Some truisms regarding linguistic communication:

1: The content of an assertion is a function of both the sentence actually uttered and its context. 

2: The context is not under the speaker’s direct and total control.

3: The content of an utterance is attributed to its speaker in a way that makes them responsible of their truth and effects.

Many philosophers (Langton, Farías, Sorensen, García Carpintero, Montemayor, Richards, myself, etc.) have noticed that there is a tension between these three facts.

The common reply as to why this is not a problem is:

4: The speaker is responsible of what they say because she is responsible of knowing how their words will be interpreted in the actual context where they are speaking.

This in turn has produced at least two broad replies:

First problem: Semantic Luck

5: Sometimes, the speaker cannot know all the aspects of their context that affects the content of its utterance.

Since ought implies can, this means that, at least in these cases, thesis (3) must be rejected.

I have written extensively on the subject so this time I want to talk about a different problem:

Second problem: Silencing

Thesis (4) is at least misleading for it suggests that:

6: For every content P and every context C there is a sentence S such that uttering S in context C would have content P.

But (6) is false. In certain cases, linguistic resources prove insufficient—a limitation aligned with Fricker’s framework of hermeneutical injustice. In other cases, contextual constraints delimit the semantic content of expressions, thereby excluding specific expressive possibilities for speakers.

A couple of examples:

A case where the material aspect of the context constrains what can be said in that context:

From the literature on semantic luck, we know that demonstratives are a simple and straightforward case where context-dependence generates problems like these. Demonstratives are useful because they allow us to refer to objects for which we have no name or identifying description. But, of course, they only work when the object is present and can be made salient in a relatively easy way. Right now, I can use “this” to talk about my computer if my interlocutor is standing a couple of meters in front of me, but I cannot use it to refer to the seventeenth line in the wood grain of the table on which my computer rests.

Negatively, this means that one cannot talk about an object unless (i) it is currently present and possibly salient, (ii) there is an available proper name for it, or (iii) there is an identifying description. Notice that (i), (ii), and (iii) may not be in the speaker’s power. As a matter of fact, they are normally not.

A case where the non-material aspect of the context – what Stalnaker has called the “common ground” – constrains what can be said in that context:

A usual example of sarcasm is when someone says, “What lovely weather we’re having!” when it is pouring rain. But what if the person actually wanted to affirm that the weather was lovely? Well, she could not in that context. In Langston’s terms, the context is silencing her opinion on the weather. Even if she added that she meant it, saying something like “I mean it”, it would most likely still be interpreted as sarcasm. (Weiner 2006) Eventually, communicating this message would require substantial changes in the context.

Thus, we have three ways in which contextual circumstances beyond our control can and do limit the sort of things we cna say:

1, When we do not have the necessary vocabulary available

2. When the material context excludes certain possible interpretations of the words we do have available.

2. When the common ground excludes certain possible interpretations of the words we do have available.