«1. Introduction In this paper, I suggest that the notion of semi-lexicality can be defined in morphological terms. I assume that a semi-lexical head ...»
Lexical particles, semi-lexical postpositions *
In this paper, I suggest that the notion of semi-lexicality can be
defined in morphological terms. I assume that a semi-lexical
head is “half lexical, half functional”, by which I mean that it is
a morphologically complex element that consists of a lexical
node and a functional suffix.
I elaborate this hypothesis through a detailed discussion of
the properties of postpositions and particles in German and Dutch. It can be shown that postpositions are functional heads that project functional phrases, whereas the projection of a particle is a lexical PP without functional structure. Although there are a number of differences between particle phrases and postpositional phrases that follow from this fact, particles and postpositions also show certain parallels that cast doubt on the claim that the functional status of postpositions mirrors that of other functional elements like tense affixes or determiners. Rather, I show that postpositions are not “genuine” functional heads, but semi-lexical elements. They are derived from lexical prepositions via suffixation of a zero-operator that alters the thematic properties of the P-element. Since this operator is a functional element, the derived postposition is a complex functional head.
However, it inherits the semantics of the lexical preposition from which it is derived. Therefore, postpositions are semilexical elements; they have the semantic content of a lexical element, but the categorial properties of a functional suffix.
The paper is organized as follows. In section 2, I discuss prepositional, postpositional and particle phrases, and I show in what respects particles and postpositions differ and how they behave similarly. Section 3 provides an analysis of the 2 Jochen Zeller morphological structure and the thematic properties of postpositions. From this analysis I conclude that postpositions are semilexical elements. In section 4, I use the results of the previous sections to compare postpositional phrases in Dutch and postand circumpositional phrases in German.
I first introduce some terminology and some notations. I treat local transitive prepositions as expressing relations between two (individual-type) entities. In (1), for example, the preposition in provides the information that the bread stands in
the IN-relation to the closet:
(1) The bread is in the closet I refer to the external argument of a preposition (the DP the bread in (1)) as the theme; its internal argument (the closet), I call the reference object. I assume that in examples like (1), the theme is generated in SpecVP and moves to SpecIP, whereas the reference object is the complement of the preposition and receives case from this lexical head.
I follow Wunderlich and Herweg (1991), Stiebels (1996),
Olsen (1999), and others in expressing the semantics of prepositions through the general form in (2):
According to (2), transitive prepositions are represented as twoplace functions. The first argument corresponds to the reference object, the second argument corresponds to the theme. The argument structure of a preposition is expressed through the lambda operators that bind the respective variables inside the formula. The predicates in the formula provide the information that the location of the theme intersects with a particular region R of the reference object y. The specific properties of R are defined by the respective preposition. The preposition in, for
example, is represented as in (3) (cf. Olsen 1999: 117):
Lexical particles, semi-lexical postpositions 3
(3) represents the situation in which the referent of the xargument is located in the internal region of the referent of y.
Notice that (3) is the locative variant of in. The directional
version is represented as in (4):
The predicate FIN maps the location of x onto a final part of a path traversed by x. The directional preposition in provides the information that the final part of this path ends in the interior region of y.
2. Prepositional phrases and functional structure
2.1. The extended projection of prepositions Van Riemsdijk (1990, 1998) and Koopman (1993) suggest that PPs are selected by functional prepositional heads that license certain lexical and grammatical properties of the lexical preposition. In terms of Grimshaw’s (1991) theory, this means that prepositions, like the lexical categories N and V, have extended projections. I adopt this idea and assume with van Riemsdijk (1990) that the structure of prepositional phrases in German looks as in (5) (cf. Zeller 1999): 1
As (5) shows, the maximal projection of P° is the complement of a functional prepositional head that I simply label F°Prep (it corresponds to p° in van Riemsdijk (1990)). In constructions
with transitive or intransitive prepositions, F°Prep is phonologically unrealized:
In (6), the PP is headed by the transitive preposition auf that takes its internal argument as a complement to its right. In (7), the PP consists solely of its head, the intransitive preposition oben.
F°Prep may also host lexical material. Following van Riemsdijk (1990, 1998), I assume that in circumpositional phrases like those in (8), F°Prep is realized by a postposition, printed in
(9) [FP [F' [ PP auf [DP den Berg]] hinauf ]] As (8) shows, German has both complex and simple postpositions. Whereas the postposition durch in (8a) is homophonous with the preposition durch, the postpositions hinauf and heraus in (8b) and (8c) consist of the prepositional elements auf/aus Lexical particles, semi-lexical postpositions 5 and a prefix her/hin (henceforth h-prefix) which has a deictic function (towards/away from the speaker). In the following, I will refer to these complex postpositions as h-postpositions;
phrases headed by h-postpositions always receive a directional interpretation. 2 Notice that in (8b) and (8c), the prepositional parts of the h-postpositions (auf in hinauf; aus in hinaus) are identical to the lexical heads of the respective PP-complements.
Constructions with this property are called pleonastic circumpositional phrases; their PP-constituents are refered to as “cognate” PPs (cf. Olsen 1996; McIntyre 1998).
Interestingly, h-postpositions may also appear without a
(12) [VP [DP ein- Tourist] [V' [FP hinauf ] schick- ]] As I will show in detail in section 3.2, the PP-complement of a postposition helps identify the internal argument of the prepositional relation expressed by the FPPrep. The h-postpositions in (10) and (11), however, are used intransitively. They occupy the F°Prep-position, but do not take PP-complements. Therefore, the internal argument of the relation expressed by the hpostpositions remains implicit; the constructions in (10) and (11) only provide information about the external arguments of hinauf and heraus, which are expressed by the DPs ein Tourist and der Hund (represented as derived subjects in (10) and as direct objects in (11)). For example, in (10b) and (11b), it is 6 Jochen Zeller clear that the dog moves out of something, but this something, the reference object, is not mentioned explicitly.
In the nominal domain, a functional head may occur without a lexical complement as well. Since Abney (1987), it is widely assumed that pronouns are realizations of a functional D°-head with no NP-complement (cf. e.g. Stowell 1991; Longobardi 1994). Analogously, we can analyze intransitive postpositions like those in (10) and (11) as prepositional proforms; they are F°Prep-elements without PP-complements.
Some h-postpositions also project postpositional phrases:
(13) Ein Tourist steigt den Berg hinauf a tourist climbs the mountain h-up/on Note first that postpositional structures like (13) must not be confused with examples like (11a). Although both sentences include a DP bearing accusative case, the DP einen Touristen in (11a) is the theme of the prepositional relation and is located outside the FPPrep (cf. (12)), whereas the DP den Berg in (13) identifies the internal argument of the relation expressed by hinauf and is hence generated inside the FPPrep. Evidence for
this latter assumption is provided by the following contrast:
(14) a. Den Berg hinauf steigt ein Tourist b. *?Einen Touristen hinauf schickt Johan In (14a), FPPrep-topicalization has moved the h-postposition and the accusative DP to SpecCP. In (14b), however, topicalization of the accusative DP and the h-postposition is ungrammatical, because the h-postposition is a proform, and the DP is the theme which is not generated inside the FPPrep. 3 It has been suggested that the DP in postpositional structures
like (14) is the complement of an empty preposition that projects the PP-complement of the h-postposition:
[FP [F' [ PP ∅ [DP den Berg]] hinauf ]] (15) Lexical particles, semi-lexical postpositions 7 Opinions differ with respect to the status of the empty preposition in (15). Van Riemsdijk (1990, 1998) argues that the head of the PP in (15) is the trace of the preposition auf that has moved to F°Prep where it combines with the h-prefix. In contrast, McIntyre (1998) suggests that the head of the PP is an independent element, a preposition without a phonological realization, but with a specific meaning. I will return to this issue in section 4 where I compare the two proposals in light of a discussion of postpositional structures in Dutch and circumpositional phrases in German. For now, it suffices to assume that the structure of postpositional phrases includes a PP whose head is not realized phonologically.
So far, I have only provided examples from German. However, the structure in (5) also accounts for the properties of pre-,
post-, and circumpositional phrases in Dutch:
(16) a. omdat Jan op de berg reed because J. up the mountain drove b. Hij is binnen/buiten/boven/beneden he is inside/outside/upstairs/downstairs c. Het vliegtuig is onder de brug door gevlogen the airplane is under the bridge through flown d. omdat Jan de berg op reed because J. the mountain up drove The transitive and the intransitive prepositions in (16a) and (16b) are heads of PPs selected by an empty F°Prep; in the circumpositional structure in (16c), the PP onder de brug is the complement of the postposition door, and in the postpositional phrase in (16d), the postposition op selects a PP-complement with an empty P°. (Notice that Dutch does not have complex postpositions that correspond to h-postpositions in German.) I will come back to postpositional structures like (16d) in section 4.
8 Jochen Zeller
In this section, I will briefly introduce the analysis of particle verbs that I develop in Zeller (1999). Based on proposals made by Koopman (1993) and Haiden (1997), I suggest that particles are represented as phrasal complements of their verbs that lack functional structure.
Thus, the prepositional particle verbs in (17) have a structure like (18):
Furthermore, I argue that particles do not undergo incorporation in order to combine with the verb. 4 Instead, I assume that the adjacency of a particle and the verb in verb-final clauses like (17) is the result of the structure of the VP in Dutch and German.
Since both languages are SOV, the PP-complement which includes the particle precedes the verb (see Zeller 1999, chapter 2, for detailed discussion):
Since particles do not move, the two terminal nodes that a particle verb consists of (i.e. P° and V°) do not form a syntactic word (a complex V°). However, the relation between the two Lexical particles, semi-lexical postpositions 9 nodes is nevertheless strictly local, due to the absence of functional structure.
I define the relevant local relation as “structural adjacency”:
(20) Structural adjacency A head X and the head Y of its complement YP are structurally adjacent.