How do different linguistic forms expressing generalizations influence reasoning performance: an experimental pragmatics perspective
Joanna Blochowiak
https://www.unige.ch/lettres/linguistique/en/collaborateurs/anciens-membres-du-corps-professoral/joanna-blochowiak
ABSTRACT
Reasoning is essential human ability consisting in drawing
conclusions based on premises. A main challenge for reasoners is to
arrive at conclusions which are _correct_, that is, logically valid
(such as MP – _Modus Ponens_). Indeed, it has been repeatedly shown that people accept logically invalid conclusions, endorsing conclusions coming from invalid inference schemas (such as AC – _Affirmation of Consequent_). Why is it so? One very influential proposal goes back to Geis and Zwicky (1971) who put forth the hypothesis of the perfection of conditionals, also known under the name of invited inferences, according to which the meaning of conditionals in natural language is regularly “perfected” to biconditionals via a process of invited inference.
The goal of this research, situated in the domain of experimental
pragmatics, is to investigate the pragmatic enrichment in relation to
conditionals by (i) comparing reasoning with different linguistic forms expressing generalizations (conditional, biconditional, generic), (ii) different types of content (abstract, realistic), and (iii) different styles of reasoning. Three research questions will be addressed: what is the meaning and the role of various linguistic formulations of the major premise, what is the mechanism underlying pragmatic enrichment during reasoning, and what is the cognitive status of generics. To investigate these research questions, three studies have been carried out: a self-paced reading experiment in English (Study 1), a self-paced reading experiment in French (Study 2), and an eye-tracking experiment in French (Study 3).
The results from Study 1 showed significant differences between
reasoning with abstract vs realistic materials and different reasoning
styles in abstract reasoning: those who systematically accept the
conclusion of an invalid argument (Endorsers) and those who
systematically reject it (Rejecters). In realistic type of reasoning, we
showed that pragmatic enrichment can be found for reasoning with a
subclass of realistic materials, namely causal conditional statements
with _few_ counterexamples.
Results from Studies 2 and 3 demonstrated that there is a difference
depending on the linguistic formulation of the major premise
(conditionals vs generics) in reading time and accuracy of
participants’ answers of valid vs invalid arguments. We also found
that in both types of reasoning participants split in two categories:
Endorsers and Rejecters. This difference in reasoning style has an
influence on reasoning performance as well.
