Natural languages have a wide variety of ellipsis phenomena.
"Ellipsis" can be defined as that an element lacks its phonological contents while it has syntactic, semantic, or pragmatic contents.
You know, it is a difficult question: how and in which component does the element lose its phonological contents?
I think, along the line of minimalist spirit, we can not take a stance of assuming that ellipsis is a purely syntactic operation (of cource, you can "assume" ellipsis as a syntactic operation in addition to MERGE and AGREE).
n.b. I refer "elliptical operation" itself in this short article. Some elliptical constructions perhaps are derived by syntactic movements (e.g. ATB-movement analysis for English gapping).
I note some ways (possiblities) of ellipsis from the view point of Distributed Morphology (If you want to know details of the model of grammar in Distributed Morphology, see Embick and Noyer (2001)).
1. Operation before Lexical Insertion (after Spell-out)
In Distributed Morphology, it is assumed that lexical items which include their phonological features are inserted post-syntactically, that is, after spell-out. The component between spell-out and lexical insertion is called "Morphology", where there are not phonological features but (morpho)syntactic features although it is a part of PF.
So, strictly speaking, the following two ways are not phonological but PF-related.
1.1. Eliminating some part of nodes at Morphology
Embick and Noyer(2007) indicates that there is an operation which add a new node to the output of syntax at Morphology. The operation "Eliminating nodes" could be a counterpart of it. Losing nodes means losing a place which lexical insertion applies to, so we elide some phonological contents.
1.2. Not occuring lexical insertion
Wilder(1997) suggests this way. He regards a certain type of ellipsis as a result of not occuring lexical insertion to some part of the tree. A diffrence from the above way is that he does not assume any elliptical operation. However, he does not refer to the detail of the process.
Ellipsis which is a result of both ways should be sensitive to syntactic or morphocyntactic constituency.
n.b. So far, I have no idea about empirical evidences to distinguish the above two. They actually could be identical process of ellipsis.
2. Operation in the phonological component: Eliminating phonological features
This operation is applied to at least after lexical insertion. I guess number of researchers suppose this way when they say "ellipsis as a PF operation".
Ellipsis which is caused by this operation could be sensitive to phonological constituency: prosodic word, minor or majour phrase, intonational phrase, and so on.
Of course, there is no necessity that every kind of ellipsis should be caused by one of the above three.
- Embick, David and Rolf Noyer(2001) “Movement operations after syntax,” Linguistic Inquiry 32: 555-595.
- Embick, David and Rolf Noyer(2007) “Distributed Morphology and the Syntax/Morphology Interface,” The Oxford Handbook of Linguistic Interfaces. G. Ramchand and C. Reiss (eds.), Oxford University Press.
- Wilder, Chris(1997) “Some Properties of Ellipsis in Coordination,” Studies on Universal Grammar and Typological Variation. Artemis Alexiadou and T. Alan Hall (ed.), 59-107, John Benjamins Pub Co.