Functions of Language - Volume 16, Issue 1, 2009

This article takes up two closely connected theoretical unclarities: the questions of what a linguistic category is, and what it means for linguistic elements to have grammatical rather than lexical status. The two issues are discussed with reference to the case of evidentiality. The point of departure is the problem of how to define a category: Definitions of categories like evidentiality tend to mix up the question of what unifies the category with distinctions between pragmatic and semantic meaning, between lexical and grammatical coding, and between main predication and ancillary modification. In this article, we try to demonstrate the drawbacks of a conception of evidentiality as a strictly semantic, grammatical or modification phenomenon. We argue that evidentiality should be understood basically as a functional-conceptual substance domain. Against this background, we reconsider the distinction between grammatical and lexical status. We present a functional account of grammaticalization according to which grammatical status arises when a meaning is coded as being secondary information. By separating the definition of functional-conceptual substance domains from the distinction between grammatical and lexical status, we try to show how a function-based approach can make the characteristics of categories and of grammatical status cohere without getting mixed up.

This paper attempts to explain the terminological and conceptual confusion of evidentiality and epistemic modality. It presents a functionally oriented semantic analysis which does not belong to a specific theoretical framework. It shows that the alleged epistemic speaker commitment of evidential expressions does not come from the specific evidential value or mode of information, but rather boils down to the speaker’s and hearer’s interpretation of the source of information. A source of information can be attributed different degrees of reliability, but these should not automatically be translated into degrees of epistemic speaker commitment. The latter involves an evaluation of the likelihood, which is quite different from the evaluation of the reliability of the source of information. Thus, the account presented here challenges previous analyses in which the labels “epistemic” and “evidential” are applied to linguistic expressions either in too broad a way or in too exclusive a way. The analysis also contrasts with accounts based on the “inclusion” or the “overlap” of the two categories. Finally, the paper also discusses Nuyts’ (2004) claim that a clause can only have one qualification at a time.

The paper is an empirical, corpus-based analysis of the functions of seem with different complementation structures and in impersonal constructions as well as in combinations with an experiencer. The focus is on describing how evidentiality (e.g. inferencing/observation or hearsay evidence) interacts with degrees of certainty or factivity of the construction. Other features are subjectivity and intersubjectivity (shared or non-shared information). Translations from English into Swedish have been used to get a more fine-grained picture of the degree of certainty expressed by seem and its relationship to the type of evidence. The different constructions with seem and their functions are explained in a grammaticalization perspective. Seem conforms to a number of principles such as layering and divergence (Hopper). Seem to is the most grammaticalized form and can be regarded as an evidential marker which is similar to the modal auxiliaries.

This paper examines the developmental relation of the use and understanding of Modern Greek evidential lexical items with varying degrees of metarepresentational ability. Drawing on early natural language production of two 3;6–5;10-year-olds, the paper shows that children’s use of evidentials follows a three-stage metarepresentational development. Comprehension experimental tests suggest a restricted linguistic ability and a reliable reasoning ability to interpret others’ evidential state of mind. McNemar tests which were run for the evidential linguistic pairs tested (Experiment 1) and for the evidential reasoning trials administered (Experiment 2) revealed a higher likelihood for age to affect linguistic development during 4;0–5;10 years of age, than for age to affect cognitive development of evidentiality during 3;0–5;11 years of age. The discrepancy between children’s fragile understanding of lexical evidentials and more robust recognition of consistent/conflicting evidential states is attributed to their immature metarepresentational abilities.