2. Particle Verbs with an and zu

Blom (2004, 2005) discusses in detail complex predicates in Dutch both synchronically and diachronically and calls the particles aan and toe in 4 orienting particles.

1. (4)

   1. a.

   2. b.

The base verbs in 4 denote intransitive activities, and the particles “express the direction toward which these activities are oriented” (Blom 2004:23). She assigns the semantic representation 5 to the verbs with orienting particles.Footnote ¹

1. (5)

Since orienting particles introduce a participant “y” in 5, the resulting particle verbs must be transitive, according to Blom. It is then expected that particle verbs with orienting particles in German always have an accusative argument. This is the case in 2a, repeated here as 6a. German in fact has many verbs with the orienting particle an. Stiebels (1996:162ff.) gives the examples in 6b, and McIntyre (2004:538) adds those in 6c. In these verbs the participant toward whom/which the activities are directed is encoded by an accusative object.Footnote ²

1. (6)

   1. a.

   2. b.

   3. a.

However, in 3a, repeated here as 7, one might regard the dative argu-ment as representing a person toward whom the action is directed as well. The question then arises as to whether the two particle verbs anlächeln and zulächeln are synonymous. We have also similar pairs like an-/zusprechen ‘speak to’, an-/zulachen ‘laugh at’, an-/zusingen ‘sing to’, an-/zublinzeln ‘blink at’, an-/zuschreien ‘shout at’, an-/zubrüllen ‘roar at’, an-/zubellen ‘bark at’, at the least.

1. (7)

Blom (2005:137) regards the Dutch particles toe and aan in 8 as orienting particles as well. That is, these particles introduce a participant toward whom the action is directed. The difference between 4 and 8 consists in the fact that in the latter this participant is expressed by an indirect object. She assigns the semantic representation 9a to the ditrans-itive verbs in 8. For example, 8a is given the representation 9b.

1. (8)

   1. a.

   2. b.

1. (9)

   1. a.

   2. b.

As is clear from 9b, the verbal phrase in 8a denotes a change of location. In this respect the indirect object de spelers ‘the players’ in 8a bears the same semantic role as the prepositional phrases naar de spelers ‘to the players’ and aan de onderwijzer ‘to the teacher’ in 10.

1. (10)

   1. a.

   2. b.

However, the semantics in 8 is more complicated, as Olsen (1997) points out on the basis of German data. Olsen shows that the dative argument in German particle verbs like 11 is structurally located higher than the accusative argument. Her main argument consists in the fact that the word order “dative – accusative” in 11 permits an interpretation in which the whole sentence is focused. As is well known, this inter-pretation is possible only if the sentence shows the basic word order. The basic word order “dative – accusative” is also found in verbs like geben ‘give’ or schenken ‘present’, which clearly express a change of pos-session. The same ordering behavior shows that both the dative arguments in 11 and those in the verbs geben ‘give’ or schenken ‘present’ are of the same type, namely the so-called high dative (see Cook 2006, McFadden 2004, 2006, McIntyre 2006, Meinunger 2006).

1. (11)

   1. a.

   2. b.

   3. c.

   4. d.

In addition, Olsen points out that the dative object in 11 can be “promoted” to the subject in so-called bekommen-passives, which indi-cates again that the particle verbs in 11 belong to the same class as the typical change of possession verb schenken ‘present’.Footnote ³

1. (12)

   1. a.

   2. b.

   3. c.

1. (13)

Olsen (1997:321) then proposes the semantic representation 14 for the German ditransitive particle verbs in 11 and suggests that the dative in these verbs corresponds both to the object of the predicate AT and to the subject of the predicate POSS, namely “z.” Note that AT expresses a goal and POSS is an abbreviation of possession in a broad sense. The latter predicate therefore stands for a beneficiary relationship between the two arguments as well.

1. (14)

The representation 14 is desirable, in so far as it correctly captures the fact that the argument y is assigned an accusative case, since in the linking system that Olsen employs the most deeply embedded argument in the representation is expected to be realized as an accusative argument in German (see Wunderlich 1997). However, it is not entirely clear from her analysis how the predicate POSS is licensed. Olsen notes that the POSS predicate overlies the locational semantics of the particle zu ‘to’, which is represented in the middle of 14, namely [BECOME(LOC (y, AT*[z]))], but she does not explicate why this locational meaning component must be overlaid by the POSS predicate.

Now I turn to the mechanism that licenses a dative in 11 and an indirect argument in 8. Note that both German and Dutch have a con-struction like 15, in which the dative or the indirect object binds the body part noun expressed in a prepositional phrase.Footnote ⁴ The dative or the indirect object here denotes the person who inalienably possesses the corres-ponding body part.

1. (15)

   1. a.

   2. b.

   3. c.

   4. d.

   5. e.

Let us assume that in 16 the suppressed ground argument represents the body of the person towards whom the action is directed and that the dative or the indirect argument binds this invisible argument. The structure of the construction is then roughly represented in 17.

1. (16)

   1. a.

   2. b.

1. (17)

   1. a.

   2. b.

This assumption is compatible with the following two facts. First, if the dative argument in 16a is attributed to the possessor dative, the semantics of the sentence is roughly described as follows: The ball was thrown to the body of the child, and since the body is possessed by the child, it is easy to infer that the child got the ball. This is exactly what the representation in 14 proposed by Olsen (1997) expresses, repeated here as 18.

1. (18)

The second piece of evidence for the analysis in 17 lies in the fact that the datives in 16 cannot be omitted. The possessor dative appears obligatorily in general, as shown in 20. If the dative in 17 represents also a possessor dative, it is expected that this dative argument cannot be omitted.

1. (19)

   1. a.

   2. b.

   3. c.

1. (20)

   1. a.

   2. b.

The obligatory occurrence shows that this dative cannot be classified as a dative (in-)commodi, which is semantically another type of the high dative. This type of free dative denotes the person to whose advantage or disadvantage the event results, and its occurrence is not obligatory. In this respect the predicate POSS in 18 that Olsen employs is insufficient, since this predicate represents not only an inalienable possession, but also a beneficent or a maleficent relationship.

1. (21)

   1. a.

   2. b.

I conclude that particle verbs with double objects in German and Dutch are based on the transitive constructions in which the possessor dative binds the body part noun inside PP.Footnote ⁵ Therefore, we should say that the dative argument in these verbs is not introduced by the particle itself, but by the suppressed ground argument inside PP, in contrast to the accusative argument in the particle verbs with an and aan in 22.

1. (22)

   1. a.

   2. b.

Now let us come back to the question of whether the particle verbs in 22a and those in 23 with a dative are synonymous.

1. (23)

I argue that the dative in 23 is not introduced by the particle zu directly, but is based on the inalienable semantics that gives rise to particle verbs like zuwerfen ‘throw to’. That is, the dative in 23 is not the person toward whom the action is directed, but the recipient of something which is created by this action. In short, the dative in 23 is not a ground argument. We have two pieces of evidence for this claim. First, the particle verbs in 23 optionally take a cognate object, which indicates that the direct object position in these verbs is not empty.

1. (24)

   1. a.

   2. b.

   3. c.

   4. d.

   5. e.

   6. f.

Semantically, the accusative object in 24 represents a product or a sign that is created by the action. It is reasonable to assume that the agentive subject in 24 causes the dative argument to have such a product or a sign. This assumption is compatible with the fact that the dative object in 24 can be “promoted” to the subject in bekommen-passives.

1. (25)

   1. a.

   2. b.

   3. c.

   4. d.

   5. e.

   6. f.

Thus we can postulate that the formation principle of the particle verbs with zu in 23 is basically the same as that of transitive particle verbs like zuwerfen ‘throw to’ or zuschieβen ‘shoot to’ in 11.

The second evidence for the claim that the dative in 23 represents a recipient comes from its semantic restriction. It is widely known that a recipient must be sentient, so a coma victim, for example, cannot occur as an indirect object in English. Wegener (1985) also counts Belebtheit ‘living’ as one of the important features of the German high dative.

1. (26)

   1. a.

   2. b.

   3. c.

This restriction seems to be found in particle verbs with zu in 23 as well. If the dative argument represents a dead person, then the verbs with the particle an are given preference over the verbs with the particle zu. For native speakers, a certain reaction is expected in the case of particle verbs with zu.

1. (27)

   1. a.

   2. b.

I conclude that activity verbs prefixed with zu like zulächeln ‘smile at’ are not synonymous with those with an like anlächeln ‘smile at’. In the former the dative is not introduced by the particle, but licensed by the inalienable semantics. In the latter the accusative is introduced by the particle and represents a ground.

3. Particle Verbs with Low Dative

Next I consider the following difference between German and Dutch: Whereas German can encode the goal, which corresponds to the ground of the particle zu—for example, dem Bahnhof ‘the station’ in 28a—in dative, the corresponding expression cannot be expressed by an indirect object in Dutch, but requires a prepositional phrase headed by naar or op.Footnote ⁶

1. (28)

   1. a.

   2. b.

The base verb in 28a denotes a motion of the subject. Other German particle verbs of this type are given in 29.

1. (29)

The dative in 28a and 29 must be differentiated from those in 30, which were discussed in the previous section. The former denotes a non-human entity, and no change of possession is expressed in the particle verbs in 28a and 29. Therefore, it is not possible to form bekommen-passives from these verbs.Footnote ⁷

1. (30)

   1. a.

   2. b.

1. (31)

   1. a.

   2. b.

The fact that the dative object in 28a and 29 represents the goal of the movement suggests that this type of German dative is a so-called “low” dative (see also McIntyre 2006:207). Typical German verbs containing a low dative are given in 32, taken from Haider & Rosengren 2003, McFadden 2004, and Cook 2006. It has been pointed out by them that low datives denote a location or a direction.Footnote ⁸ This semantic feature leads McFadden (2004:114f.) to propose that the low dative is in fact a PP with a hidden prepositional head. The VP of the verb aussetzen ‘expose’, for example, is according to him represented roughly as in 33.

1. (32)

1. (33)

My claim that the dative in 28a and 29 represents a low dative is supported by the following fact. As Wegener (1991) and McFadden (2004) point out, both verbs with a low dative and transitive verbs expressing a change of position like legen ‘lay’ show the same behavior with respect to sentential negation. As shown in 34, the negation adverb nicht ‘not’ should be placed before the dative and the directional PP, respectively.

1. (34)

   1. a.

   2. b.

This ordering restriction on the negation adverb is also found with the German particle verbs in 28a and 29. It is shown that the dative argument in 35 behaves like a directional phrase.

1. (35)

   1. a.

   2. b.

Now note that the Dutch expressions that correspond to German verbs with a low dative are given in 36.

1. (36)

   1. a.

   2. b.

   3. c.

   4. d.

   5. e.

   6. f.

As is clear from here, the German phrases with low datives cannot be expressed by double object constructions in Dutch, even if morpho-logically closely related verbs are used.

In conclusion, the low dative, which often denotes the goal of a movement, is lacking in Dutch, and this is the reason for the difference between German and Dutch discussed in this section, namely the differ-ence of how the goal argument is expressed with motion verbs with zu and toe. Footnote ⁹

4. Unpredicated Particles

In this section I discuss the contrast between German and English displayed in 37 and 38.

1. (37)

   1. a.

   2. b.

1. (38)

   1. a.

   2. b.

The sentence in 37a denotes the situation in which the subject poured something into the glass, and the sentence in 37b denotes the situation in which the subject loaded something onto the truck. In both cases the semantic subject of the particle ein ‘in’ and auf ‘on’, namely the figure, is suppressed, and the semantic object of these particles, namely the ground, is realized as a direct object of the verbs. In contrast with German the corresponding constructions in English are ruled out, as shown in 38.Footnote ¹⁰ In most of the transitive particle verbs in English the direct object corresponds to the figure, and the ground appears as a direct object only in some verbs with the particles off and out, as seen in 39. Since they are not predicated of direct objects, Levin & Sells (to appear) call them unpredicated particles.

1. (39)

   1. a.

   2. b.

Levin & Sells (to appear) indeed mention the German sentences in 40, cited from McIntyre (2007:354), but they mainly discuss the question of why the particles on and in in English cannot be unpredicated par-ticles, in contrast with off and out, as shown in 41.

1. (40)

   1. a.

   2. b.

1. (41)

   1. a.

   2. b.

According to Levin & Sells, unpredicated particle constructions like 39 should be analyzed in parallel with resultative constructions with a subcategorized DP. In connection with the general observation that resultative constructions describe events with a definite endpoint, Hay et al. (1999) and Wechsler (2005) point out that the adjectives in resultatives with subcategorized objects are restricted to the ones with a maximal endpoint closed-scale, such as flat or clean. In contrast, open-scale adjectives such as long or wide and minimal endpoint closed-scale adjectives such as dirty or wet cannot head the result phrase. We thus find contrasts like hammer the metal flat/*long or wipe the shelf clean/*wet. Maximal endpoint closed-scale adjectives are associated with a closed scale with a maximal value and can be used only if this maximal value is reached. Therefore they can delimit the event described in the resultatives temporally. In contrast, open-scale adjectives like long do not have a maximal or minimal value, and the value of this type of adjectives is always context dependent. Minimal endpoint closed scale adjectives like dirty have a scale with a minimal value, thus this type of adjective can be used as soon as an entity minimally has the property in question. For example, a towel can be dirty as soon as there is a speck of dust on it. However, in normal cases this minimal standard is ignored and a more reasonable contextual standard takes priority. This type of adjective is thus de facto equivalent to open-scale adjectives. In short, both types of adjectives do not provide a definite endpoint and thus cannot head the result phrase.

Levin & Sells (to appear) argue that the particles off and out can be analyzed in parallel with maximal endpoint closed-scale adjectives like clean or flat, whereas the particles in and on are just like minimal endpoint closed-scale adjectives like wet or dirty.

1. (42)

   1. a.

   2. b.

The event described in 42a contains a scale defined by the amount of some material like dust. Such an event is over when the surface is clear, and thus the scale has a maximal endpoint. In contrast, the event denoted in 42b is not one of clearing, but one of covering a surface. According to Levin & Sells, the particle on could be applied as soon as a bit of material is placed on the surface. However, just like the adjectives wet or dirty in resultatives, the particles on and in do not provide a temporal boundary, since we normally neglect the minimal value and these particles de facto belong to the same class as the open-scale adjectives.

To sum up, the point of the analysis of the unpredicated particles in English by Levin & Sells (to appear) is twofold. First, they analyze unpredicated particle verbs in terms of resultatives with a subcategorized object. Second, they attribute the impossibility for the particles in and on to serve as unpredicated particles to the type of scale inherent in these particles. This analysis encounters the two problems, though.

First, Levin & Sells fail to mention verbs with unpredicated particles whose object is not subcategorized.

1. (43)

In order to accommodate the particle verb in 43, it is reasonable to assume that the direct object in all constructions with an unpredicated particle is somehow moved out of the prepositional phrase, as McIntyre (2007:356) suggests.

1. (44)

Second, it is clear that the explanation based on the scale type of the particles cannot be applied to German, since the particles ein and auf are semantically almost synonymous with in and on, respectively, and have nevertheless unpredicated usage, as we have seen in 37, repeated here as 45.

1. (45)

   1. a.

   2. b.

As far as a ground is realized as a direct object of the verb, as represented in 44, the German sentences in 45 are rather similar to the so-called applicative constructions like 46, as McIntyre (2001:296) points out.

1. (46)

   1. a.

   2. b.

We are now confronted with the following questions. First, what is the difference between applicative constructions like 46 and constructions with unpredicated particles ein ‘in’ und auf ‘on’ like 45? Second, why can the German particles in and auf be used as unpredicated particles, whereas the corresponding particles in and on in English can not.

McIntyre (2001) extensively discusses unpredicated particles in German and provides examples with the particles ein ‘in’ and auf ‘on’ like 47. Other examples of this type are given in 48. For some of them corresponding expressions can be found in Dutch as well (49).

1. (47)

1. (48)

1. (49)

We notice here that the direct object in constructions with the unpredicated particles ein ‘in’ and auf ‘on’ must be three-dimensional.Footnote ¹¹ In addition, themes must be transferred to this three-dimensional object incrementally. That is, these constructions denote situations where themes like liquid, cream, or goods are continuously moved into a “container.” These restrictions are not found in applicative constructions. Whereas the prepositional object den Zettel ‘the paper’ in 50a and die Leine ‘the rope’ in 51a can be realized as the direct object in the applic-ative construction in 50b and 51b, respectively, these prepositional objects cannot be the direct object of the particle verb aufschreiben ‘write on’ in 50c and aufhängen ‘hang on’ in 51c respectively, since it is not possible to regard these objects as three-dimensional.

1. (50)

   1. a.

   2. b.

   3. c.

1. (51)

   1. a.

   2. b.

   3. c.

The difference between 52c and 53b can be treated in the same way. The prepositional object das Heft ‘the notebook’ in 52c cannot appear as the direct object of the particle verb einschmieren ‘smear in’, since it is not a three-dimensional container. However, it is easier to consider the entity das Gesicht ‘the face’ in 53b as three-dimensional, so that this entity may be a direct object of this particle verb.

1. (52)

   1. a.

   2. b.

   3. c.

1. (53)

   1. a.

   2. b.

Furthermore, the distribution of the particle auf ‘on’ can be explained in a similar manner. The sentence in 54 describes an event in which themes like boxes are piled up on the truck. The direct object der Wagen ‘the truck’ here is a three-dimensional entity. In contrast, the sentence in 55a denotes that a theme like water is moved to the surface of the laundry. The direct object die Wäsche ‘the laundry’ cannot be charac-terized as a container. However, if the particle auf ‘on’ is exchanged with ein ‘in’, which expresses a three-dimensional relation by nature, the resulting sentence is acceptable, as shown in 55b.

1. (54)

1. (55)

   1. a.

   2. b.

1. (56)

   1. a.

   2. b.

   3. c.

1. (57)

   1. a.

   2. b.

   3. c.

1. (58)

   1. a.

   2. b.

The unpredicated particles ein ‘in’ and auf ‘on’ serve to telicize the described events, just like the result phrases. Then, the object in the constructions with these particles must be a container, because two-dimensional entities such as a piece of paper (see 50c) or a rope (see 51c) cannot store an incremental theme and thus cannot guarantee the telicity.

Next we come back to the following difference between German and English.

1. (59)

   1. a.

   2. b.

As we have seen, the explanation by Levin & Sells based on the type of scale cannot be applied to German. We rather pay attention to the referential property of the implicit ground of the particles in question in both languages. To begin with, we consider the sentences in 60 and 61. Notice that the particles here are predicated of the figure.

1. (60)

   1. a.

   2. b.

1. (61)

In the case of the particle verb pour in, it must be clarified what the drinks are poured into. Without a preceding context, the sentence in 60a is thus not appropriate. By contrast, the sentence in 60b is felicitous, since the goal of the drinks is contextually given and thus explicated referentially. Now notice that the German particle verb eingieβen ‘pour in’ is free from such a contextual restriction. The sentence in 61 can be used even if we do not know into what the beer is poured. Note here that in German the deictic prefix hin ‘away from the speaker’ or her ‘toward the speaker’ is added to the particle if the semantic object of the particle is contextually given. In 62 the prefix hin unambiguously refers to the glass that is mentioned in the preceding sentence. In other words, the prepositional element with the prefix hin or her refers to a particular token of a path and is thus fully referential, whereas the ground of the particle in German is non-referential, it is just a type (see McIntyre 2001:261ff. and Zeller 2001:515ff. for further discussion).

1. (62)

The difference between eingieβen and pour in with respect to the referentiality of the suppressed ground can be demonstrated by the following pair as well.

1. (63)

   1. a.

   2. b.

McIntyre (2001) points out that in 63a there can be more than one empty glass, since the ground of the particle ein is non-referential; it is just a type. In contrast, the sentence sequence in 63b shows that in the case of pour in there must be only a single specific glass, since the ground of the particle in is referential.

It is now obvious that the referential status of the suppressed ground of the English particle verb pour in is different from that of the German particle verb eingieβen and that the particle verb pour in corresponds to the German complex verb hineingieβen . Zeller (2001a,b), following Riemsdijk (1990), proposes that PPs are selected by a functional head F⁰, as represented in 64a, and that this functional head can be filled with postpositions in German, which consist of a deictic prefix hin or her and a prepositional element. The functional head F⁰ can select not only a PP, as shown in 64b, but also a DP, as in 64c. It can occur even without a PP or a DP, as in 64d.

1. (64)

   1. a.

   2. b.

   3. c.

   4. d.

The important difference between particles and regular locational prepositions, including postpositions, is that the former lack a functional structure. The German particle verb aufsteigen ‘climb up’, for example, is represented as in 65, according to Zeller (2001a,b).

1. (65)

What is significant here is that verbs with a postposition headed by hin or her do not permit the ground of the particle to be realized as a direct object of the verb, as McIntyre (2001:295ff.) clearly shows. Zeller (2001a,b) argues that this is attributed to the functional structure that resides in the postposition.

1. (66)

   1. a.

   2. b.

   3. c.

We are now in a position to conclude that the difference between 59a and 59b, repeated here as 67, is due to the fact that the English particle verb pour in contains a functional structure, since this particle verb corresponds to the German verb hineingieβen, whereas the German particle verb eingieβen lacks it. The most important point is that the functional head is responsible for assigning case (see Koopman 2000 and Zeller 2001a,b). If the ground of the particle ein ‘in’ in 67a, namely das Glas ‘the glass’, is merged in the structure of the particle, it must then move to a position where this DP receives structural case, resulting in 67a, as Zeller (2001b:513f.) suggests. In contrast, there is no reason for the ground of the particle in to move, since the particle verb pour in is provided with a functional structure. In addition, the referential differ-ence of the suppressed ground between the two particle verbs can be explained from this perspective. Zeller (2001a:135ff., 2001b:515ff.) argues that a prepositional phrase is referential only if it contains a functional structure, just as a noun phrase is referential only if it is provided with a functional head D⁰ (see also Longobardi 1994).

1. (67)

   1. a.

   2. b.

The contrast between 68a and 68b is explained in the same way. If the sentence in 69a is used felicitously, the ground of the particles in and on must be established referentially, which suggests that the English particle verbs load in /on have a functional structure and thus the ground cannot move. The German particle verbs einladen ‘load in’ and aufladen ‘load on’ in 69b can be used even if the ground of the particle is not established contextually. That is, the suppressed ground in 69b is not referential. If the ground such as der Wagen ‘the truck’ appears in the structure of the particle ein ‘in’ and auf ‘on’, this DP is not assigned case and must move to the direct object position, resulting in 68b.

1. (68)

   1. a.

   2. b.

1. (69)

   1. a.

   2. b.

The referential contrast of the implicit ground with load in and einladen ‘load in’ can be also demonstrated by the following sentence sequences. The ground in the former must be a single specific vehicle.

1. (70)

   1. a.

   2. b.

Here one might raise the question of how the difference between 71a and 71b can be explained. Note that the sentence in 71c is used felicitously even if the ground of the particle in is not mentioned in the preceding sentence. One might then consider the ground not to be referential and expect that the sentence in 71a would be acceptable, just like the German counterpart 71b.

1. (71)

   1. a.

   2. b.

   3. c.

   4. d.

However, the ground in 71c necessarily refers to a particular body part of the subject. In contrast, the ground in 71d need not refer to a single token. Thus, the sequence in 72a is consistent, whereas the sequence in 72b is contradictory.

1. (72)

   1. a.

   2. b.

As we have seen, Levin & Sells (to appear) argue that the scale type inherent in the English particles in and on is responsible for the fact that these particles do not show unpredicated usage. However, we argue that this follows from the syntactic structure of the particle verbs in question, that is, they involve a functional projection. Therefore, the implicit ground of the particles on and in in 73, which Levin & Sells mentioned, must be referential. Such a restriction with respect to the reference of the ground is not placed on particle verbs like wipe off and rinse out. This difference is demonstrated again by the contrast between 74 and 75.

1. (73)

   1. a.

   2. b.

1. (74)

   1. a.

   2. b.

1. (75)

   1. a.

   2. b.

As we have already seen, the Dutch particle verbs ingieten ‘pour in’, inladen ‘load in’, and opladen ‘load on’ have unpredicated usage as well.

1. (76)

   1. a.

   2. b.

   3. c.

It is assumed that the implicit ground of these particle verbs is not referential. Therefore, the following sentence sequences are consistent.

1. (77)

   1. a.

   2. b.

   3. c.

It follows that the particle verbs in Dutch ingieten ‘pour in’, inladen ‘load in’, and opladen ‘load on’ lack a functional projection and thus the ground is allowed to be promoted to the direct object position, just like the German counterparts.

In this section I have discussed unpredicated particles in German and English. I have shown that in German verbs with the unpredicated particle ein ‘in’ and auf ‘on’ the object must be a three-dimensional container, because the particles telicize the event described by the verb and only a container can fulfill this function. Furthermore, I have argued that the functional structure inherent in the English particle verbs pour in and load on is responsible for the fact that these particles in and on do not show unpredicated usage, in contrast with the corresponding verbs in German and Dutch.

5. Conclusion

In this paper I discussed the following three topics with respect to German particle verbs with a ground argument, which is defined as a semantic object of the particle. First, I compared activity verbs prefixed with an like anlächeln ‘smile at’ and those with zu like zulächeln ‘smile at’. At first glance, we might have the impression that the object in both particle verbs represents a ground, but the formation of these particle verbs is different. Whereas the accusative argument in the former is introduced by the particle an and thus represents a ground, the dative argument in the latter is licensed by inalienable semantics. The ground representing a body part is suppressed and bound by the possessor dative. Second, I discussed German motion verbs prefixed with zu like zueilen ‘hurry toward’, in which a ground can be realized as a dative argument. I considered the question of why the preposition naar or op is always necessary in the corresponding verbs in Dutch, whereas in German such a preposition is not necessarily required. I argued that this difference comes from the availability of the so-called low dative in German. Finally, I discussed a difference between English and German with respect to unpredicated particles in the sense of Levin & Sells (to appear). As opposed to them, I argued that the functional structure inherent in the English particle verbs pour in and load on is responsible for the fact that the particles in and on do not show unpredicated usage.