The secret nominal life of Afrikaans intransitive adpositions

Languages like Afrikaans are sometimes said to feature a class of adpositions that can be described as “intransitive” (e.g. binne “inside”, bo “upstairs”, agter “in the back”, buite “outside”, onder “downstairs” voor “in the front”). It is argued that such elements in fact instantiate a “hybrid” category – the locative noun category – which is not missing a Ground external argument but is actually lexicalising the nominal structure associated with that Ground argument itself. Such locative nouns pattern with R-pronouns and home-class nouns and therefore probably share a similar internal structure with these better-studied elements of language.

Though members of the binne-class alone supply Afrikaans with its inventory of intransitive spatial adpositions, each member of this class has a transitive counterpart. 2That is, each optionally takes a DPGROUND complement (3). 33) Jan sit binne / agter / voor / buite die konservatorium.Jan sits inside/ behind/ in front of / outside the conservatory "Jan is sitting inside/ behind/ in front of/ outside the conservatory." [AFRIKAANS] I argue that the difference between the "intransitive" incarnation of binne-class adpositions, illustrated in (1), and the transitive incarnation, illustrated in (3), involves a micro-categorial distinction, and that binne-class adpositions are syncretic (=systematically homophonous) between these two micro-categories. 4The micro-category instantiated by the transitive adposition is the better-studied locative preposition (PLOC), whereas that instantiated by the "intransitive" adposition is a "hybrid" category, namely locative noun.I argue that locative nouns are importantly different from "ordinary" referential nouns in that they carry deictic information.This information is structurally encoded by a node AXPART (axial part; Svenonius 2006a).Locative nouns are also importantly different from "ordinary" PLOC-expressing adpositions in that they themselves encode the Ground argument.By implication, then, "intransitive" binne-class adpositions are not underlyingly intransitive, nor are they adpositions in the typical sense.

Jan sit [op die stoep] / [buite] / [daar].
Jan sits on the porch / outside / there "Jan is sitting one the porch / outside / there." (5) Jan draf [om die dam] / # [buite /daar].Jan jogs around the dam / outside / there "Jan is jogging around the dam / outside/there."(Where "outside/there" is locative only) (6) Jan draf buite / daar om.Jan jogs outside / there around "Jan is jogging (all the way) around the outside.""Jan is jogging (all the way) around there (=the perimeter of that space)." The fact that the locative noun buite and the R-pronoun daar felicitously substitute (i) the (DPGROUND-containing) PP in (4) as well as (ii) the DPGROUND of the directional adposition om in (5) evidences a noun-like status.
A further distributional property of R-pronouns (in Dutch, and also in Afrikaans), is that they must surface to the left of the selecting P, even when that P is otherwise exclusively prepositional (van Riemsdijk 1978).Locative nouns also share this distributional property.Consider for instance that in the "full" PP om die dam in (5), the DPGROUND die dam is to the right of the selecting adposition om; in (6) where the locative noun buite and the R-pronoun daar substitute the DPGROUND, both surface to the left of the adposition.
Finally, locative nouns and R-pronouns pattern the same in circumpositional phrases: (7a) shows the route-circumposition met…langs with a regular DPGROUND; in (7b) the DPGROUND is substituted with a locative noun and an R-pronoun.
we stroll always with the river along "We always stroll along the river."

(b)
Ons wandel altyd binne / hier langs.we stroll always inside / here along "We always stroll along the inside / here As (7b) shows, the preposition met is omitted when the DPGROUND of a circumposition is a locative noun or R-pronoun.Home-class nouns (Collins 2007), which in Afrikaans include place names (e.g.Bellville) and "highly-frequented" spaces (e.g.huis, "home"), share this distributional property.For instance, contrast (7a) above with the expression containing the home-class noun in ( 8) and see that it patterns with the locative noun and R-pronoun-containing expressions in (7b).
we drive always Bellville / ground-road along "We always drive via Bellvile / along the dirt road."binne binne

Analysis
An important component of the analysis is that it should capture, in a systematic way, the syncretism of binne-class adpositions; that is, their tendency to function as either "ordinary" locative adpositions (recall (3)) or locative nouns.I claim an adequate understanding of this phenomenon requires an interface perspective; that is, one on the interaction between syntactic structure and lexical material.I propose the late-insertion analysis detailed in (9-10), where (9) represents the "shape" of binne-class lexical entries, (10a) represents the structure these elements lexicalise when functioning as locative nouns, and (10b) the structure they lexicalise when functioning as locative adpositions.In the diagrams, PLOC represents a place-denoting node typically lexicalised in Afrikaans by locative adpositions; AXPART (axial part; Svenonius 2006a) represents a space that is projected around some part of the Ground and is the locus of deictic information.More detailed discussion of axial part follows below.The lexical entry in (9) should be read as follows: there is a lexical entry for a linguistic element binne, which is pronounced /bǝnǝ/, which is associated with the encyclopaedic content "INTERIOR", and which qualifies for insertion into a contiguous (sub)set of the hierarchical formal sequence N*<AXPART<PLOC, where N* is a lexical noun which denotes a PLACE and which is "deficient" in the sense that it lacks the ability to project functional structure associated with the category D. 7,8 The conceptual underpinnings of this sequence are important: I follow 7 In this sense N* could equated with (a projection like) PLACE in the work of Katz and Postal (1964) and Kayne (2004).Cf. the discussion below for a short elaboration. 8It is not possible to include discussion of the technical details surrounding the mechanisms governing matching and insertion, or the larger theoretical assumptions implicit in the analysis.Briefly, however, matching is governed by the Superset Principle (Caha 2007), and insertion takes place according to the Elsewhere Condition (Kiparsky 1973).Furthermore, the presence of a D layer in structures underlying expressions with "full" DPGROUND arguments like (10b) prevents the entry in (9) from lexicalising the N in such expressions.Cf.Pretorius 2017 for detailed e.g.Ramchand and Svenonius (2014) in adopting a system where the formal features comprising the functional spine are syntactico-semantic in nature.These are probably drawn into the linguistic system during language learning from a conceptual system outside of the language faculty.The fixed order 9 in which these features combine reflects their conceptual origin: as each instance of merge must preserve compositionality, features cannot be merged in a sequence that turns out a conceptually non-valid structure.I take AXPART to represent a typeshift from the region occupied by an object to a vector space projected around (a part of) that object, and PLOC to identify a point within that space. 10I argue here that the AXPART node is also a locus of spatial deixis, which is sensitive to speaker perspective. 11 The insertion scenarios in (10) should be read as follows: As regards (10a), there is a derivation in which syntax merges AXPART with a "decifient" noun N*, and binne is inserted to lexicalise [AXPART [N*]].This insertion site makes binne a locative noun.
As regards (10b), there is a derivation in which syntax merges AXPART with a DP and subsequently merges PLOC with AXPARTP, and binne lexicalises AXPART and PLOC.DP is lexicalised as a "regular" DPGROUND, probably as part of a distinct phase.Whereas the N*P in (10a) constitutes a "defective" phase, allowing one and the same element binne to lexicalise structure that would otherwise fall on two sides of a phase boundary, the DP in ( 10b) is a nondefective phase.In line with Kayne (2004), the defectiveness of the N*P in (10a) could be understood to follow from the nature of a "silent" lexical noun PLACE embedded in such structure (cf.e.g. ( 11)).Nouns like PLACE are unable to identify the extended structure associated non-defective (phasal) DPs.
The structure proposed in (10a) for binne-class locative nouns is in line with that proposed by c for locative adverbials/R-pronouns in English (here, there, where; (11)), which is built upon by Kayne (2004).It is also in line with what Collins (2007) proposes for home-class nouns.
discussion and implementation of these mechanisms in a working system modelling Spellout.The architecture of the grammar is in line with that assumed by Nanosyntax, although the analysis is not carried out in that framework. 9Ramchand and Svenonius (2014) argue that the order of the functional sequence is not rigid across the board. 10Cf.Zwarts and Winter (2000), Kracht (2002;2008) and Svenonius (2006a,b;2008) for discussion and formal definitions.Note that although there is consensus around the idea of (a) syntactic node(s) denoting the projection of vector space in the structure of locative expressions, there is still debate around the precise nature and position of such (a) node(s). 11As a brief illustration of how this deictic component functions, consider the expression there is a lizard on top of the hat.The axial part in this expression is top which differs from the nominal the top in the sense that the latter is inherent to the Ground object (i.e. the intrinsic top of the hat), whereas the former relies on speaker perspective (i.e. the part of the hat facing upward relative to the speaker's perspective).In the case of an upturned hat, the lizard may in fact be on the hat's intrinsic bottom.Cf.Svenonius (2006b) for discussion.
(11) (Katz and Postal 1964;from Pantcheva 2008: 307) In line with what Svenonius (2006a) proposes about the nature of axial parts, the AXPART node in (9-10) could be considered equivalent to the locus of proximal/distal information in Katz and Postal's (1964) structure, i.e. the Definite complex in spec-Noun (this/that).In locative nouns, this "deictic" node is endowed with richer encyclopaedic content, but the function is the same: a mapping from the region occupied by the Ground object to vectors in space which, crucially, takes speaker perspective into account (cf.fn. 9 for illustration).It is this deictic quality of AXPART especially which makes the proposed analysis of locative nouns amenable to existing accounts of R-pronouns and home-class nouns.The next section continues with a short discussion of the relevance of this deictic quality, and the position of the AXPART node in the functional hierarchy.

Axial Part, Speaker Perspective, and Category Hybridity
The deictic quality of the AXPART node mentioned above ties in with Biberauer, van Heukelum and Duke's (this volume) argument that there is a Speaker/Hearer-perspective node/zone at the edge of every phase.If Biberauer et al's argument is right, we expect to find Speaker/Hearerperspective related structure "on the cusp" between what might be considered the "N" and the "P" zones/phases of syntax.Arguably that is precisely where much cross-linguistic research on AXPART locates this node.12Furthermore, being "on the cusp", we might expect AXPART to be lexicalised by "nominal" lexical material in some languages/contexts, and by "adpositional" material in others.I have presented some evidence in this paper that AXPART in Afrikaans is lexicalised by prototypically "P" lexical material. 13Svenonius (2006a) originally shows, however, that axial parts in English are syncretic with nouns.Such "on the cusp" structural positions therefore seem to play a crucial role in category hybridity in and across languages, and may also turn out to be target positions for Speaker/Hearer perspectives in syntactic structure, and influence the route that grammaticalization takes in different languages.

Conclusion
This short paper underscored the fact that binne-class adpositions function as members of distinct micro-categories, i.e. "regular" locative adpositions as well as locative nouns (the socalled "intransitive" adpositions).It was shown that these binne-class locative nouns, which are relatively understudied, exhibit much of the same syntactic behaviour as R-pronouns and homeclass nouns, which are better understood.On a par with what has been argued for the internal structure of R-pronouns and home-class nouns, binne-class locative nouns were analysed as a "hybrid" N-P category incorporating a Ground argument and a deictic component (AXPART).
The paper also proposed a "framework" for understanding the dual category membership of the binne-class.Namely, it is useful to conceptualise category as an interface phenomenon arising as a result of how lexical material is mapped onto syntactic structure.Finally, attention was drawn to the fact that the node AXPART in languages like English is associated with typically nominal lexical material, whereas it is associated with typically adposititional material in (languages like) Afrikaans.