Linguistic Variation - Volume 21, Issue 1, 2021

In some limited cases, English allows a particular preposition to combine with a certain kind of subordinate clause, as exemplified by in that in “I take the proposal seriously, in that I loathe it”. In contrast, Norwegian systematically allows prepositions to combine with subordinate clauses (as in Det resulterte i at vi vant, literally “It resulted in that we won”). I argue that the English case should be handled as the subcategorization for a certain complement class by a particular lexical entry, while the Norwegian case indicates that the extended projection of clauses can continue up to the preposition. This highlights an important difference between lexical selection and extended projection, revealing a hitherto underappreciated source of parametric variation, and sheds light on several properties of extended projections as well as of prepositions. Specifically, the extended projections of N and V may “converge” at P, challenging the notion of extended projection as being confined to a single lexical category.

Circumpositions in Afrikaans present several puzzles: (i) they always encode spatial paths, but spatial paths can also be encoded by prepositional phrases; (ii) they can be doubling or non-doubling, and (iii) they exhibit disharmonic word order of the kind that appears to violate the Final-over-Final Condition (FOFC). In this paper, I argue that circumpositions offer support for the existence of a directional head [dir] in the fine structure of the Afrikaans verbal domain, and that this head is lexicalised by adpositional material in circumpositional expressions. I show that Afrikaans grammar distinguishes Route-paths from Goal-/Source-paths, and argue that whereas [dir] selects a [pathP] in the structure underlying Goal-/Source-paths (circumpositional expressions), Route-paths (prepositional expressions) are ‘bare’ [pathP] structures. I argue that since circumpositions identify structural components in different Spellout Domains, double-insertion of adposition-like material is required to exhaustively lexicalise the structure, and the disharmonic word order is understood as a direct consequence of the fact that [dir] is located in Afrikaans’ head-final verbal, which addresses the concern arising around FOFC. Finally, given that the adpositions in circumpositional expressions are shown to occupy structural positions that are distinct from that of de-adpositional V-particles, the paper also addresses the structural relation between circumpositions and particle verbs in which adposition-like material lexicalises a resultative [res] node in the verbal domain.

This paper proposes a minimalist analysis of locative prepositions in Central Catalan, from a comparative perspective. Specifically, I claim that certain semantic and syntactic properties that are usually considered part of the field of the extended projection of PPs in cartographic approaches (categories like Place, Degree, K, and AxPart, for example) are in fact properties of the DP in the complement of a preposition. This claim takes the view that adpositions are a functional projection that relates two DPs, the Figure and the Ground, and not a lexical head that projects a functional domain on its own, as Ns, Vs or As (cf. den Dikken 2010; Koopman 2000). The final part of the paper proposes a model to account for the variation that locative prepositions exhibit across Romance languages following the Conjecture of Borer (1984) (known as the Borer-Chomsky Conjecture since Baker 2008). More precisely, I propose a model in which microparametric differences among Romance simple locative prepositions depend on the particular composition of features in p.

Based mostly on the Romance languages, we provide evidence for the conclusion that oblique adpositions involved in the encoding of location and direction do not contribute a specific, fixed spatial meaning. On the contrary, they are general relators, relating a complement to an event by establishing an inclusion relation between them. Locatives are specializations of the basic relational inclusion content. State-in, motion-to and motion-from interpretations depend on the interaction of these simple relators with the structure of the event. Specifically, the relator may attach at the level of the Result phrase (goal, motion-to) or at the level of the Cause layer (source, motion-from). Furthermore, the Romance languages provide evidence for differential encoding of non-animate vs. animate location, which we refer to as locative DOM, presenting and discussing various instances of it.

The focus of this paper is the possessive relation arising in several configurations between the complement of a locative preposition (u ‘at/by’ and k(o) ‘towards’ in Russian, bij ‘by’ in Dutch and la ‘to/at’ in Romanian, henceforth, u-preposition, heading a u-PP) and another NP in the same clause. I will show that u-PPs can introduce a number of distinct possessive relations in function of the syntactic context and that languages differ subtly in the matter of which such relations are available in which contexts. I will attribute this variation to the different semantic domains of these possessive PPs (locus-modifiers as opposed to event-modifiers) arising from the lexical specification of the possessive relators lexicalized by these prepositions.

This paper deals with Talmy’s (2000) typological distinction between satellite- and verb-framed systems by comparing the expression of negative meaning through Source prefixes in Latin and Spanish complex verbs. In particular, the claim is made that the different scope relations established between the Source prefixes and the verb root in each language are the reflection of their different typological nature. The core proposal is that Latin Source prefixes lexicalize a Path head that defines a phase, whereas the Path head lexicalized by the Spanish Source prefix is not phase-defining. This has consequences on the timing of Spell-Out as well as on the position in which roots are merged, which naturally accounts for the distinct lexicalization patterns shown by these prefixed constructions in both languages. The negative meaning of Source prefixes, in turn, is derived from the context in which they are embedded.

Pazar Laz (pl) has a rich set of spatial prefixes, which encode path information as separate morphemes. However, unlike satellite-framed languages, pl spatial prefixes are incompatible with manner of motion verbs, which do not inherently encode path information, but they can only be used with directed motion verbs where the path information is available in the lexical semantics of the verb. Given this pattern, complementing Acedo-Matellán (2016)’s classification of weak vs. strong satellite-framed languages, we argue that verb-framed languages are also of two types. While languages like Romance are strong verb-framed languages encoding Path directly in the verb root, pl constitutes an example of a weak verb-framed language, where Path is affixal and forms a single word with the verb. We argue that spatial prefixes spell out the underassociated Path projection of the verb à la Ramchand (2008b).