Case and PPs
5.5 What this analysis tells us about the case vs. adposition debatedebate
5.5.3 Integrating the proposal into the Path > Place decomposition
Given my decomposition of Hungarian PPs in (127)–(128) and the Path > Place decomposition of spatial expressions agreed on in the literature, the question that I would like to address now is how the two decompositions can come together into one functional sequence.
25Note also that if naked Ps are true adpositions, which is the view advocated in this thesis, then the phonological clue is even less reliable:t´ul‘beyond’ and´atare monosyllabic, too.
26E. Kiss (2002); Trommer (2008) and Sur´´ anyi (2009b) do not extend this claim to naked Ps, though.
27This approach nicely ties in with the fact that case markers have evolved from postpositions. Example (i), from theDeed of Foundation of the Abbey of Tihany(1055), one of the oldest written texts that contain Hungarian words, shows that today’s Sublative case marker used to be a postposition. The contemporary equivalent is given in (ii).
(i) Feheruuaru Feh´erv´ar
rea sublat
meneh going
hodu military
utu road
rea sublat
‘the road going to Feh´erv´ar’
(ii) a the
Feh´erv´ar-ra Feh´erv´ar-sublat
men˝o going
hadi military
´ ut-ra road-sublat
‘the road going to Feh´erv´ar’
In the present approach the change fromreato-racan be described in purely (morpho)phonological terms, without positing a categorial change. See Heged˝us (2010a) for a generative analysis of the postposition-to-case-marker cline.
(127) dressed Ps
At the beginning of the thesis I argued that four variables must be settled before a complete functional sequence of any extended projection can be set up: i) how the functional sequence is mapped onto the interpretive component (syntax-semantics mapping), ii) how the functional sequence is lexicalized, iii) how Agreement is represented in syntax and iv) how the functional sequence is linearized. I assumed that the syntax-semantics mapping proceeds in a seamless com-positional fashion: that particular interpretations are tied to particular projections in the structure, and that the meaning of a syntactic structure sheds light on its internal composition. That is, if a particular interpretationIis available in projection P r, and a certain structure has the meaning componentI, then that structure also must containP r. Conversely, if a certain structure does not haveI as a meaning component, then that structure does not containP r.
The projections Path and Place have very easily identifiable meaning contributions to the structure, therefore we will begin with identifying what serves as their spell-out in spatial dressed and naked Ps. Path and Place correspond to the node I labeled P in the foregoing discussion, that is, my P has an internal structure with two sublayers: Path over Place.
As I have already mentioned, the dressed vs. naked divide does not correlate with a semantic difference, and we find place, path, and other abstract Ps in both groups. Some examples are given in below.
‘to under the bridge’
naked P: Path
‘through a bridge ’ dressed P: abstract
‘together with the child’
For now I am going to focus on spatial P only, the more abstract adpositions will be taken up at the end of the chapter. Given the interpretive evidence, I will take both (129) and (130) to be PlacePs and both (131) and (132) to be PathPs (apart from the obvious difference of what kind of location or path is being referred to, there is no salient meaning difference between (129) and (130) on the one hand and (131) and (132) on the other, therefore it is not warranted to posit more structure for naked Ps than for dressed Ps or vice versa).
Combining my proposal with the Path > Place decomposition means that place denoting dressed Ps spell out K and Place, and path denoting dressed Ps spell out K, Place and Path.
(135) Place denoting dressed P
PlaceP KP
DP K
Place
dressed P
(136) Path denoting dressed P PathP PlaceP KP
DP K
Place Path
dressed P
In a similar fashion, place denoting naked Ps spell out only Place and path denoting naked Ps spell out Place and Path.28
(137) Place denoting naked P
PlaceP KP
DP K
case marker
Place naked P
(138) Path denoting naked P PathP PlaceP
KP
DP K
case marker
Place Path
naked P
According to the structures in (135) through (138), the Place node does not directly combine with the DP. Instead, this relationship is mediated by a projection I labeled K. The literature, in fact, is full of proposals that hypothesize a projection (or projections) between PlaceP and DP. Of the decompositions in (108) to (112), I would like to give prominence to the following two, which come with a very precise characterization of the semantic contribution of the various projections proposed.29
(139) Pdir > Pstat > Paxialpart > Ploc (Kracht, 2008) (140) DegP > PlaceP > AxialPartP > KP (Svenonius, 2008b) Given the obvious parallel between the structures, the semantics attributed to the projections of (139) and (140) are also very similar. Let us begin with Kracht’s projections and their semantic characterization. Simplifying matters a bit, his Ploc is applied to the DP and yields a region.
This region is compressed to a point and serves as the origin of the coordinate frame. Paxialpart
establishes the coordinate frame and Pstat(my Place) picks out a location. This yields a function
‘goodness of fit’, which tells for each coordinate point how well it fits. Pdir (my Path) describes a
change in the goodness of fit through time.
Let us now turn to Svenonius’ functional sequence. His K(ase) shifts objects to eigenplaces (the region occupied by the object, see Wunderlich, 1991). AxPart is a function from eigenspaces to regions that are normally subparts of the eigenspaces (front, interior, top, and so on). Place (which is identical to my Place) picks out spaces on the basis of the subparts established by AxPart. In front of, for instance, comes about when the subpart of the object’s eigenspace chosen by AxPart is the front part of the object, and Place picks out a space on the basis of this. DegP serves as the locus of degree modifiers.
In sum, it is common to both Kracht (2008) and Svenonius (2008b) that a Place denotation is built up gradually, via the mediation of two projections between DP and Place. The first one
28There is perhaps room for variation here, it is possible that all or some path denoting naked Ps spell out only Path and the case marker spells out both K and Place. For my purposes nothing important turns on this; the point is that naked Ps spell out the same higher bit of the structure that dressed Ps do.
29(140) has also been shown to be able to handle temporal and other abstract non-spatial PPs, to which we are going to return in Section 5.5.8.
of these yields a spatial region ( Ploc/K), and the second one (Paxialpart/AxPart) establishes an extended location on the basis of this region. A specific spatial location is not created before Place is merged into the structure.
In (135) to (138) I suggested that the functional sequence contains a part that is intermediate between D and Place and I labeled it K. We have now seen that my K has various analogues in the literature, and that it can be decomposed into two sublayers (at least): AxPart and K (using Svenonius’ labels). (135) to (138) therefore have the revised structure in (141) through (144).
(141) Place denoting dressed P PlaceP AxPartP KP
DP K
AxPart
Place
dressed P
(142) Path denoting dressed P PathP PlaceP AxPartP KP
DP K
AxPart
Place Path
dressed P
(143) Place denoting naked P
PlaceP AxPartP KP
DP K
AxPart
Place naked P
case marker
(144) Path denoting naked P PathP PlaceP AxPartP KP
DP K
AxPart
Place Path
naked P
case marker