Monday, May 18, 2020

A Remark on Identity Conditions in Nominal Ellipsis

This short squib was triggered by the following joke that I saw on Facebook (in the group “I Love Mathematics”):

Wife: Please could you go to the shop and get a carton of milk, if they have avocados get six.
Husband: [Returning from the shop with six cartons of milk]: They had avocados.

When the husband heard the phrase “get six”, there were two possible ways to resolve it: “get six avocados” or “get six cartons of milk”. Only the former makes pragmatic sense, since there is no reason why buying cartons of milk should be contingent on seeing avocados at the store. What is funny is that the husband misinterpreted the request to mean: “If they have avocados, get six cartons of milk.”

Although the husband’s interpretation does not make pragmatic sense, it seems to be OK syntactically. In other words, it is possible for “a carton of milk” to serve as the antecedent of an elided “six <cartons of milk>” (<…> shows elision). This fact can be seen more clearly in the following example:

John bought one carton of milk, and I bought six <cartons of milk>

The facts suggest the following generalization about nominal ellipsis:

A singular noun can be the antecedent of an elided plural noun.

Here I would simply like to point out that this fact does not fall easily under either a syntactic or semantic identity theory of ellipsis. Under a syntactic identity theory, the antecedent of an elided constituent must be identical to it syntactically. But in (1), carton of milk is not syntactically identical to cartons of milk.

Under a semantic identity theory, the antecedent of an elided constituent must be identical to it semantically. But certain the denotation of carton of milk (the set of cartons) is distinct from the denotation of cartons of milk (the set of sums of cartons).

So, (1) suggests that neither syntactic nor semantic identity theories are correct. In the remainder of this squib, I will outline several possible analyses that maintain some form of identity:

One possibility is that there is syntactic identity, but then the plural suffix is deleted independently. So, the partial representation of (1) would be (3):

John bought one carton of milk, and I bought six <carton>-s

In this representation, carton is elided, stranding the plural suffix -s. Now, the following assumption is crucial:

A stranded plural suffix does not cause ungrammaticality, rather it is deleted.

So, in (3), the plural suffix -s is also deleted, in addition to the noun carton. There are really two separate instances of deletion.

I do not know of any independent evidence for (4), and in fact, there have been proposals for constraints that seem to contradict (4) (Lasnik’s Stray Affix Filter). Minimally to maintain (4), all of Lasnik’s data would have to be reanalyzed. So, although I do not reject 4, and the account of (1) based on it, I will explore an alternative.

An alternative account of (1) is based on the concept of Ghosting in Collins and Postal 2012 (Imposters, MIT Press). The idea is that the antecedent of the deleted cartons is contained in a non-pronounced topic phrase such as as for cartons of milk. Under this account, the representation of (1) would be:

John bought one carton of milk, and <as for cartons of milk> I bought six <cartons>.

So, the topic phrase as for cartons of milk is ghosted, but the plural noun phrase in the topic serves as the antecedent of the elided phrase in “six <cartons of milk>”.

The problem with this account is that it then brings up another Pandora’s box: What licenses ghosting of null topics in examples like (5)?

Yet a third possible analysis is that the plural -s is actually semantically vacuous. Rather, the semantics of plurality are related uniquely to the numeral six. After all, in many languages numerals and plural marking do not cooccur. Under these assumptions, carton of milk in (1) would actually be semantically equivalent to cartons of milk, and deletion would be licensed under semantic identity.

This last theory will need to face the issue of how bare plurals are interpreted. In the following sentence, if -s is semantically vacuous, how is cartons of milk interpreted as plural:

I bought cartons of milk.

Perhaps in this case there is a null existential quantifier that plays the role of the numeral in licensing the semantically vacuous -s:

I bought SOME cartons of milk.

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.