Negative strengthening

↵ Back to class homepage

With the concept of strong implicatures and the "Competence Assumption" (i.e., the assumption that people have an opinion one way or the other), we can explain another type of implicatures. Let's start off with an example.

Once I was chatting with an acquaintance I didn't know very well, and this person asked me what I did that weekend, and I said that I had played badminton. She then asked me who I played with, and I said "some Japanese friends", and this person then said, "Oh, I don't like Japanese people." We had a big fight after that; I said that was a mean thing for her to say, and I don't remember exactly how she replied but she basically said something along the lines of, "What, I didn't say I dislike them, I just said I don't like them! I didn't say anything bad about them, I just said something neutral."

What exactly happened here? The speaker said "I don't like Japanese people", and I interpreted that as meaning that she dislikes Japanese people.

Here's another episode. In one episode of the TV show Better Call Saul, the police are searching for some incompetent criminals, Mr. and Mrs. Kettleman, who have recently disappeared. The main characters of the show, Jimmy and Kim, are discussing what to do (Jimmy is worried that he will get in big trouble if the Kettlemans aren't found, so he hopes the police will find them soon). Kim asks Jimmy what will happen if the police never find the Kettlemans. Jimmy replies, "The Kettlemans? You've met these people, right? They're not exactly masterminds. They will be caught, if the cops are looking for them." The sentence in bold shows a similar phenomenon as what we saw in the first example. Here, Jimmy only said "They're not exactly masterminds", but what he really means is that they are stupid.

These are both examples of a type of implicature called negative strengthening. In this module we'll look at how negative strengthening works, why it's relevant for our analysis of pragmatic meaning, and how to explain how it happens.

Negative strengthening and formal notation

Remember how in the previous module I introduced some confusing-looking notation, such as ¬BELR(P) to mean "Rebecca doesn't believe P" and BELRP) to mean "Rebecca believes not-P"?

Why did I torture you with that complicated notation? Negative strengthening is why! If I say something like "I don't think Mahmoud is coming to the party", in plain English, it's ambiguous: it might mean that I actually think he's not coming, or it might just mean "it is not true that I think he's coming" (which could be because I don't think he's coming, or because I just have no opinion). This is the same difference we saw between weak and strong implicatures. The bottom line is, plain language is ambiguous here, precisely because we have a tendency to re-interpret "I don't think Mahmoud is coming to the party" ¬BELI(Mahmoud is coming to the party) as meaning "I think Mahmoud is not coming to the party" ¬BELI(Mahmoud is not coming to the party). Using the fancy formal notation helps us show the two different meanings more explicitly.

How negative strengthening works

Negative strengthening happens with words whose meanings are "scalar". "like", from our first example, is a good example of this. We can imagine that the amount we like something in reality can fall anywhere on a scale between absolutely loving something and absolutely hating them (with "having no opinion" in the middle of the scale), as shown in the graph below.

The word like indicates some section of the scale. If I say I "like" something, what I literally mean is that my amount of liking something is on the upper part of that scale: at the very least, it's more than "no opinion".

So, if I say I "don't like" something, that literally means that the amount I like something is not on the "liking" part of the scale, it's somewhere else. I might dislike it, or I might just have no opinion. For instance, if I say "I don't like the DAB", that literally means that it cannot be true that I like them; I might dislike them or I might have no strong feeling one way or another, as shown in the figure below.

People often will interpret the sentence in a stronger and more specific way, however. They will "narrow" the potential meaning so that it only includes the "dislike" part of the scale: in other words, they will take "I don't like the DAB" to mean not just that I don't like them, but in fact that I dislike them. If I drew a circle around how much of the scale I might mean (as in the above figure), you can imagine that circle shrinking down to a narrow interpretation, as in the below figure:

As mentioned above, negative strengthening can happen with pretty much any kind of word that describes a location on some sort of gradient scale. We've seen that "I don't like X" can be taken to mean "I dislike X", but this works with many other verbs that express opinions. It also works with all kinds of adjectives, as long as they also express meanings on a gradient scale. For example, above we saw an example in which "The Kettlemans are not exactly masterminds" is interpreted as meaning that the Kettlemans are stupid (or, more accurately, that the speaker believes the Kettlemans are stupid). This could be explained in the exact same way as we explained the like examples above. "Smartness" is a scale (from totally stupid to super-genius mastermind); saying "The Kettlemans aren't masterminds" literally just means they're not masterminds but they might be anywhere lower on the scale (they might be stupid, they might be of average intelligence, they might even be pretty smart but not quite reading the level of masterminds). But we would often strengthen the meaning of this to mean not just that they're not masterminds, but actually they're stupid. (You can try drawing a graph of the "scale" of smartness, like we did above, and making a circle and "shrinking" it just like we did with the like example.)

An explanation of why and how negative strengthening happens

We can straightforwardly explain negative strengthening based on the "Competence Assumption" that we introduced in the last module, and which seemed to already be needed to explain those kinds of implicatures. Recall that we started out by seeing that an utterance like "Josh is smart" often implies that Josh is not brilliant, but then we further broke this down into several steps: first, there is an implicature that the speaker does not think Josh is brilliant (a weak implicature, based on the Gricean maxim of quantity), and then there is a further step that we assume the speaker has an opinion and therefore, if the speaker has an opinion and doesn't think Josh is brilliant, then the speaker does think Josh is not brilliant (a strong implicature, based on the Competence Assumption). That second part is an example of negative strengthening.

In other words, it seems that all that is needed to explain negative strengthening is the assumption that the speaker has an opinion (but see the discussion topics for a complication). For example, if a speaker says that they don't like X, that means that either the speaker dislikes X or they have no opinion. So if we believe the speaker has an opinion, then the only possibility that's left is that the speaker dislikes X. In this way, negative strengthening can be seen as just normal application of the "Competence Assumption". This also has an interesting consequence that instead of describing scalar implicatures (i.e., the interpretation of "Josh is smart" as meaning Josh is not brilliant) as consisting of a "weak" and a "strong" implicature, we could instead say that it consists of a quantity implicature that then gets subjected to negative strengthening.

(Note that, as is often the case in this class, there are other possible explanations available for the same phenomenon. In another module we are going to see how another theory explains negative strengthening as a sort of "I-implicature" [which is a concept we haven't learned about yet but will learn about in that module]. Alternatively, we could try to explain these implicatures using the maxim of relation: for instance, in the example about the Kettlemans, the idea that the Kettlemans are stupid is relevant to the broader point Jimmy is trying to make [the point is that they should be easy for the police to find] whereas the literal meaning that the Kettlemans are not-masterminds-but-could-be-anywhere-else-on-the-smartness-scale is not really relevant to the point.)

Negative strengthening vs. scalar implicatures

At first glance, negative strengthening may look very similar to the sorts of scalar implicatures we analyzed in the previous module. Both involve narrowing down the literal meaning into something more specific, e.g. from "I dislike or I have no opinion" down to just "I dislike" (in negative strengthening) or from "at least smart and possibly even brilliant" down to "just smart but not brilliant" (in scalar implicature).

There is, however, a crucial difference between them. A scalar implicature involves rejecting something that the speaker did not say: for example, the speaker said "Josh is smart" but chose not to say Josh is brilliant, therefore we infer that the speaker actually means Josh is not brilliant (or at least doesn't mean Josh is brilliant). On the other hand, negative strengthening involves accepting something that the speaker did not say: for example, the speaker said "I don't like the DAB" and chose not to say they dislike the DAB, but we infer that the speaker dislikes the DAB anyway!

So, despite the apparent similarity, negative strengthening and scalar implicatures are actually opposite: scalar implicature involves assuming that the speaker doesn't mean something they didn't say, whereas negative strengthening involves assuming that the speaker does mean something they didn't say.

"Neg-raising" and negative strengthening

A phenomenon very similar to negative strengthening is "neg-raising", also sometimes called "negative transportation". Neg-raising is when negation that is interpreted in an embedded clause ends up being pronounced in the matrix clause; for example, if I say "I don't think Mahmoud is coming" and actually mean that I think Mahmoud is not coming, then the negative word ("not") is interpreted within the embedded clause "Mahmoud is not coming" but actually pronounced in the matrix clause "I don't think...".

The idea behind "neg-raising" is that a negative word can "move" to another part of the sentence. The idea of movement within a sentence should be familiar if you've ever studied syntax: a word that's interpreted at one part of a sentence can end up moving to another. A famous case is question formation in English (and French and a bunch of other languages), where one of the ways we can turn a statement into a question is by moving certain verbs to the front of the sentence: e.g., "You are angry" can become a question "Are you are angry?" (here I have used the strikethrough text "are" to indicate where the "are" originally was in the sentence). The claim about "neg-raising", then, is that it works in the same way: the sentence has a "not" in an embedded clause, but that "not" syntactically moves up to the matrix clause.

It should be obvious, though, that so-called "neg-raising" could be explained as just another example of negative strengthening. If I don't think Mahmoud is coming to the party, there are two possible reasons why: maybe I think he's not coming, or maybe I just have no thoughts one way or the other. But if you apply the Competence Assumption and believe that I do have some thought about the issue, then the only possibility left is that I think he's not coming. For this reason, many researchers have questioned the syntactic view of neg-raising and argued that instead it's just more negative strengthening (for instance, Bartsch 1973; I don't know German but I'm pretty sure the title of that paper means "'Negative transportation' is not a real thing); see, however, Collins & Postal (2018) for a review of evidence for and against both views.

There is a subtle difference between "neg-raising" and negative strengthening, but to me it seems just surface-level. In "neg-raising" examples, we could explain the interpretation by moving the word "not" (e.g., "I do not think Mahmoud is coming" gets interpreted as "I think Mahmoud is not coming")—although, for the reasons discussed above, I'm not convinced that we should explain it that way. Negative strengthening examples, on the other hand, cannot be explained in this way (the interpretation of "I don't like the DAB" cannot be paraphrased as "I like not the DAB", that's not even grammatical English!). Personally I think this is a good piece of evidence that the Competence Assumption, which can explain both, is a better explanation than "neg-raising", which can only explain the "I don't think..." sort of examples but cannot explain the "I don't like..." sort of examples. (But that, of course, assumes that the Competence Assumption can explain both of these sorts of examples; we will question that claim in the discussion activities).

Video summary

In-class activities

Here's an article headline: "印度政府“不希望”官员参加达赖喇嘛纪念活动". Translate it into English, or ask a Chinese-speaking friend to help you translate it. Hopefully you will notice an interesting negative strengthening (or "neg-raising") phenomenon in it.

Personally, I would translate this as "The Indian government 'doesn't hope' its officials will participate in even commemorating the Dalai Lama". It's clear from the context and the article, though, that the government hopes its officials will not participate.

This example highlights an interesting cross-linguistic difference: it seems like which words are susceptible to "neg-raising" (or negative strengthening) differs across languages. Anecdotally, I feel like in Mandarin it's not weird to say "I don't hope X will happen" to mean "I hope X won't happen". In English, though, this would be very weird and awkward, and it would be much more natural to say "I hope X won't happen", without neg-raising; it seems like "neg-raising" doesn't happen with hope in English (even though we've seen it can happen with very similar verbs, like think and belive.

Have students discuss this. Any way of explaining why there is this difference between English and Chinese? Does it have to do with different grammars, different alternatives available in the language, or something else? (Are there any other verbs that behave differently between these languages?) And does this pose a challenge for pragmatic theory or for my claim that "neg-raising" is a form of negative strengthening and negative strengthening is a conversational implicature? (An important thing to note here is that conversational implicatures are supposed to be cross-linguistically universal: they follow from general principles of communication, so those principles should work in any language. Therefore, if something is a conversational implicature, then the same line of reasoning should produce the same implicature in other languages as well, unless there's some special reason for it not to [e.g., "smart" might not imply "not brilliant" in a language that doesn't have a word for "brilliant"]. So, if the logic that gets "neg-raising" in the Chinese equivalent of "hope" doesn't get neg-raising in the English equivalent of the same verb, that would be a problem.)

Another thing to discuss can be trying to brainstorm other examples (in any languages) where "neg-raising" or negative strengthening is different for equivalent expressions in two different languages.

In the module I said that all we need to explain negative strengthening is the Competence Assumption. But that was a bit of an oversimplification.

If I say I don't like X, the Competence Assumption alone is enough to get the interpretation that I dislike X. But remember that, as we saw in the module, if I say e.g. "Josh is not brilliant" I might actually mean Josh is stupid. Can the Competence Assumption alone get us there?

No, it cannot. If I say "Josh is not brilliant", I literally mean that he's not brilliant but he might be stupid or he might be average or he might be smart-but-not-brilliant; he can be anything other than brilliant. Maybe if you assume I think he's different-than-average (which is already a new assumption, separate from the Competence Assumption), you can rule out the average part, but how do you decide whether I think Josh is just smart or I think he's stupid?

I don't know the answer to this (but see, e.g., Gotzner et al., 2018, for a great discussion of how negative strengthening works). It does seem to me like there is some extra implicature going on here as well. Specifically, "Josh is stupid" doesn't seem to me like the most usual default interpretation of "Josh is not brilliant"; without any special context or special pronunciation, I can also easily imagine that this sentence could mean he's not brilliant but he's at least smart. To get the interpretation that Josh is stupid, I feel like there needs to be some additional context, or the speaker has to pronounce the sentence with special stress or intonation and/or special discourse markers added(like, "Wellll... Josh is not exactly brilliant, if you get my drift!"—we saw this with the example about the Kettlemans, where the speaker said "They're not exactly masterminds"), and once we have that we can more easily analyze this as a Manner implicature (an implicature based on the speaker's violating the maxim of Manner).

Nevertheless, I think this example does point out a big problem with the analysis I described in the module. Using the Competence Assumption to explain negative strengthening seems most straightforward for examples with epistemic verbs (verbs that describe an epistemic state, i.e., what a person thinks or believes about something) or verbs that otherwise express my feelings about something. In those cases, the Competence Assumption works relatively well because it rules out the possibility that I have no feeling one way or the other. On the other hand, it's not so easy to use the Competence Assumption to explain negative strengthening of adjectives. Even if we imagine that smartness only has three levels (stupid, average, and brilliant), then the Competence Assumption still cannot explain why "Josh is not brilliant" would implicate that Josh is stupid, because assuming I have an opinion does not rule out the middle level (it's possible that I have an opinion and my opinion is that Josh's intelligence is average!). This is fundamentally different from the state of epistemic verbs, where the middle level is ruled out by the Competence Assumption (e.g., I can believe X is true, believe X is false, or have no opinion one way or the other; but the Competence Assumption rules out the possibility that I have no opinion).

So, under that view, maybe the Competence Assumption is not enough, and some other mechanism is needed to explain how negative strengthening happens. (As mentioned in the module, our later module on the Q- and I-principles will discuss one possible explanation—although, to preview that module a bit, personally I don't think that explanation is satisfactory either, because it's more just stating what type of implicature it is rather than explaining just how that implicature happens.)

Have students discuss and figure out if they can explain how negative strengthening happens for adjectives where the Competence Assumption is not sufficient.


⟵ Weak and strong implicatures
Disjunction and ignorance ⟶

by Stephen Politzer-Ahles. Last modified on 2022-04-25. CC-BY-4.0.