Introduction

Ever since Aristotle and Plato (The Categories; Cratylus), linguists have considered language to be the pairing of form (sounds or gestures or written strings) and meaning. This is true for all meaningful linguistic units from morphemes, through words, phrases and sentences, to discourse. Generally speaking, semantics is the study of how form and meaning are related. However, semantics is more narrowly construed as excluding those meanings that derive from speaker intensions and psychological states, as well as sociocultural features of the context. Furthermore, the boundary between semantics proper and pragmatics is intensely debated and to some researchers constitutes an empirical question. Formal semantics came into being as a system describing formal languages, that is, the mathematical and logical languages of computing machines as opposed to the natural languages of human beings. However, in the late 1960s the philosopher Richard Montague argued that natural languages such as English could be fruitfully described using the same rigorous rules and correspondences utilized in the description of formal languages. Modern formal semantics was born and is currently prospering as a branch of linguistics.

Before we embark on considering how form–meaning correspondences are acquired in a second language (L2), it is important to consider what is there to be acquired. In other words, what are the semantic differences between two natural languages, for instance between English and Spanish? The study of semantic universals (von Fintel & Matthewson Reference von Fintel and Matthewson2008; Bach & Chao Reference Bach, Chao, Maienborn, von Heusinger and Portner2012) is based on the strong Effability Hypothesis (Katz Reference Katz1976: 37).

Assuming this hypothesis as a general guide or heuristic forces us to see semantic universals as general laws delineating and describing meanings articulated in all human languages in one way or another. The main question of language variation then shifts to: How are these universal meanings expressed? If all languages have, in principle, the same expressive power, the differences must lie in the type and especially in the size of the linguistic sign needed to encode a given meaning. One language may possess a single morpheme or word to express a given meaning while another language would need a whole story to encode that same meaning. In anticipation of linguistic Minimalism, van Benthem (Reference van Benthem and Zaefferer1991) conceives of systemic semantic universals as arising from the fact that languages are constrained by reflecting human cognition.

What is the task of the second (and additional) language learner? This learner already has one language in her mind/brain, her native language, and all possible meanings are thus available to her. If all meanings are in principle expressible in all languages, then it is only the various expressions of universal meanings that need to be acquired. Note that looking for variation in meaning expressions should be across linguistic levels, as suggested above: what one language conveys with a functional morpheme, another language may denote with an open-class lexical item or with a whole sentence. A simple example comes from a comparison of the means of marking evidentiality in Bulgarian and English.

1. 1. Ivan vid-ja Maria. (Bulgarian)

2. Ivan see-3sg.Past.Indicative Maria

3. ‘John saw Mary.’ (and I am asserting this because I witnessed the event)

4. 2. Ivan vid-ja-l Maria.

5. Ivan see-3sg.Past-Evidential Maria

6. ‘John saw Mary.’ (but I was not a witness of the event; it is hearsay).

A comparison between the two meanings indicates that, unlike in the example in (1), the speaker of sentence (2) does not have direct evidence for the proposition expressed. While Bulgarian marks evidential modality on the past verbal form, English needs a whole additional proposition to convey the meaning. Note that the English sentence John saw Mary is actually vague with respect to evidentiality; discourse knowledge or a direct enquiry will disambiguate, if needed.

To extend this example to acquisition, consider the learning task of the English native learner of Bulgarian. At first, this hypothetical learner might not notice the functional morpheme –l in the verbal paradigm, especially because there is a lot of idiosyncratic variation in the paradigms. Transfer from English would ensure that this learner considers the Bulgarian sentence in (1) and all the sentences like it as vague with respect to whether the propositional event is hearsay or not. Though this particular learning situation has not been investigated in second language acquisition (SLA) to my knowledge, it is likely that the evidential meaning will come into the grammar at post-intermediate proficiency levels, when the learner acquires the meaning contrast between vid-ja and vid-ja-l. It is not enough that the learner recognizes the existence of the two forms: she must pair each one of them with their distinct meanings.

The example demonstrates that languages vary not in the meanings they express but in the means of expression. In this research timeline, I will examine how the pairing, or mapping, of form and meaning has been investigated in SLA. In order to impose some system on the various meanings, I will largely follow the outline of Semantics: An international handbook of natural language meaning edited by Claudia Maienborn, Klaus von Heusinger & Paul Portner, vol. 2. Since not all meanings studied by formal semanticists yield themselves easily to empirical study, and not all meaning expressions constitute interesting cross-linguistic variation, only those investigated in SLA will be represented in the timeline. In order to delineate the subject-matter, only studies that consider knowledge of meaning will be included, disregarding studies looking at the (un)acceptability of morphosyntax.

1. A. Constraints on possible lexical items

   1. A1. Causative and inchoative verbs

   2. A2. Dative and double object structures

   3. A3. Psych(ological) predicates

   4. A4. Unaccusative and unergative verbs

   5. A5. Motion verbs

2. B. Noun Phrase (NP) semantics

   1. B1. Pronouns

   2. B2. Definiteness and Specificity

   3. B3. Quantifiers

   4. B4. Bare and mass noun phrases, plurals

   5. B5. Genericity

3. C. Verbal Phrase semantics

   1. C1. Lexical aspect (aspectual classes of verbs)

   2. C2. Grammatical aspect (perfect and progressive)

   3. C3. Verbal mood (subjunctive)

   4. C4. Deverbal nominalization

4. D. Semantics of Adjectives and Adverbs

5. E. Clausal-level semantics

   1. E1. Tense

   2. E2. Modality

   3. E3. Conditionals

   4. E4. Scope and negation

   5. E5. Questions

   6. E6. Ellipsis

6. F. Discourse-level semantics

   1. F1. Topic and Focus

   2. F2. Discourse effects of word order (WO) variation

The broad categories range from lexical meaning to discourse meaning. While it has been argued that lexical items rarely have exactly the same meaning (denotation as well as connotation) in language after language, there are solid candidates for universal constraints on possible lexical items. One such universal are aspectual lexical classes of the Vendler (Reference Vendler1967) type (van Valin Reference van Valin, Mairal and Gil2006), or at least the building blocks from which event types are composed (von Fintel & Matthewson Reference von Fintel and Matthewson2008). Meanings pertaining to different types of noun phrases (e.g. definite, specific, generic) come next. Verbal phrase meanings start from lexical aspectual classes but also include the grammatical morphemes operating on those classes to change their meanings in predictable ways. The broad topic of adjective modification is not very widely studied in SLA, unlike clausal-level semantics, where there is much more research. The final category deals with discourse-level meanings of Topic and Focus. Justifiably, the stronger emphasis in SLA research has been on the meanings whose expressions vary from language to language (e.g. tense, mood, aspect) than on meanings that may be truly universal (e.g. compositionality, presuppositions), and so this bias is reflected in the timeline.