Published: April 22, 2022

Author: Peter Cooper
Nominator: Orin Hargraves
Course: LING 3430 Semantics, Fall 2021
LURA 2022

Consider the following sentences:

(1) a. Everything I don’t eat is food for the dog.
      b. All I don’t eat is food for the dog.

These sentences appear very similar on the surface, but have very different meanings. (1a) means the speaker gives the dog their leftovers, while (1b) means that the speaker is willing to eat anything but dog food. These constructions are labeled all-clefts and everything-copular clauses respectively, and are two different types of copular clauses which I hoped to investigate.

Copular clauses - clauses whose main verb (in English) is the copular verb be - have been the subject of much work over the decades. Relevant here is the taxonomy first developed by Higgins (1979) and further developed in subsequent years. This differentiates several different types of copular clause based on their syntactic and semantic behavior; the two most important classifications are predicational and specificational copular clauses. Following are examples of each:

(2) a. The person with the umbrella is walking alone. (Predicational)

b. What I don’t eat is food for the dog. (=The dog eats my leftovers, from Hedeburg (1993))

(3) a. The person with the umbrella is my sibling. (Specificational)

b. What I don’t eat is food for the dog. (=I don’t eat dog food, from Hedeburg (1993))

The semantic definition of the mentioned taxonomy, though, is somewhat vaguely-defined. The broad strokes are that predicational sentences further specify the pre-copular constituent, while specificational sentences denote some attribute of it. However, a more rigorous definition is still not agreed upon. Two proposed definitions are outlined below:

Mikkelsen (2005, 2011) and Geist (2007) use a definition which bases the taxonomy on the referentiality of a given sentence’s pre- and post-copular constituents – essentially, it considers whether each constituent is a definite noun or not. A predicational clause has a referential constituent at the start and a non-referential constituent at the end; a specificational clause has the opposite.

A more nuanced definition was given in Heller (2005), based on discriminability. An item is said to be more discriminable the closer it is to a specific, discrete object in space – In essence, this is just a sliding scale of referentiality. In this framework, specificational clauses increase in discriminability, while predicational clauses decrease.

I sought to compare and evaluate these definitions using my previous work in all-clefts – crucially, I had shown in this work what can be seen in (1): all-clefts must be specificational, while everything CCs must be predicational. These two constructions can therefore be used as an ‘anchor’ to specificational and predicational clauses, and by comparing the postcopular parts of these two constructions, we can compare the behavior of specificational and predicational clauses. I used a web corpus to gather examples of these constructions for comparison, as well as only-clefts, another similar construction which patterns with all-clefts, to ensure that specificational clauses do act similarly to each other, and the data is not random noise.

The data collected did not support Mikkelsen’s metric. It did hold up for predicational clauses – everything-CCs universally showed non-referential postcopular constituents; however, the purely binary definitions offered would predict that all-clefts must always have a referential postcopular constituent. This was directly refuted by examples such as (4), where the postcopular constituent (bracketed) was a full clause:

(4) All that means is [that they aren’t looking hard enough].

However, these results do show a relative trend towards Mikkelsen’s criteria. The specificational constructions analyzed, although they are far from entirely referential, do show a higher percent of referential and noun-phrase postcopular constituents than their predicational counterparts, which show a much higher proportion of adjectival (i.e. less referential) constituents. In fact, this is very close to Heller’s definition – the sliding scale of referentiality mentioned above. I argue that the data supports Heller’s definition over Mikkelsen’s and Geist’s.

As a side note from the main findings of the study, it can be seen in the table that there were two instances of adjectival all-clefts. These are shown in (5):

(5) a. All that he says is true; he does not speak empty and false words.
      b. All that he does is true and actual.

These read more like everything-CCs than all-clefts. These sentences were both drawn from a religious website, and seem archaic in nature. It is possible that the specificational use of all-clefts is a new phenomenon, and sentences like these are relics of the construction’s past usage. Hopefully, more research will be able to shed some light on this in the future.

Cooper, P. (2020) ‘The Syntax of All-Clefts and All-RC Constructions’, BA Thesis, University of Sheffield

Geist, L. (2007) ‘Predication and equation in copular sentences: Russian vs. English’ In: I. Comorovski, K. von Heusinger (eds.) Existence: Semantics and Syntax, Berlin: Springer

Hedeburg, N. (1993) ‘On the subject-predicate structure of pseudoclefts’, In: Eid, M. and G. Iverson (eds.) Principles and Prediction: The Analysis of Natural Language, Amsterdam: John Benjamins

Heller, D. (2005) ‘Identity and Information: Semantic and Pragmatic Aspects of Specificational Sentences’ PhD Thesis, Rutgers, The State University of New Jersey

Mikkelsen, L. (2005) Copular clauses: specification, predication, and equation, Amsterdam: John Benjamins

Mikkelsen, L. (2011) ‘Copular clauses’ In: Maienborn, C. K. von Heusinger, and P. Portner (eds.) Semantics: An International Handbook of Natural Language Meaning, 2 vols. Berlin: Mouton de Gruyter, pp. 1805-1829