## Alternatives and context

↵ Back to class homepage

As we saw in the previous module, any utterance has an infinite number of stronger alternatives, but only implies the denial of a few of them. How can we solve this problem?

First of all, it's clear that we only consider alternatives which are relevant to the context. So, uttering "That's a cat" doesn't imply anything about a stronger alternative like "That's a cat and I like ice cream" if "I like ice cream" is not relevant to the conversation. So alternatives like this, which involve juxtaposing two totally unrelated propositions, are not considered unless they're in a special context that makes them relevant (e.g., "I'm all out of bubblegum" can imply "I'm going to kick your ass" only when the context indicates that chewing bubblegum and kicking ass are related). This constraint can already rule out most of the possible alternatives.

Secondly, we probably don't consider alternatives that are substantially more complex or wordy than what the speaker said (see Geurts, chapter 6, for more discussion). Keep in mind that the logic of quantity implicatures goes like, "The speaker could have said X but didn't say X, so I guess that means the speaker doesn't believe X". This logic only holds if X is approximately as easy to say as whatever the speaker really did say. If X is substantially more complex, then the logic doesn't work anymore; when we ask ourselves "Why didn't the speaker say X?", the answer doesn't have to be "because the speaker doesn't believe X", it could instead just be "because X is too wordy or too much work to say." This can also help to explain why uttering a short utterance doesn't necessarily imply a denial of lots of much longer utterances.

This second constraint is a bit tricky, because sometimes we do make implicatures about longer utterances, as we saw in the previous module. It's not exactly clear how much longer or more complex an alternative needs to be before it stops being considered as an alternative; it's also not exactly clear when this complexity constraint does or does not apply. And, finally, it's not clear to me how we can measure complexity—surely mere number of words is not sufficient, because an utterance with a few long words may be more "complex" than one with many short words, and things like syntactic and semantic structure may be additional sources of complexity even when number of words is held constant.

For the rest of this module, we will examine some interesting ways that language and context influence which alternatives become available for implicatures.

### Availability of alternatives

One of the things that affects which alternatives we consider is our knowledge about which words are commonly used in a given situation. Geurts (chapters 2 and 6) discusses several interesting examples of this. For instance, in English, saying "I hurt my finger" seems to imply that the finger I hurt was not my thumb, but saying "I hurt my toe" does not seem to imply that the toe I hurt was not my big toe. (In other words, I can say "I hurt my toe" when I hurt my big toe, but it seems weird or misleading to say "I hurt my finger" when I actually hurt my thumb.) This pattern is probably due to the fact that in English we have a one-word expression for "thumb" and we don't have any common, colloquial one-word expression for "big toe". (There is a word hallux, but only specialists would ever use a word like that. It's certainly not thrown around in everyday conversation.)

We can see a similar example with family words. In English, if I say "Joe's brother is an architect", this usually would not imply that I don't know whether Joe's brother is older or younger. Because we don't have any word for "older brother" or "younger brother", it seems that we don't consider alternatives like "Joe's older brother is an architect"; these alternatives aren't relevant, so the original utterance doesn't raise any implicatures about them. In Chinese, however, if we used the word 兄弟 ("brother"), we might imply that we don't know if the brother is older or younger. This is because Chinese does have words for older brother (哥哥) and younger brother (弟弟), and in fact these words tend to be more common than the unspecified word for "brother". So, if someone uses that word, a listener would treat these as relevant alternatives, and would wonder why the speaker went out of their way to use a less common word when a semantically stronger and more common word was available; in this situation a listener may well infer that the speaker doesn't know if the brother is younger or older.

The last example from Geurts that we'll discuss is animals. If I say "I saw an animal outside last night", I probably imply that I'm not sure what kind of animal it was (maybe it was too dark to see clearly). This implicature happens because there are stronger alternatives available; for instance, if I had seen a dog, I could have just said "I saw a dog outside last night". This is the typical sort of quantity implicature we have been discussing. But now imagine I do say "I saw a dog outside last night"... does this imply that I don't know what kind of dog I saw? It doesn't seem to imply that (or, at least, this implicature seems less likely than the implicature from "I saw an animal" to "I don't know what kind of animal it was"). Why doesn't this implicature happen? The same logic we saw with animal and dog should work here as well; for example, if I knew that I had seen a pug outside, I could have said "I saw a pug outside" (pug is a stronger alternative than dog). It seems like we don't get this implicature, though, because pug is more specific than usually necessary for conversation. It seems like we have some default, prototypical level of specificity; in a "normal" context it seems we are expected to be at least specific enough to say if something is a dog, cat, fish, or whatever, but we don't necessarily need to say what breed of dog, what species of fish, etc.

This example also leads us towards examining the role of context, because these default expectations about specificity may be overriden by context. For example, at a dog show we do expect people to be more specific about breeds of dog, so if someone just says "I saw a dog" we might indeed be more likely to infer that they didn't get a good enough look to know what kind of dog it was.

### The difficulty of determining alternatives and contexts

You may have noticed by now that we've been quite vague about what constitutes "context" and what makes alternatives "relevant". While this may be a problem, we also should acknowledge that these things are not always accurately determined in real conversation, either. People do sometimes get the wrong implicatures, or fail to get the right implicatures, because they consider different alternatives than what the speaker intended.

A prominent example is discussion around the Black Lives Matter movement. Some people consider this slogan offensive because, they claim, they believe it implies that not all lives matter. This is a standard quantity implicature: "all lives matter" entails "Black lives matter" and is thus a stronger alternative.

That interpretation is misunderstanding the intended message, however, because "all lives matter" is not a relevant alternative. The slogan "Black lives matter" was not introduced in response to a question asking whether or not all lives matter; it was introduced in response to the question of whether or not Black lives matter. In other words, it was not uttered to deny the proposition "all lives matter"; it was uttered to deny the proposition "Black lives don't matter". (While people might not have been uttering "Black lives don't matter" explicitly, it is nevertheless clear that much of our society assumes Black lives don't matter, as evidenced by the way Black people are treated by police and legal systems in places like the US and Canada; see, e.g., books by Michelle Alexander and Robyn Maynard.)

By now, everyone and their grandmother has written explainers of this issue; white people explaining "Black Lives Matter" to other, even more clueless white people has become a super popular genre of online writing. Here's one by Taylor Jones and one by me. Anyway, the bottom line is, this is an example of a situatiton where context determines the relevance of alternatives; here a semantically stronger alternative (in terms of the entailment-based definition we saw in the previous module) is one that's not relevant to the context.

### In-class activities

One complication I did not mention with the "Black lives matter" example is that nowadays "all lives matter" also has an implicature strongly associated with it.

Many people who support the Black Lives Matter movement find the expression "all lives matter" offensive (particularly when it's uttered in response to someone saying "Black lives matter"). For example, in the time leading up to the 2016 US presidential election, presidential candidate Hillary Clinton got criticized for saying "all lives matter".

From a purely semantic perspective, this shouldn't happen. "All lives matter" entails "Black lives matter". So, the interpretation that people dislike must be something that comes from pragmatics, rather than semantics.

Discuss and figure out what the implicature commonly associated with "all lives matter" is and how it is derived.