help’, alahindama ‘underestimate’, armastama ‘love’, huvitama ‘interest’, kahtlustama ‘suspect’, jätkama ‘continue’, nautima ‘enjoy’, pooldama ‘be on the side of, support’, sallima ‘tolerate, stand’, and õigustama ‘justify’. Most, but not all, particle verbs are culmination-oriented, and the sentences with the particle ära are always culmination-oriented. The second aspectually relevant distinction, that of homogeneity and nonhomogeneity, divides the data on object case alternation. Homogeneous sentences have partitive objects, and nonhomogeneous sentences have genitive/nominative objects. The phenomenon of object case alternation is indirectly related to culmination-orientedness; naturally, object case alternation is possible only if there is an object in the verb frame. Object case alternation, therefore, strictly reflects the opposition of homogeneity, which equals to telicity in the sense of Van Hout (2000). It is important to note that homogeneity and verb frames do not display relevant correlation. Verb frames with and without an object can yield equally homogeneous sentences. On the one hand, the sentence (8) with ehitas masinat ‘was building a machine’, or sentences describing states, processes and activities, such as (48) and (49), have homogeneous reference as well as objects: (48) Mari alahindas Tooma M.nom underestimate.past3sg Thomas.gen ‘Mari underestimated Thomas’s cleverness.’ kavalust. cleverness.part 21 (49) Mari saatis lauljat M.nom accompany.3sgpast singer.part ‘Mari accompanied the singer on the piano.’ klaveril. on the piano On the other hand, intransitive sentences can be as homogeneous as the above sentences – for instance, the sentence in (50): (50) Mari seisis. M.nom stand.3sgpast ‘Mari was standing.' In addition, verb frames with and without an object can yield equally nonhomogeneous sentences; for instance, compare the sentence (40) with klouni ära mängima ‘finish playing the clown’ and the sentence (37) with the oblique loobuma ‘decline’. Therefore, even if the object tends to be related to culmination-orientedness, it is not true that homogeneity has any strict relationship to culmination-orientedness or objects. There is one more facet of the correlation between culmination-orientedness and objects in Estonian. More specifically, sentences expressing pure delimitedness without measuring out occur more frequently without objects than the sentences that express measuring out and delimitedness. Typical sentences that express delimitedness without measuring out contain achievement verbs. Even some schoolbook examples of achievement verbs, which occur with an object in English, such as ‘reach the top’, do not have object complements in Estonian, but obliques instead: (51) Mari jõudis tippu/tipuni/tipule. M.nom reach.3sgpast top.illat/termin/allat ‘Mari reached the top.’ Pure delimitedness correlating with the lack of objects still has the status of a slight tendency in Estonian, since achievement verbs with objects are abundant. 3.2. Objects occur also if the object is not the measuring argument The second tendency differs from the account of Tenny (1994) and confirms the insights of Van Hout (2000). More specifically, Estonian objects, as opposed to obliques, tend to occur when the sentence measures out, even if the measuring out is not related to overt paths or objects. Estonian has a case-alternating object, for instance, in sentences with non-measuring objects, as in Bill pushed the cart to the store (see sentence (52)). (52) Bill lükkas vankri B.nom push.3sgpast carriage.gen ‘Bill pushed the carriage to the store.’ poodi. store.illat 22 Verbs with case alternating objects can form sentences with objects that diverge in their measuring out properties. The objects of those verbs can be either moved objects, which are not measuring arguments, or affected objects, which are measuring arguments. Consider the verb vajutama ‘squeeze, push, press’ in (53) and (54). (53) Mari vajutas tuubi M.nom squeeze.3sgpast tube.gen ‘Mari squeezed the tube empty.’ tühjaks. empty.transl The object is an affected object and a measuring argument. (54) Mari vajutas tuubi M.nom squeeze.3sgpast tube.gen ‘Mari squeezed the tube into the box.’ karpi. box.illat The object is a moved object and not a measuring argument. The tendency in Estonian is that any type of measuring out or culmination-orientedness correlates with the overt case-alternating object. 3.3. Volitional sentences have ‘volitional measuring scales’ The third tendency concerns the special status of volitional sentences. This sentence type displays a different aspect-syntax relationship than covered by the accounts of Tenny and Van Hout. If the measuring out in a sentence is not related to (implicit) paths and objects, however, volition becomes a crucial constraint in enabling such measuring out. The data in 2.2.2. have shown that volitionality corresponds to the possible presence of a measuring out scale in sentences. Therefore, volitionality corresponds often, but not necessarily, to the possible presence of a case-alternating object in the sentence. However, volition-induced measuring out scales have frequently no relation to the mapping to objects or physical paths in the observed data (cf. tutvuma ‘become acquainted with something or someone’, süvenema ‘concentrate on something’, pühenduma ‘become devoted to something’). The presence of such a ‘volitional measuring out scale’ is demonstrated by the compatibility of the volitional sentences with durative adverbials. Another indicator for a measuring out scale in volitional sentences with the discussed verbs is the possibility of indicating more precisely the non-measuring reading of the sentence with the particle ära: (55) Mari tutvus materjaliga/Katiga ära. M.nom get-acquainted material/K.comit ära ‘Mari accomplished her task, her task being getting acquainted with the material/Kati.’ If the verb would be unambiguously non-measuring, the particle ära would have no function. In the volitional readings of the sentences of section 2.2.2., therefore, culmination-orientedness and measuring out can be expressed. Thus, the volitional sentences with these verbs can have homogeneous reference. Homogeneity is possible even if the sentences are composed of verbs 23 and verb-particle combinations that typically express only delimitedness and have nonhomogeneous reference. On the other hand, the less volitional, experiencer readings of those sentences can express only culmination and delimitedness. Those sentences have always nonhomogeneous reference. Therefore, the measuring out scales that constitute the measuring out in nonvolitional sentences must be constrained to be only object- or (implicit) path related. This is exactly what Tenny’s theory accounts for. On the contrary, the measuring out scales that make up measuring out in volitional sentences must not be seen as just object or (implicit) path related – volitional sentences have the possibility of an extra ‘volitional measuring scale’. Only a tentative, intuitive explanation can be offered here why volitional sentences have this possibility to refer to more abstract culmination-oriented activities in their temporal progress. The measuring out is then seen as the mapping between the event and the culmination-oriented activities – in other words, subevents that constitute the event described by the sentence. Let us consider sentence (12) again, repeated here as (56): (56) Töömees lõhkus aiaväravat maha. Repairman.(nom) break.3sgpast gate.part down ‘The repairman was breaking/taking down the gate.’ Naturally, the object-related subevents correspond to the increments of the gate coming down in the course of the event. However, the repairman might be choosing his tools for fulfilling his task, or he might be consolidating the constructions around the gate in preparation for the concrete action of taking down the gate. Also, those activities can be referred to as part of the scenario of taking down the gate. Because what is inflicted by the storm is non-volitional, preparatory activities cannot be understood to be part of what is happens to the gate in sentence (57): (57) *Torm lõhkus aiaväravat maha. Storm.(nom) break.3sgpast gate.part down ‘The storm was breaking/taking down the gate.’ Only a volitional participant in the event is felicitously understood to carry out complex tasks on his way to the result described in the sentence. A volitional participant can subordinate those preparatory tasks to the fulfillment of the overall task as relevant parts of it. In particular, the fulfilling of complex tasks requires (often causally) related activities that remove the sequence of obstacles and hazards that lie in the way of obtaining the result. Nonvolitional participants are not felicitously understood to involve in those preparatory activities on their way to a further goal. 3.4. Some achievements have no ‘volitional measuring scales’ However, volitionality does not always license the ‘volitional measuring scale’; this is the case with sentences containing genuine achievement verbs. Achievement sentences lack the possibility of a ‘volitional measuring scale’ (cf. loobuma ‘give up, decline’, võitu/jagu saama ‘overcome, win’). The fact that these sentences lack a measuring scale is shown by their 24 incompatibility with the durative adverbial test (see 2.2.2.). Perhaps another suitable test for singling out (volitional) achievement verbs with oblique complements involves combining them with the particle ära. Volitional achievement verbs with no ‘volitional measuring out scale’ tend not to combine with the verbal particle ära in sentences: (58) *Haigekassa nõukogu loobus otsustamisest ära. ‘The Council of Public Health give-up.3sgpast deciding.elat ära (‘The Council of Public Health finished declining to decide; the Council finished its task of declining to make the decision.’) Sentences with those verbs are delimited and have nonhomogeneous reference; they do not allow any interpretation along the lines of several culmination-oriented activities. There can be activities related to the ultimate goal, but they cannot be referred to by the same verb in this particular grammatical form. Only a tentative, intuitive explanation can be offered here as to why these achievement verbs lack the possibility of a ‘volitional measuring scale’. It is not possible to construe any subevents that would constitute the event described by that specific lexical verb in a sentence. Therefore, the temporal mapping between the event as described by the sentence, and any of imaginable related subevents, except the event itself, is lexically rendered impossible. 4. Conclusion This article has lead to the conclusion that case-alternating objects typically occur in Estonian sentences expressing telicity (Van Hout 2000) or measuring out (Tenny 1994). However, the relation between these aspectual properties and objects has the status of a tendency and not the status of an absolute rule. That tendency has been outlined in further detail in this article. Contradicting the explanation of Van Hout and confirming a possible interpretation of Tenny’s theory, it is not nonhomogeneity, or delimitedness, that correlates with syntactic objects. Rather, the presence of objects tends to depend on the presence of culmination-orientedness in the sentence – in other words, on the possibility of measuring out. However, differently from the account of Tenny and according to the insights of Van Hout, Estonian objects tend to occur in measuring out sentences even if the measuring out is not related to (implicit) paths or objects. In those cases, volition becomes a crucial constraint in allowing such measuring out. Some achievement verbs lack the possibility of a ‘volitional measuring scale’. A tentative explanation for the distribution of volitional measuring scales has been offered in terms of the possibility of a mapping between the event and its subevents. The further articulation of the relations between verbs, their objects, obliques, aspect and thematic roles is a promising task for future research in the crossroads of Estonian lexicon, semantics, and syntax. Bibliography Comrie, Bernard (1976) Aspect. An Introduction to the Study of Verbal Aspect and Related Problems. Cambridge University Press. Cambridge, London, New York, Melbourne. Dahl, Östen (1985) Tense and Aspect Systems. Oxford. Depraetere, Ilse (1995) ‘On the necessity of distinguishing between (un)boundedness and (a)telicity’, Linguistics and Philosophy 18/1: 1-19. Levin, Beth and Malka Rappaport Hovav (1995) Unaccusativity: At the Syntax-Lexical Semantics Interface. Number 26 in Linguistic Inquiry Monographs. MIT Press. 25 Metslang, Helle (2001) ‘On the Developments of the Estonian Aspect: the Verbal Particle ära.’ In Östen Dahl and Maria Koptjevskaja-Tamm (eds.): The Circum-Baltic Languages: Their Typology and Contacts. Studies in Language Companion Series 55. Amsterdam, Benjamins: 443-479. Kiparsky, Paul (1998) ‘Partitive Case and Aspect’. In Miriam Butt and Willem Geuder (eds.) The Projection of Arguments. Stanford, CSLI Publications: 265–307. Rätsep, Huno (1978) ‘Eesti keele lihtlausete tüübid.’ [Types of Estonian simple sentences.] ENSV TA Emakeele Seltsi toimetised 12. Valgus, Tallinn. Smith, Carlota (1990) The Parameter of Aspect. Kluwer, Dordrecht. Tamm, Anne (1998) ‘Unaccusativity. Some notes on the phenomenon in Estonian.’ Handout for LingDoc98, Szeged, Nov. 20, 1998. Tenny, Carol (1994) Aspectual Roles and the Syntax-Semantics Interface. Studies in Linguistics and Philosophy, Vol. 52. Kluwer Academic Publishers. Dordrecht/ Boston/ London. Van Hout, Angeliek (2000) ‘Event-Semantics in the Lexicon-Syntax Interface: Verb Frame Alternations in Dutch and their Acquisition.’ In Carol Tenny and James Pustejovsky (eds) Events as Grammatical Objects. Stanford, CSLI Publications: 239-282. 26