Are scalar alternatives context dependent?

In the context of what I’m learning across my classes right now, these articles were fascinating. In examining differences in understanding languages, we are able to make cognitive models of inference and understanding. The article titled “Accessing the Unsaid” explores pragmatic inference on a lexical level. From “some” to “many” to “most” to “all,” language includes a scalar progression of quantifiers. When most adults hear one of these words, the word is usually understood to mean solely that word and not a quantifier of a different scale. However, each of these words can signify their own meaning or the meaning of any word at a higher scale (i.e. the word “some” can describe a situation where “all” of something occurs). Children tend to use this latter understanding of the words. In explaining why this difference exists, both articles actually propose that children fail to access the scalar alternatives that are negated for specificity, not that working memory or processing constraints are the limiting factor. This explanation supports certain cognitive models that help explain the phenomenon of human inductive reasoning. In SymSys 1, we are currently studying Bayesian Inference models. Through a Bayesian lens, we always want to maximize our probability of a hypothesis given an event, and so we tend to pick the most restricted hypothesis given that event. When adults hear the word “some,” a hypothesis that the speaker meant “only some” is the most restricted, and therefore the most probable. If children had access to alternative quantifiers, they too most likely would choose the most restricted meaning of a sentence.

With the Bayesian model in mind, I expected to see similar results for the studies of ad hoc implicatures that were studied in the other article. Real life examples would provide the alternative for the children and so they should be able to use cognitive reasoning that can be modeled. The results showed this. Even three-year olds were able to us ad hoc scales to clarify the referent of a logically ambiguous statement. Adults still outperformed children, however, and so pragmatic competence probably grows over time. I’m curious about the intentional choice to sometimes include scalar alternatives in intended meaning of adults’ speech. Okay, that’s probably confusing, but let me explain. Let’s say person A eats “all of the cookies.” Now let’s say person B says to person A, “I said you could only eat some of the cookies.” Person A may respond, “Well, I did eat some.” The original use of the word “some” was used and understood as “only some,” but then in person A’s response, it was used to signify “some or more.” Both people were aware of both meanings, but my question is how? When we as adults use the word “some” to mean “only some” most of the time, how do we know when an individual intended the broader meaning?