…in which I try to convince you that I was actually wrong in convincing you that groups are atomic. Yeah, sorry about that.
When I first conceived the idea of writing about my research in a way understandable to non-semanticists, I imagined I’d write an attractive new update every two months or so, sharing all my interesting recent ideas in fascinating ways and making everyone who read it wish they had my job. Then reality happened, in which very few of my interesting ideas still make sense after a month or so, and I’m usually too lazy to write blog posts about them anyway (I already wrote a book about them, which was exhausting enough as it is).
But with said book as good as finished, and some extra time for frivolities on the side due to several well-timed weeks of maternity leave (which I have mostly been spending on the couch, tied down by little Roanne’s completely unpredictable feeding and crying schedules), I can finally give you the promised update on British English. Enjoy!
In Part 1 of this post, I pointed out that the two sentences in (1) and (2) have different meanings in spite of appearing very similar:
- The linguists are walking or cycling.
- The group (of linguists) is walking or cycling.
Imagine a situation in which half of the linguists are walking while the other half are cycling. Most speakers of English would say that sentence (1) is true in this situation, but sentence (2) is false: sentence (2) is only true if either all the linguists in the group are walking, or they are all cycling. I explained this by claiming that the group of linguists refers to an atomic entity, which means that the formal semantics cannot access the ‘parts’ of the group when calculating a meaning for (2); as a consequence, anything that the sentence says about the group of linguists must be true of the entire group. In (1), on the other hand, the plural the linguists refers to a set of entities. A set has parts that are accessible to our formal interpretation mechanisms, which means that it can be ‘taken apart’ by a semantic mechanism known as quantification in order to say something about the individual members of the set – namely, that each of them is walking or cycling.
So, the contrast between (1) and (2) confirms that plurals like the linguists and groups like the group of linguists refer to a different kind of semantic object, despite being intuitively quite similar in meaning. But does it really? One might point out that the grammatical number of the subject is not the only variable distinguishing (1) from (2): another difference is the fact that the predicate is plural in (1), but singular in (2). Could it be possible that it is this distinction, rather than the plural/group contrast, that is actually responsible for the meaning difference?
This may seem like a question that is impossible to answer, because English usually requires the subject and the verb to agree in grammatical number, which makes it hard to look at the role of subject number and predicate number separately. Fortunately, there is an exception to this: in British English, singular human group NPs like the group of linguists are allowed to appear with a plural predicate, as in (3):
- The group of linguists are walking or cycling.
And sure enough, once you replace the singular verb with a plural verb in this way, the sentence becomes identical in meaning to the one in (1) (I checked this with a whole bunch of British English speakers). In other words, if the verb is singular, the group subject is treated as a single atomic entity, but if the verb is plural, that same subject is treated as if it were a ‘real’ plural, referring to a set of entities.
So, what does a noun phrase like the group of linguists (or the team, the class, the choir, my family, etc) refer to? When the same expression can have different interpretations depending on the formal context (I am not talking about words with different meanings – for example, the word nail as used by either a carpenter or a manicurist, which are essentially just two different words that happen to sound the same – but about the same word playing different ‘mathematical roles’ in the derivation of the meaning of the sentence as a whole. In part 1 of this post, for example, I mentioned the fact that the adjective huge can be applied to a person (as in “John is huge”), but it can also combine with a noun like idiot to form the meaning ‘someone with a huge degree of idiocy’. It’s clearly the same word, but with different semantic functions depending on the words that surround it.), semanticists like to treat one of these meanings as ‘basic’, and the other(s) as ‘derived’. If we follow this line of thinking, we could either say that the atomic interpretation of group NPs is basic and the set interpretation is derived (in other words, group NPs refer to atomic entities by default unless something in the context actively turns them into set-referring expressions), or, conversely, that the set interpretation is basic and the atomic interpretation derived. Concretely:
- Possibility 1: a group NP refers to an atomic entity, but if it appears with a plural verb, this atom is ‘broken up’ into a set of entities.
- Possibility 2: a group NP refers to a set of entities, but if it appears with a singular verb, this set is ‘fused’ into an atom.
In my dissertation, I take the second approach, because I think it is much prettier (my supervisor won’t allow me to use those words in my academic writing, but fortunately this is my personal blog where I’m completely free to rank formal theories according to their prettiness). One reason for this is that the story above does not apply to all collective expressions in British English, but only to those that refer to human (or at least animate) groups. Inanimate collectives like the stack of plates or the list cannot appear with a plural predicate and only ever receive an atomic interpretation. Possibility 2 provides us with an easy way to implement this difference: we simply treat inanimates like stack or list on a par with ordinary entity-referring nouns like cat or table – in other words, by treating them as a different kind of set-theoretical object than animate set-referring nouns like family or team. Because our semantic toolbox, under Possibility 2, lacks the ability to break up atomic entities, stack and list and cat can only refer to atoms and are not expected to trigger plural agreement. On the other hand, Possibility 1 offers us no simple way to distinguish inanimate from animate collectives, since they all refer to the same kind of set-theoretical object (an atomic entity). In order to guarantee formally that the latter can occur with a plural predicate and be interpreted as a set while the former cannot, we would have to build an entire new formal layer on top of our set-theoretical system, where animacy is somehow formally coded in such a way that it determines which atoms can be broken up into sets and which cannot. This is not impossible, of course, but it would require a lot of extra work that is not needed if we adopt Possibility 2.
So we end up with a formal theory according to which animate group NPs (like my family) are analysed as sets, while inanimate group NPs (like the stack of plates) are analysed as atoms. Of course, now we want to know why animacy seems to correlate so neatly with an NP’s ability to refer to a set and to occur with a plural VP. If all goes according to plan, I will address this question in the third and final research update!