In the last several Network Grammar pieces we’ve simply assumed up to be a particle. It can of course be a preposition instead. Indeed in some sentences both readings seem to be possible. This week I try to make some sense of the situation.
The analyses in LanguidSlog 27 that say sentences (20a) and (20b) fail must be wrong because proximity to Poppaea or Olivia means up is treated as a preposition. Sentence (103) shows this, using nicer semantics.
(103) Nero is giving Olivia up Vesuvius
Before Vesuvius is encountered, up can be treated as a particle. The junction giving__Vesuvius would be consistent with that but the resulting double-object sentence is ungrammatical, like (20b). The junction up__Vesuvius must apply instead, treating up as a preposition. However attachment of the up-phrase is uncertain.
We’ve never made any assumption that a nearer junction (Olivia__up) trumps a farther one (giving__up). If that were so, the interpretation of (103) would be that the gift is a slave-girl located on top of a volcano, rather than that the giving is enacted there. The latter interpretation seems more natural, so how NG deals with attachment of adjuncts needs to be addressed.
Again, up would be a particle in (104) if the sentence ended without Vesuvius.
(104) Nero is giving up Vesuvius
With the continuation, up can be preposition (Nero is dispensing charity at the rim of the volcano) or particle (he is surrendering a piece of his empire to someone else).
Sentence (105) shows that such ambiguity doesn’t necessarily arise in this type of construction.
(105) Nero is giving beneath Vesuvius
Following a give form, up can be a particle or a preposition, but beneath only the latter. The challenge is in combinations such as give…up. In (103), attachment of the up-phrase is uncertain, while in (104) the uncertainty is whether up is particle or preposition. The uncertainty persists even when the sentences are complete. In both sentences, before completion, the role of up may change as processing works from left to right.
We’ll now look at (103), starting with an analysis of the sentence without Vesuvius.
Nero is giving Olivia up
Until now LanguidSlog analyses have had up only as a particle, the PARTICLE / SACRIFICE proposition being immediately completed with six units of activation. But, as has just been shown, allowance must be made for up to be a preposition. The proposal is that the six units of activation are split between two alternative propositions pending consolidation on one or other.
(7b) Nero is giving Olivia up
The split is asymmetric. Then at sentence-end, with nothing else happening, the activation consolidates on the PARTICLE / SACRIFICE proposition.
Provisionally I’ve annotated the RELs differently for the split propositions from giving__up. Presentation would be improved if the two propositions clearly shared the same REL concept as well as the same QUO. However, LS25 proposed that, for a particle-derived proposition, REL should be simplified as PARTICLE with the SIC concept carrying all the semantic weight. This seems wrong for a preposition-derived proposition because the SIC concept is the complement of the preposition and the particular semantics of the preposition must be in the REL.
But perhaps one REL concept can be shared. It wouldn’t matter if it contained the locative semantics needed for preposition up because that is trumped by SACRIFICE when up turns out to be a particle.
(7b) Nero is giving Olivia up (revised)
Nero is giving Olivia up Vesuvius
If the sentence continues with Vesuvius, somehow the proposition with SACRIFICE as SIC must remain incomplete while the other is delivered to cognition.
(103) Nero is giving Olivia up Vesuvius
This analysis raises issues. First, the activation of preposition phrases. What is shown above is inconsistent with what has been seen for to– and by-phrases in earlier analyses. In those, the preposition brought all six units and its complement none at all. Clearly in (103), up as preposition can only bring part of the activation because the rest is needed for up as particle. The principle could be that a word brings six units of activation when it occurs as dependent (only once, that is). Then preposition phrases are over-supplied with activation but with no ill effects.
The second issue concerns the GIVE / UP / SACRIFICE proposition in (103). This remains incomplete but doesn’t cause ungrammaticality. I’ll assume this is because none of the words in the sentence is left unaccounted for.
Third, why is up the dependent of giving, not Olivia? I’ll look at this again in LS31, so the analysis above is provisional.
Still to come
We also need to analyse sentence (104) in a similar manner. That will be the subject of next week’s LanguidSlog piece because this one has already become rather long.