9. Constructional meaning and compositionality

One of the major motivations for constructional approaches to grammar is that a given rule of syntactic formation can sometimes, in fact often, be associated with more than one semantic specification. For example, a pair of expressions like purple plum and alleged thief 25 call on different rules of semantic combination. The first involves something closely related to intersection of sets: a purple plum is a member of the set of purple things and a member of the set of plums. But an alleged thief is not a member of the intersection of the set of thieves and the set of alleged things. Indeed, that intersection is empty, since only a proposition can be alleged and a thief, whether by deed or attribution, is never a proposition. This 30 chapter describes the various ways meanings may be assembled in a construction-based grammar. 1. Constructions and compositionality It is sometimes supposed that constructional approaches are opposed to compositional 35 semantics. This happens to be an incorrect supposition, but it is instructive to consider why it exists. A foundation of construction-based syntax is the idea that rules of syntactic combination (descriptions of local trees) are directly associated with interpretive and use conditions, in the form of semantic and pragmatic features that attach to the mother or daughter nodes in these descriptions (Kay 2002; Sag forthcoming). This amounts to the 40 claim that syntactic rules mean things. Meaning, of course, is generally viewed as the exclusive purview of words, and in the prevailing view of meaning composition, syntactic rules do no more than determine what symbol sequences function as units for syntactic purposes. So while syntactic rules assemble words and their dependent elements into phrases, and the phrases denote complex concepts like predicates and propositions, the rules 45 cannot add conceptual content to that contributed by the words; nor can they alter the combinatoric properties of the words. On this view, which Jackendoff (1997: 48) describes as the “doctrine of syntactically transparent composition”, “[a]ll elements of content in the meaning of a sentence are found in the lexical conceptual structures [...] of the lexical items composing the sentence” and “pragmatics plays no role in determining how [lexical 50 conceptual structures] are combined”. To embrace a construction-based model of semantic composition is not to reject the existence of syntactically transparent composition but instead to treat it, as per Jackendoff (1997: 49), as a “default in a wider array of options”. That is, whenever a class of expressions can be viewed as licensed by a context-free phrase structure rule accompanied by a rule composing the semantics of the mother from the semantics of the 55 daughter, a construction-based approach would propose a construction that is functionally equivalent to such a rule-to-rule pair. But constructional approaches also provide a revealing way to represent linguistic structures in which semantics of the mother does not follow entirely from the semantics of the daughters. A case in point is the pattern exemplified by the attested sentences in (1), retrieved from google. We will call such sentences pseudo60 conditionals, and we will refer to the if-clause and main clause as the pseudo-protasis and pseudo-apodosis, respectively. (1) a. If you’re 3Com right now, you’re considering buying add space in next week’s issue. b. If you’re George Bush, you’re now allowed to lie in the faces of trusting 65 young voters. c. [I]f you’re Betty Ford right now, you’re probably thinking, you know, I hope everybody’s OK. d. More than one able program director thinks commercials, promos and features is not an all-news station,] but if you’re new CBS President Dan 70 Mason right now you’re going to leave well enough alone. Example (2) shows that the pseudo-apodosis, like a true apodosis, can be extended beyond the bounds of the initial sentence. (2) If you are George W. Bush and this vending machine represents Iraq, you keep 75 putting money into the machine. When you have none left and it is obvious to all rational persons that trying again is not going to result in a different outcome, you borrow more and keep going. Syntactically the sentences in (1) and the first sentence in (2) appear to be ordinary 80 conditional sentences like (3). (3) If you’re pleased with the outcome, you may feel like celebrating. But the sincere speaker of the protasis of an ordinary conditional sentence does not hypothesize a patently impossible state of affairs, while the if-clauses of (1)-(2) appear to 85 pose the manifest impossibility that the addressee is identical to Peter Angelos/ Betty Ford/ George Bush/ Dan Mason/ etc. Of course that is not what is being said in (1)-(2). Exactly what is being said is difficult to pin down with certitude. The syntactic form is roughly given by (4). (4) If you are x, p(x). 90 The semantics seems to assert the proposition expressed by p(x), qualified in different examples by a number of different illocutionary forces or speaker attitudes. In any case, no hypothetical situation is posed; it appears that a categorical judgment is expressed (possibly hedged or epistemically qualified in some way) and the subject of that judgment is not the 95 addressee but the person identified as x; e.g., example (2) is clearly about George Bush, not about the consequences of a hypothetical identity between George Bush and the addressee. Pseudo-conditionals have the same form as (one type of) vanilla conditional but entirely distinct semantics. If the grammar accords to a sentence a different interpretation from what could be built up 100 piece by piece from its words and constituent phrases, syntactically transparent compositionality scores this as an instance of non-compositionality. As such, the pseudoconditional pattern could appropriately be called an idiom, but, as numerous proponents of construction-based approaches have observed, idiomaticity is not the same thing as inflexibility (Fillmor, Kay & O’Connor 1988, Michaelis & Lambrecht 1996, Culicover 105 1999). The pseudo-conditional pattern is evidently a productive one, and an adequate grammar must describe the interpretive and combinatoric constraints that define it. In a construction-based grammar, the pseudo-conditional sits on a continuum of idiomaticity (or generality) of expressions, somewhere between tightly bound idioms and fully productive processes. A construction grammar models this continuum with an array of constructions of 110 correspondingly graded generality (Sag forthcoming). Doing so obviously requires many more rules of composition than are countenanced in most non-constructional approaches— roughly as many as there are constructions listed in an (ideal) traditional grammar. A construction-based grammar sees nothing special about any part of the syntactic structure of sentences like (1)-(2); the syntax of (1)-(2) is the same as the syntax of (3)—that of a 115 common, garden-variety conditional sentence. But the meaning is different, and not obviously derivable by conversational implicature. So one posits a special construction with the syntax of a vanilla conditional, constrained as in (4), but with a semantic form unlike that of an ordinary conditional: a hedged categorical judgment is expressed—one whose subject is not denoted in the pseudo-protasis. 120 The pseudo-conditional is important for our purposes because the existence of this interpretive affordance appearance to undermine one of the foundational assumptions of syntactically transparent composition, as expressed by the following quote (from the online Stanford Encyclopedia of Philosophy): (5) “If a language is compositional, it cannot contain a pair of non-synonymous complex 125 expressions with identical structure and pairwise synonymous constituents” (Szabó (2007) If we use Szabó’s diagnostic, the existence of pseudo-conditionals entails either that English is not compositional or that pseudo-conditionals are syntactically distinct from 130 ordinary present-tense conditionals. A view of compositionality this narrow also presumably necessitates different syntactic analyses for any pair of readings attached to sentences in the large class illustrated by (6)-(7). Each such sentence yields both an idiomatic and a composed interpretation: (6) My yoga instructor sometimes pulls my leg. 135 (7) I’m afraid he’s going to spill the beans. A constructional approach welcomes a single syntactic analysis in all of these cases and posits constructions in the case of the idiomatic readings that attach semantic interpretations directly to certain relatively complex syntactic objects. In short, constructional approaches 140 recognize as instances of compositionality cases in which two different meanings for the same syntactic form are licensed by two different collections of form-meaning licensers, i.e., by two different collections of constructions. Construction-based grammars are nevertheless compositional in a quite usual sense: if you know the meanings of the words and you know all the rules that combine words and phrases into larger formal units, while simultaneously 145 combining the meanings of the smaller units into the meanings of the larger ones, then you know the forms and meanings of all the larger units, including all the sentences. The ‘bottom-up’ procedural language used here is intended only heuristically: most constructional approaches are explicitly or implicitly declarative and constraint based, notwithstanding the tempting metaphorical interpretation of construction as denoting the 150 building of big things out of little things. Constructional approaches tend to pay special attention to the fact that there are many such rules, and especially to the rules that assign meanings to complex structures. And such approaches do not draw a theoretical distinction between those rules thought to be of the ‘core’ and those conside