Hans-Christian Schmitz

Optimal Accentuation vs Focus Accentuation

The following examples, which were already given in the Introduction, show that the use conditions of a sentence can be influenced by accentuation: (1) a. Who did John introduce to Sue? — John introduced BILL to Sue. b. To whom did John introduce Bill? — John introduced Bill to SUE. The declarative sentences of the examples can be uttered as answers to the questions indicated. The declarative sentence of example (1-a) — with stress on “Bill” — however, is not an adequate answer to the question of (1-b), nor can the declarative sentence of example (1-b) — with stress on “Sue” — be used as an answer to the question of example (1-a). On the basis of the different stress patterns the sentences require different discourse contexts (here: questions under discussion); that is, the sentences have different use conditions. Within a theory of optimal accentuation, the difference is easily accounted for: the two stress patterns are optimal in different contexts. If a recipient knows the question under discussion, he needs only recognise the word “Bill” for understanding the answer of example (1-a), so that exactly this word must be accentuated. Analogously, in example (1-b), “Sue” needs to be accentuated, because a correct understanding requires that this word be recognised.

Introduction: Pragmatic and Semantic Effects of Accentuation

The following examples show that accentuation can change the conditions under which a sentence can be used: (1) a. Who did John introduce to Sue? — John introduced BILL to Sue. b. To whom did John introduce Bill? — John introduced Bill to SUE. The declarative sentences in the examples (1-a) and (1-b) can be used felicitously as answers to the questions preceding them. The senctence with stress on “Bill” (example (1-a)), however, is not a felicitous answer to the question of example (1-b), nor can the sentence with stress on “Sue” (example (1-b)) be used as a reply to the question of example (1-a). Due to the different stress patterns both sentences require different utterance contexts; they have different use conditions.

Reconstruction of Messages

Following the model developed here, information exchange through natural language functions as follows: a speaker sends a message by uttering a word or a sequence of words. The message denotes either a proposition or a question, regardless of whether the words form a complete sentence. A recipient recognises some of the uttered words; he might even recognise all of them. On the basis of which words he recognises, and on the basis of his knowledge of the discourse context, he tries to reconstruct the entire message as a sentence of QL. For this, he translates the recognised expressions to QL and, eventually, applies context-sensitive or context-insensitive operations of semantic enrichment. It may happen that by doing so the recipient can reconstruct several non-equivalent messages. Each of these messages can be evaluated on its contextual adequacy. The recipient possesses a representation of the discourse participants’ common ground. (That is, he makes assumptions about the common ground; these assumptions can eventually be false.) The criteria of adequacy describe how a reconstructed message should relate to this common ground representation. If a reconstructed message is considered inadequate, the recipient can either reject the reconstructed message (and possibly choose a different reconstruction), or he can accommodate his representation of the common ground, thereby “making” the message adequate. Both the possibility of reconstructing and the possibility of accommodating are restricted, so that a given utterance cannot be interpreted in any arbitrary way. Ultimately, a recipient can update his common ground representation with the reconstructed message.

Cooperative Information Exchange

According to the communication model described in the previous chapter, a message is encoded into a signal by the speaker and transmitted to the recipient. The recipient recognises (decodes) the signal. The recognition may be imperfect; nonetheless, the recipient may still be able to reconstruct and interpret the complete message.