Verb forms in the clauses of conditional sentences are commonly seen a an important aspect of the overall characterization of conditionals. However, most analysts' interest has centered on subjunctive and “counterfactual” sentences, where they signal the speaker's negative commitment to the proposition. The unmarked, indicative verb phrases in conditionals are consequently assumed to be interpretable along with straightforward, non-conditional indicative declaratives. Consequently, sentence (1) below is usually claimed to belong to a different class of conditionals than (2) and (3). The analysis advocated below assumes that all verb forms in conditionals contribute in a significant way to the overall interpretation of the construction, and, in particular, reveal the specific nature of the non-factuality of the assumption in the scope of if and of the relation between p and q.With respect to these criteria the sentences exemplified in (1)–(3) will be claimed to form a uniform class, which I will refer to as predictive conditionals:

1. If it rains, the match will be canceled.

2. If it rained, the match would be canceled.

3. If it had rained, the match would have been canceled.

Predictive conditionals such as (1)–(3) will be discussed in the sections to follow with respect to their temporal interpretation and type of modality, as well as the kind of unassertiveness signaled and the relation between the assumptions in p and q.

The purpose of this book, as stated in the introduction, is to describe complex sentences whose subordinate clauses are introduced with if. This is the class of sentences which I have been calling “conditionals.” Different analysts, however, broaden or restrict the application of the term in a number of ways. For instance, logical accounts disregard constructions with questions and imperatives as q's, some analyses (e.g. Ramsey 1931) do not consider even if sentences to be conditionals, Austin (1961) doubts the conditional status of speech act sentences, and even Jespersen talks about “pseudo-conditionals.” Using yet another set of criteria, it is common to apply the term “conditional” to any sentence which can be paraphrased as a conditional.

In what follows I will try to show that, taking if to be the only explicitly conditional form in the English conjunction repertory, we can still offer a consistent explanation of if-compounds (such as even if) and of other conjunctions traditionally linked to conditionality: unless (as a negative marker of the protasis) and then (as a marker of the apodosis). In the final chapter I will also discuss the ways in which conditional meanings can arise without the presence of any overt markers.

Concessive conditional constructions

We have seen throughout this work that if … then constructions share important aspects of form and interpretation with constructions containing other types of adverbial clauses, in particular temporal and causal ones. As was noted by König (1986), concessive clauses should also be added to this list.

In the preceding chapters conditional constructions were characterized with respect to three major parameters: the use of verb forms, the relation between constituent clauses, and the epistemic status of the apodosis. In what follows I will discuss structural features of the construction: the form of the clauses and clause order.

Constituent clauses of the construction

A standard grammatical characterization of a conditional construction describes it as a complex sentence consisting of the main clause, or the matrix clause, and an adverbial clause of condition which is subordinate to it. Typically, both clauses are structurally complete (that is, the frame provided by the construction requires no deletions or special syntax in the clauses), and they are both declarative in form. In the sections to follow I will give an overview of some varieties of conditional constructions not conforming to this generalization.

Sentence types in the apodosis

Declarative apodoses are often seen as more typical for conditional constructions than interrogative or imperative ones. Linguistic descriptions concentrate on declarative conditionals, and this is carried over to pedagogical approaches, where conditionals with questions or imperatives as q's are discussed separately as exceptions. In truth-functional or logical works on conditionals, on the other hand, non-declaratives, which are uninterpretable in terms of truth value, are often entirely disregarded.

From a purely descriptive point of view, seeing declaratives as more typical apodoses is apparently a matter of statistics rather than of genuine differences in the status of declaratives and non-declaratives as parts of a conditionals structure.

In chapter 2 I distinguished two major classes of conditionals on the basis of their usage of verb forms. I argued that the verb forms are indicative of the overall interpretation of a conditional construction as predictive or non-predictive. I further suggested that the choice of verb forms also helps to specify (or at least constrain) the nature of the relation between the protasis and the apodosis. In the present chapter I will discuss the types of such relations and their relationship to other parameters of conditionality.

The protasis/apodosis relation seems to be a particularly important aspect of conditional interpretations. As I argued above, neither of the clauses of a conditional construction is asserted in the construction, even though the content of a clause may be brought into the construction after being asserted in the context (e.g. in If she is giving the baby a bath, I'll call back later the protasis is not asserted by the speaker, but it quotes a statement of the other participant in the conversation). The if-clauses of predictive sentences thus cannot be interpreted as asserting their protases (or, consequently, their apodoses). As I have argued, the presence of if in the construction marks the assumption in its scope as unassertable. As a result, the assumption in the apodosis, which belongs to the same mental space as the protasis, is not treated as asserted either.

Constructions, conventional meaning, and the grammar of conditionals

This book is an attempt to provide a description of a certain fragment of the grammar of English, namely, conditional sentences. By “conditional,” I will mean primarily the sentences so labeled by grammarians (rather than logicians): complex sentences, composed of the main clause (sometimes also called q, or the apodosis) and a subordinate clause (p, or the protasis). The subordinate clause is introduced by a conjunction, the least marked of English conditional conjunctions being if.

The analysis of conditionals attempted here will focus on providing an explanation of how aspects of conditional form give rise to a variety of meanings that conditional sentences express. That is, following the framework of cognitive linguistics, I will not treat the “grammar” as an autonomous formal description of linguistic structure, but rather as a representation of the speaker's knowledge of linguistic convention. In the cognitive approach (advocated by Fillmore 1977, 1982, Lakoff and Johnson 1980, Langacker 1987, 1991a, 1991b, Lakoff 1987, Fillmore, Kay, and O'Connor 1988, Fillmore and Kay 1994, and many others), it is not possible to speak of grammar in isolation from meaning, on the contrary, grammar is meaningful and essentially symbolic in nature. In Langacker's Cognitive Grammar, for example, lexicon, morphology, and syntax form “a continuum of symbolic units serving to structure conceptual content for expressive purposes” (Langacker 1987: 35).

As I argued in the earlier chapters, if as a marker of a conditional protasis instructs the hearer to treat the assumption in its scope as not assertable in the usual way. There are various reasons for the speaker to mark an assumption as non-assertive; most of these reasons, except in the rather exceptional case of commentary on linguistic form, are related to the speaker's state of knowledge. As I have tried to show, the type of unassertability influences the choice of verb forms in the sentence. Thus in predictive conditionals the speaker may mark two different types of lack of commitment with two types of backshift. The if-backshift marks the content of the protasis as in a sense “not knowable” – it is not factual and thus cannot be known to be true or false, and it is also not predictable on the basis of the present situation and its potential consequences. The hypothetical backshift marks both the protasis and the apodosis as even more difficult to assert, because of some other beliefs which the speaker holds. Thus, in hypothetical conditionals, the protasis is still not knowable or predictable, and is then further distanced because of the clash between its content and the content of other assumptions the speaker holds or entertains. The apodosis is then marked with the same distance, for its assertability crucially depends on that of the protasis.

Predictiveness and a prototypical conditional

Recent typological work has expressed an understanding of linguistic categorial universals of cognitive structure. Researchers such as Kemmer (1988/1993) and Pederson (1991) have examined the way in which similar meaning categories are encoded again and again, in unrelated languages' “middle” and “reflexive” markers. Even more interestingly, not only are certain meanings cross-linguistically more likely to be encoded than others, but there seem to be a range of more or less closely related meanings which are likely to share a marker cross-linguistically. The same seems to be true of conditional constructions. When we identify a construction as “conditional” in some previously undescribed language, we do so on the basis of its meaning, and, I would claim, predictive conditional relations at the level of content are the meaning most likely to be identified as centrally “conditional” by most analysts. Yet we have an intuitive understanding that the expression of conditional relations in other domains may well share at least some aspects of formal expression with content-level prediction, in other languages as well as English.

Within English, evidence for the centrality of the predictive function to the definition of conditionality comes from several sources – beginning with the observation that native grammarians and teachers invariably light on these cases first. Most importantly, however, the expression of this kind of conditional meaning involves the tightest complex of entrenched, grammaticized constructional relation of form to meaning.

Slavic Prosody is a book about language change and language structure. It seeks to provide a coherent account of the Slavic languages at the time of their differentiation and later, and to relate some of the findings to issues in phonological theory. In demonstrating that many persistent problems of Slavic historical linguistics, such as the development of nasal vowels, the well-known but little understood innovations in the liquid diphthongs, and the various accentual changes may be solved in a theoretically interesting way by reference to the syllable, the book offers a new analysis and a different view of the Slavic languages. This is not a historical or comparative grammar in the usual sense, nor does it contain the complete individual histories or synchronic descriptions of the Slavic languages. But by exploring some intriguing problems of Slavic historical linguistics and their synchronic resonances within a contemporary theoretical framework, the study aims to engage both Slavists and general linguists, as well as those simply interested in the Slavic languages or in language per se.

Given that the objective of Slavic Prosody is to examine phenomena within the syllable and the role of the syllable within the word, it is particularly rewarding to look at how these relationships changed over time. The study therefore focuses on Late Common Slavic, the period of significant upheaval and differentiation of the Slavic dialects, but it looks at Late Common Slavic in relation to both its past and its future.

Background

Slavists agree that the syllable played a major role in the history of the Slavic languages and there are various principles, tendencies, and laws to explain how sound change related to syllable structure in Common Slavic. But our understanding of the syllable in Slavic is still fairly rudimentary. After Leskien (1909: 52ff.) observed that several developments in Proto- and Common Slavic had the effect of converting closed syllables to open ones, Iljinskij (1916, 1926) associated the loss of final consonants, the simplification of consonant clusters, the monophthongization of diphthongs, the changes in liquid diphthongs, and the creation of nasal vowels and syllabic liquids with a “law of open syllables.” By formulating this “law” as a “principle of rising sonority,” van Wijk (1931) was able to include the prothesis of glides among the series of innovations consistent with a certain syllable type. Other changes, including palatalization and vowel fronting, seemed to produce a similarity between sounds within a syllable and in 1929 Jakobson extended our understanding of the syllable as a phonological domain by referring to these changes as being consistent with a “tendency for syllable synharmony.” Slavists since then have viewed the development of Common Slavic from Proto-Indo-European (PIE) as governed by open syllable structure or rising sonority, and syllable synharmony. It is also widely claimed that the loss of certain vowels (jers) in Late Common Slavic (LCS) thoroughly restructured the phonological system of Slavic, creating newly closed syllables.

The preceding discussion argued that syllable structure changed in Slavic before the loss of weak jers and that intrasyllabic and intersyllabic reorganizations in Late Common Slavic (LCS) had a far more important role in its evolution than did the loss of jers. Not only did the reorganization of syllable structure produce major isoglosses in LCS, but it also had far-reaching implications for subsequent developments in the individual Slavic languages, both on the segmental and the prosodic level. Several new proposals were made with respect to the nature of phonological change in Common Slavic (CS).

The syllable structure changes that occurred in CS were the result of several constraints. The earliest innovation was the emergence of the Moraic Constraint. It resulted in the elimination of syllable-final obstruents and in the neutralization of length in diphthongs. A No Coda Constraint came into play later in CS. Historical accounts of the Slavic languages do not distinguish between the two, citing the “law of open syllables” as responsible for the loss of syllable-final consonants or a movement of the syllable boundary. By Late Common Slavic the high ranking of the No Coda Constraint was a regional phenomenon. The developments in (North) East LCS, for example, may be described as the interaction of two other constraints, one on maximal syllable weight and another on the moraicity of sonorant consonants without reference to the No Coda Constraint. (North) West Slavic, in fact, had syllable codas as an outcome of constraint interaction in that area.