Publications and Research
Document Type
Article
Publication Date
2026
Abstract
What we hope to do is pose questions that, to our knowledge, have not received attention in the literature. The broad form of the question is all too familiar: how do speakers of a language—say, English—come to possess the grammatical knowledge that they demonstrate? The questions that guide this paper are narrow instances of this broad one, narrowed in three ways. First, they concern three particular languages: Japanese, English, and Russian. Second, they concern tenses specifically, and more specifically still, constructions containing embedded tenses. Third, they concern the semantics of tense. The learning challenge of interest to us is not merely that the learner must come to know, of certain structures, that they are grammatical or ungrammatical; the learner must come to know what meanings the relevant structures can and cannot have. For instance, one acquiring English must learn that certain instances of embedded past tense permit simultaneous readings, while one acquiring Japanese must learn that they do not. The evidence bearing on such knowledge must consist of a kind of pairing between sentences and the situations they are used to describe, and it is from such pairings (in tandem with their existing grammatical beliefs) that a learner must operate. Before we proceed, we must make it clear that our intended contribution is not specific to the semantics of tense. The issues that we explore are general to the acquisition of any piece of semantic knowledge. We choose embedded tense as a case study.
Any answer to a question about acquisition must begin with explicit assumptions about the learner. The learning framework we describe operates locally (both with respect to individual utterances and individual hypotheses) in a sense that we describe and defend below (Section 3); with supporting evidence drawn in part from the experimental literature on children's word learning. A further assumption we highlight concerns where learning begins. Drawing on evidence from Hollebrandse (2000) and Demirdache and Lungu (2011) we assume that children carry an initial belief that the past tense is indexical while the present tense is vacuous—a state which matches none of our three target grammars. The challenge, then, is to arrive at the (adult) grammars of Japanese, English, and Russian from this starting point using our local learning procedure.
Our finding, such as it is, is that the challenge can be met: for each of our languages there is a path from the initial state to the target grammar, provided that certain hypotheses are available to the learner and that certain critical constructions are sufficiently attested in the input. Identifying those critical inputs is, we think, one of the primary contributions of the paper. This is because the inputs that the logic of our account singles out are not necessarily the ones that inspection of the adult grammars would intuitively suggest, and perhaps only come into focus when semantics and acquisition are considered together.
Comments
To appear in the Proceedings of the 56th Annual Meeting of the North East Linguistic Society (NELS 56), GLSA Publications, University of Massachusetts Amherst