guests, and one must politely refuse twice before accepting, or if one refuses a third time, the host assumes that the guest really is full. Pickering, who grew up in a different culture, where food was much less central to the expression of caring, would perceive repeated offerings of seconds to be borderline badgering of guests. Conversely, if someone who shares Attardo’s cultural background is offered food by someone from Pickering’s background, he/she would politely refuse, only to see the plate being withdrawn.
This is why all these components of context, from presuppositions to shared cultural background, are crucial in understanding and negotiating communication. Note that even the understanding of what is happening, of what activity the participants are engaging in, is affected by their cultural backgrounds: when a speaker refuses an offer of seconds, is she being polite or is she full? There is no way of knowing, unless you know what her cultural background is (and even then, she may be accommodating to the different culture in which she is currently operating). The idea of context is much broader than this short discussion allows. See also Section 8.3 for more detail.
Context is also important from another perspective. Traditionally, the assumption has been that semantics deals with meaning per se and that pragmatics deals with meaning in context. While this is a good first approximation, we will see in what follows that the dichotomy cannot be maintained.
1.1.4 The Semantics/Pragmatics Boundary
As just mentioned earlier, one approach has been to argue that semantics is truth-functional, whereas pragmatics would deal with non-truth-functional aspects of meaning, as famously expressed in the formula:
Pragmatics = meaning - truth conditions
This definition has the advantage that it is neat: semantics would be responsible for the literal meaning of a sentence, whereas pragmatics would be responsible for any “extra” meaning. Consider the following example: The horse is not in the barn. Setting aside the fact that the notion of literal meaning has itself been seriously challenged, we could describe the literal meaning of the sentence as The location of the horse is not the barn. Put it differently, the horse may be anywhere but not in the barn.
What could the pragmatics of the utterance be? It depends on the context: suppose that there are horse thieves roaming in the area. Uttering The horse is not in the barn could imply that it has been stolen. If the hearer is a person whose job was to make sure that the horse was put in the barn, the utterance may be a threat of punishment. If the speakers are spies or agents of the secret service, this could mean that a subject has not reached their destination.
The problem, of course, is that the “literal” meaning of The horse is not in the barn depends, in part, on the disambiguation of its components. Let’s start with “is.” “Is” has many meanings; contrast
(3) This is Paul.
The book is yellow.
The book is 5 dollars.
The book is 300 pages long.
It’s two hours to Dallas.
Paul is home/in the barn.
In these examples, “is” expresses identity, quality, cost, length, duration, location, and many more concepts. In fact, the verb “to be” does not have much meaning (unlike a verb like “kick”) and thus must rely heavily on context to disambiguate it. In the case of our example, which of these meanings applies to “being in the barn”? Probably the location one (compare the example: Paul is home). But of course, disambiguation is, at least in part, a pragmatic process. Then there is the problem of the determinative article: it is “the horse” not “a horse,” which means it is a horse known to the speakers or such that it can be uniquely identified. Needless to say, “known to the speakers” is a pragmatic notion. So the idea of a neat boundary between semantics and pragmatics seems unlikely.
1.1.5 Modularity
Another way of arguing about the boundary between semantics and pragmatics is to resort to modularity arguments. As a teacher, you want your student to “think” in the L2; however, to do so they cannot refer to a rule and figure out what to say based on that, nor can they monitor what they say word by word. All those strategies are way too slow to produce fluent output. The learners must rely on automatic processing. Cognitive psychology has introduced the idea of “fast” and “slow” mental processes. In order to understand the difference between fast and slow processing, we must understand modularity.
Modularity is the idea that some mental processes are “encapsulated,” that is, they are based on an input and do not take into consideration anything else. Consider reflexes, for example, blinking when something approaches your eye does not depend on the will of the subject. You cannot will yourself into not blinking. Therefore, we can assume that the blinking reflex is a module, which takes as input the proximity to the eye of an object and is encapsulated, that is, does not consider other inputs, such as the will of the subject not to blink.
Visual illusions are likewise examples of modularity: consider the Müller-Lyer figures in Figure 1.7. Even after you have been told that the two horizontal lines are of the same length, you cannot help seeing them longer and shorter. This tells us the perception of the length of the lines is a module (i.e., is encapsulated), vis-à-vis the knowledge of their length.
Figure 1.7 The Müller-Lyer figures.
Within language, we know that all the meanings of a word get activated when it is first read, albeit for a very short time (about 200 milliseconds, well below the level of consciousness). Later the non-contextually relevant meanings are suppressed, but the activation of the irrelevant meanings cannot be stopped; hence it is a module. While modularity has much to contribute to cognitive science, as a theory of the boundary between semantics and pragmatics, it is a nonstarter (Récanati, 2010, pp. 1–6).
Some theories, for example, Relevance Theory, would see semantics as a code, since it is part of the language system and hence a module, since a code can only consider the information encoded within it. Conversely, pragmatics is the interplay of the semantic code, with contextual information, logical inferences, and pragmatic principles such as Relevance. However, leaving aside Relevance Theory, we know that disambiguation and determining the truth of an utterance rely on pragmatic factors, hence semantics, which deals with disambiguation (i.e., finding the meaning of the sentence) cannot be modular vis-à-vis pragmatics.
Furthermore, even the concepts themselves that are expressed by words are sensitive to context. Consider the word cut: we can define cut as sever, but it is clear that cut in cut the grass is rather different from cut the paper and cut his/her finger: first, in cut the grass and cut the paper, the severing is complete: at the end of the action, the two parts are no longer attached. In the cutting of the finger, we would need to specify that he/she cut his/her finger off, to achieve full severing. Second, in the cutting of the grass, the use of a lawnmower (or a scythe) requires a largely horizontal motion, whereas paper and finger cutting can happen irrespective of their orientation. So there are “sub-concepts,” restricted forms of cutting (completely vs. partially; horizontally vs. omni-directionally). These are called “ad hoc concepts” and are activated by pragmatics (context).
Grammaticalization and Pragmatics
Some authors have tried to distinguish
