There Are Data, However, That Indicate That Elements Do Sometimes Side Switch

Side Switching

1Is elbowing a universal strategy?

the big start/little star approach predicts that losers will always stay on the side of the host preferred by the highest ranked alignment constraint, though further from the host than the winners

2Hungarian pre-V

there are data, however, that indicate that elements do sometimes side switch

pre-V V* V pre-V

neg V pre-V* neg pre-V V* neg pre-V V

Foc V pre-V* Foc pre-V V* pre-V foc V

wh V pre-V* wh pre-V V* pre-V wh V

X V pre-V Y Z? X V Y pre-V Z?? X V Y Z pre-V

the pre-V prefers a pre-verbal position but loses this to a number of elements

when it loses, it prefers an immediately post-verbal position

the further away from the verb it gets, the worse it is

3Possible analysis

something must come in front of the verb (vFx)

the pre-V wants to follow the verb (pvFv)

but when nothing else comes in front of the verb, the pre-V must (vFxpvFv)

thus side switching is achieved due to conflicting requirements of different constraints

4Finnish focus

Finnish also has possible side switching data

S V O* other orders

ST V OS V OT* other orders

SF V OS V OF* other orders

S V A O* S V O A

S V O AF* S V AF O

XC V* V XC

XC S V* S XC V* XC V S

XC ST V* ST XC V* XC V ST

XC ST V* ST XC V* XC V ST

XC V SF* XC SF V* SF XC V

Finnish has a basic SVO word order: when S and O are unmarked they come either side of the verb

Simple topic and focus do not seem to change this

the focus prefers either to be first (when subject) or last (when object)

A contrastive element always comes first

a first contrastive element does not affect the topic subject, which comes before the verb

a contrastive element does affect the focus subject, which comes last in the presence of a contrast

5Analysis

there are a number of factors here that are hard to capture with simple alignments using the big star notation

the concepts of ‘first’ and ‘last’ are puzzling as they suggest big time losers rather than winners – but then why would this be a strong requirement? The contrast obviously competes for first position and wins!

Both object and non-subject focus prefer a last position (suggesting losing), but the focus wins (how can a loser win?)

Why would a subject, which prefers a pre-verbal position (with or without topic marking) prefer a last position when focus marked? – but only in the presence of a contrast!!

trying to ignore this, we might try the following

subject precedes verb

focus follows verb, but not when focus (sPvfFv)

contrast precedes verb and precedes subject (sPvcPv)

when subject is focussed and there is a contrast, focus follows verb is important (fFvsPv) – we thus derive a contradiction in ranking: fFvsPvfFv

we won’t really be able to solve this problem until we have introduced other concepts

6English adjectives

English adjectives show side switching phenomena

a proud father* a father proud

a very proud father* a father very proud

a father proud of his children* a proud of his children father

adjectives typically precede the noun

but they follow the noun when they have a post head element

7Discussion

again this is going to be difficult to account for – can’t do it the way we did for Hungarian pre-V as it doesn’t seem to be anything to do with the noun as to why the adjective precedes or follows it

so the adjective prefers a pre-noun position if it has no element following it and a post-noun position if it does (aPnaFn and aFnaPn = contradiction)

what appears to be going on is that the adjective likes a position next to the noun and it prefers a position in front of the noun. But when it has something following it, it can’t be both in front and next to the noun, so it switches sides and prefers the adjacent but following position

but such side switching is exactly what the big start notation predicts should be impossible

the question is how to accommodate a preference for elbowing and a preference for side switching in the same system?

8Analysis

suppose we replace the big star system with a more refined system

xP/Fyordering constraints that only evaluate order, not adjacency

these are categorical – violated once by each instance of a wrong order

xyz, xzy, xz..y all satisfy the constraint

yxz, yzx, yz...x all violate the constraint to the same extent

xAyadjacency constraint that only evaluated adjacency to host, not order

these are gradient constraints – violated to a more or less degree depending on how far the target is from the host

xy and yx are equivalent (not violated)

xzy and yzx are equivalent (one violation)

now we have the constraints

adjective wants to be adjacent to nounaAn

adjective wants to be in front of nounaPnaFn

complements follow headscFhcPh (= head initial)

complements are adjacent to headscAh

that the complement follow the head is more important than that the adjective precede the noun cFhaPn

that the adjective be adjacent to the noun is more important than that the adjective precedes the noun aAnaPn

the ranking we end up with is

cFhcAhaAnaPn > the rest

cFh / cAh / aAn / aPn
=>A N / *
N A / *!
cFh / cAh / aAn / aPn
A C N / *!
A N C / *!
C A N / *!
C N A / *! / * / *
=>N A C / *
N C A / *! / * / *

9How to get elbowing