Modeling incomplete knowledge in courtroom
Izabela Skoczeń
Jagiellonian University in Kraków, Poland
Abstract
The objective of our paper is to answer the question how and to what extent can speakers influence the belief content of their hearers. We take particular interest in courtroom situations. In 1975, in his groundbreaking article ‘Logic and conversation’ Paul Grice noticed that while uttering sentences in natural language we often want to convey more than just the amalgam of the meanings of words that we utter. (Grice, 1975) The question: ‘Could you pass me the salt?’ is not just an inquiry into the capabilities of the speaker. Rather, it is a polite request to pass over the right condiment. What is meant through an utterance but not quite said is called a conversational implicature. Quantifiers such as ‘some’ also convey implicatures. The lexical meaning of ‘some’ allows it to include ‘all’. However, when you hear ‘some of the apples are red’ you will infer that ‘not all of the apples are red’. This inference is based on the reasoning that if the speaker had wanted to convey ‘all’ he would have used the word ‘all’ instead of ‘some’. This is called a scalar implicature because the words are placed on a scale in terms of their strength. Using ‘some’ implicates that the speaker wants to convey neither a stronger word on the scale (all) nor a weaker word on the scale (few). Analogously, numerals in natural language convey scalar implicatures. For instance ‘three’ implicates ‘not more than three’ etc. (Horn, 2006) N. Goodman and A. Stuhlmüller have made an experiment to confirm a model that was supposed to predict the probability of implicature formation in contexts where speaker and hearer do not have full knowledge of the situation. (Goodman and Stuhlmüller, 2013) During the experiment the speaker saw for example three apples. However, he had information concerning the color of only two of the apples (red). The hearer was aware of the speaker’s state of knowledge. The utterance ‘some of the apples are red’ cancelled the implicature ‘not all’ since hearers inferred a similar probability of two and three apples being red. If the speaker saw two out of three apples and said ‘one of the apples is red’, then the hearer inferred a similar probability of one and two of the apples being red and an extremely low probability of three apples being red. Thus, the implicature ‘not more than one’ was cancelled. During the experiment various scenarios were used and the speaker had a varied access to objects and their features, while the hearer was aware of the speaker’s state of knowledge. The experiment confirmed a finegrained interaction between the state of knowledge about the world and pragmatic inference. It also provided a strong argument against modular theories of mind since the data showed that the language faculty and the inference about world state were strongly connected. However, there remained the question whether the effects were cross linguistic or rather confined to the English language.
In our paper we present the replication of the above-described experiment in the Polish language. We used the same scenarios and the same quantifiers and numerals as Goodman and Stuhlmüller. Namely, the Polish words ‘niektóre’ for ‘some’ and ‘wszystkie’ for ‘all’ as well as the numerals ‘jeden, dwa, trzy’ for ‘one, two, three’ were used. The data proved to be a replication of Goodman and Stuhlmüller’s experiment with some differences. In the paper, we investigate hypotheses that explain the differences.
The second aim of our paper is an attempt to set ground for a model of courtroom inference where the speaker has some knowledge of the world state but the hearer does not know what knowledge the speaker has. Thus, we aim at a slightly different model than the one made by Goodman and Stuhlmüller where the speaker and hearer shared a certain (sometimes incomplete) knowledge of the world state. Imagine the following situation: we have three suspects A, B, and C. We know that A and B were at the crime scene the day it was committed. The judge does not know whether C was at the crime scene. What should C’s defendant tell the judge? Should she say ‘some of the suspects were at the crime scene’ or rather ‘two of the suspects were at the crime scene’? Will the judge cancel the scalar implicatures with equal probability? Our aim is to work on a model that will answer these questions. We employ ideas from rational speech act theory (Bergen et al., 2016) (Frank and Goodman, 2012), game-theoretical pragmatics (Clark, 2012) (Franke, 2009) as well as gametheoretical legal reasoning (Załuski, 2013).