Discourse connectives are words or fixed expressions that express a discourse relation between the host and the preceding or following context, such as by the way, however, on the other hand, and so on, and so on. Importantly, the relations they express are not between sentences but rather between Discourse Units, which can be either sentences or hierarchical assemblies of sentence meanings. Many discourse connectives show no evidence of a syntactic category. Connectives vary somewhat unsystematically in their degree of freedom of position; for instance, by the way is extremely free, while so is restricted to initial position. Overall, then, discourse connectives behave asyntactically, like the parenthetical constructions of previous chapters. The chapter formalizes a representative collection of discourse connectives in Parallel Architectural terms: by the way, therefore, for example, in general, first, finally, too, however.

Parentheticals, it is fair to say, are everywhere; and, as can easily be shown, they come in a huge variety of flavors. Two characteristics set them apart from ordinary arguments and adjuncts: they tend to show some freedom of position within their hosts; and they tend to be semantically opaque to scope of quantification, negation, and long-distance dependencies. This study demonstrates that parentheticals are independent pieces of language that are jammed into their host, subject to prosodic constraints, and they stand in some discourse relation to the host, serving as a side comment or as supporting material. But crucially, parentheticals are “asyntactic”: they are not syntactic constituents of their host. The analysis is couched in terms of Jackendoff’s Parallel Architecture, in which phonology, syntax, and semantics are independent but linked generative systems, and in which words and rules of grammar are encoded in a common format. In support of the asyntactic treatment of parentheticals, this study develops accounts of prosody and discourse structure, components of the grammar that are new to the Parallel Architecture.

This chapter analyzes two classes of parentheticals with similar semantics: sentential commentaries like (1) and nonclausal commentaries (a.k.a. speaker-oriented sentence adverbs and adverbials) like (2) and (3). (1) Joe—you’ll be shocked to hear this—arrived on time this morning. (2) The economy, sadly, is tanking. (3) Parentheticals, to Marc’s chagrin, are confusing. These classes are relatively freely positioned in their host (with some exceptions), and both are opaque to quantification, negation, and extraction. They are understood as asserting new information, but they are not “at issue.” This chapter formalizes them in terms of the Parallel Architecture, as augmented in Chapter 2 to include prosody and discourse structure. They are shown to support the Asyntactic Hypothesis: they are attached to their host in prosody and discourse structure but not syntax.

This chapter addresses appositive relative clauses like that in (1) and nonclausal appositives like those in (2). (1) The economy, which had been doing so well, now is in trouble. (2) a. Seymour, a famous character in syntactic theory, sliced the salami. b. Angry at her boss, Beth decided to quit. Aside from needing to be right-adjacent to their antecedent, appositive relatives behave like other types of parentheticals, and unlike restrictive relatives. In particular, they are opaque to quantification, negation, and extraction. Nonclausal appositives such as the AP appositive in (2b) are freer in position than appositive relatives. It is shown that appositives can be treated in conformance with the Asyntactic Hypothesis: they are connected to their host in prosody and discourse structure but not in syntax. Their configuration is statable in the Parallel Architecture framework, but not in traditional generative grammar.

Complement as-parentheticals like (1) typically have a gap in the complement of a verb of propositional attitude. VP as-parentheticals like (2) have an empty VP, similar to VP ellipsis. (1) (2) Tom read War and Peace, as did Amy. The former class allows very free positioning within their hosts, the latter less so. They are both opaque to quantification, negation, and extraction. The speaker asserts both the host and the parenthetical, but the parenthetical is not “at issue.” This chapter formalizes a substantial collection of semantic properties of as-parentheticals, in terms of the Parallel Architecture framework. It also develops a Parallel Architecture account of VP ellipsis, drawing on the domain-general relation same-except. The upshot is that as-parentheticals are connected to their hosts in discourse structure and prosody, but not in syntax. They behave like independent clauses that happen to be jammed into the interior of the host.

AbstractTraditional generative grammar assumes without argument that only syntax is ‘generative’, that is, that the combinatorial complexity of language arises entirely by virtue of its syntactic organization. This chapter presents a framework in which phonology, syntax, and semantics are equally generative. Syntax is thus only one of several parallel sources of grammatical organization. The generative components communicate with each other through ‘interface’ components; it is shown that these interfaces are of nontrivial complexity. It is also shown that many of the alternative frameworks for generative grammar share this sort of parallel organization.

We propose a Complexity Hierarchy of grammars that map between sound and meaning, beginning with relatively trivial one-word grammars and culminating with the grammars of modern human languages. We argue that the levels in this hierarchy are plausible and necessary stages in the evolution of the contemporary human language faculty. We further argue that these levels leave traces in syntactically complex languages, as well as in home sign and emerging sign languages, where they can be thought of as "living fossils." An important stage in the hierarchy is what has been called "protolanguage" or "linear grammar"-a form of language that maps meaning to strings of words, but that, crucially, lacks syntactic and morphological structure. This gradualist scenario has implications for the division of labor between grammar and pragmatics. The simpler grammars in the Complexity Hierarchy place a strong reliance on pragmatics for many aspects of utterance meaning, including even basic questions such as who did what to whom. As grammars move up the Complexity Hierarchy and become more complex, these relatively simple interpretive factors become more systematic and less dependent on pragmatic inferences. However, pragmatic processes do not disappear. Rather, they change in character: Syntax, semantics, and the lexicon trigger highly structured pragmatic phenomena such as presuppositions and implicatures in a systematic and reliable way. In a sense, the more complex the grammar, the more opportunity for such pragmatic niches.

Epistemic parentheticals (“slifting parentheticals”) like (1) have nearly free positioning and are opaque to quantification, negation, and extraction. They have a gap that is understood as the content of a propositional attitude, as in (1). (1) The economy, Krugman says, is in trouble. The speaker of (1) is asserting that Krugman says that the economy is in trouble. But the speaker is not committed to the host itself: (1) may be followed by … but he’s wrong. There also nonclausal epistemic parentheticals with similar semantics: certain “sentence adverbs” and “adverbials” such as those in (2). (2) The economy, supposedly/apparently is in trouble. The rain, in Beth’s opinion/to the best of my knowledge/most likely, will stop soon. This chapter formalizes both clausal and nonclausal varieties in Parallel Architecture terms, and shows that they conform to the Asyntactic Hypothesis: they are connected to their hosts in discourse structure and prosody, but not in syntax.

This chapter illustrates the diversity and ubiquity of parenthetical expressions in spoken and written language. It reviews approaches in which parentheticals are constituents of the syntax of their host, and approaches in which parentheticals are not constituents of their host at all. It proposes the Asyntactic Hypothesis, in which parentheticals are constituents of their host in prosodic structure and discourse structure, but not in syntax. The chapter lays out a collection of distinctive properties of parentheticals: comma intonation, freedom of position, invisibility to anaphora that targets the host, opacity to scope of quantifiers and negation, no clefting or extraction, and no Condition C violations. In each case, the parenthetical acts as though it is an independent sentence from the host, not a syntactic dependent.

The Parallel Architecture is a conception of the organization of the mental representations involved in language and of the role of language in the mind as a whole. Its basic premise is that linguistic representations draw on three independent generative systems-phonological, syntactic, and semantic structures-plus a system of interface links by which they communicate with each other. In particular, words serve as partial interface links that govern the way they compose into novel sentences. It is shown that this architecture also provides a natural way to account for our ability to talk about what we see: semantic structure in language has to communicate via interface links with a level of spatial representation that encodes understanding of the physical world. It is suggested that such configurations of independent but linked representations are a widespread feature of cognition.

The formalism of Relational Morphology (R. Jackendoff and J. Audring, The Texture of the Lexicon, Oxford University Press, 2020) offers a straightforward way to encode the lexical entry of a one-off blend such as spork, and to relate it to words it is built from, e.g. spoon and fork. The approach extends easily to cases of overlap, such as when Spanish and English are blended to form Spanglish. Blending with overlap occurs not just with these one-off items, but also with certain affixes, for instance -ery. Nouns such as mock-ery and nunn-ery simply concatenate the base and the affix. But if the base ends in -er, for instance in flatter, the derived form is not ✶flatterery but flattery. We argue that this haplology is not the result of a truncation process, but rather that the stretch -er- is an overlap of the base with the affix. We show that the formal principles that account for the form of one-off blends and overlaps are readily generalized to affixes such as -ery that can overlap with their bases. This generalization is expected in the Relational Morphology framework, but not in more traditional procedural approaches to these phenomena.

Many years after the fact, I finally got around to reading Piattelli-Palmerini (1980), the chronicle of the 1975 debate between Jean Piaget and Noam Chomsky, with the participation of many other interested parties. Being a deeply committed Chomskian, though not an altogether orthodox one, I was surprised by the vehemence of the debate. After all, Piaget and Chomsky have a great deal in common. Both believe in complex unconscious mental processing. Both believe that the structure of the world we experience is in large part determined by internal mental constructs of potentially great abstraction. Both are firmly opposed to behaviorism.

What concept does the word ‘belief’ express, and how does it fit into the Theory of Mind? In its repertoire of syntactic niches (some not noticed in the literature) and in its range from stereotypical to marginal cases, "believe" is quite a typical word. The analysis offered here reveals striking grammatical and semantic parallels to expressions that denote depictions, such as "picture," "map," and "performance." These parallels provide unexpected solutions to classical philosophical puzzles about "belief." The analysis is extended to other propositional attitude predicates, to speech–act predicates, to predicates of actional attitude such as "intend" and "vow," and to commitment to norms of all sorts. These conclusions provide a sense of the richness of folk psychology and of how detailed linguistic analysis can uncover it, presenting challenges for future research.

Lexical semantics is to be treated as a mental phenomenon, deeply connected to and supported by the human conceptualization of the world. In this light, the unlimited number of possible word meanings must be accounted for in terms of a generative system that constructs word meanings from a finite set of primitives and principles of combination. This chapter sets out some basic elements of this system: the contribution of spatial cognition; nonperceivable features such as modality and the type-token distinction; semantic decomposition into other than necessary and sufficient conditions, such as prototypical conditions; a rich ontology including objects, events, locations, trajectories, amounts, and manners; polysemy that straddles ontological categories; “dot-objects” and the notion of person; argument structure and thematic roles; and the relation of lexical meaning to world knowledge.

A recent trend in psycholinguistic research has been to posit prediction as an essential function of language processing. The present paper develops a linguistic perspective on viewing prediction in terms of pre-activation. We describe what predictions are and how they are produced. Our basic premises are that (a) no prediction can be made without knowledge to support it; and (b) it is therefore necessary to characterize the precise form of that knowledge, as revealed by a suitable theory of linguistic representations. We describe the Parallel Architecture (PA: Jackendoff, 2002; Jackendoff & Audring, 2020), which makes explicit our commitments about linguistic representations, and we develop an account of processing based on these representations. Crucial to our account is that what have been traditionally treated as derivational rules of grammar are formalized by the PA as lexical items, encoded in the same format as words. We then present a theory of prediction in these terms: linguistic input activates lexical items whose beginning (or incipit) corresponds to the input encountered so far; and prediction amounts to pre-activation of the as yet unheard parts of those lexical items (the remainder). Thus the generation of predictions is a natural byproduct of processing linguistic representations. We conclude that the PA perspective on pre-activation provides a plausible account of prediction in language processing that bridges linguistic and psycholinguistic theorizing.

We discuss the conventionalization process in Central Taurus Sign Language (CTSL), an emerging sign language in its initial stages, and present evidence from four studies for the amount of lexical and structural variation it harbors. The four studies together portray the architecture of a young system under construction. Study 1 looks at signs for everyday objects and finds considerable variation in their degree of conventionalization. Study 2 examines modifier-head and negation-head relationships, showing substantial conventionalization. Study 3 addresses the emergence and conventionalization of various word orders as argument structure markers. It shows that conventionalization increases over successive cohorts of signers, but it decreases as argument structure becomes more complex. Study 4 demonstrates the emergence and conventionalization of distinctive morphological markers to signal the subtle semantics of symmetry and reciprocity. Finally, we raise the issue of whether CTSL presents evidence of syntactic structure, or whether its properties can be characterized in terms of a simpler “linear grammar” that maps directly between phonology and semantics.

Conceptual Semantics is a thoroughly mentalist theory of meaning, language, and thought (Miller and Johnson-Laird, 1976; Jackendoff, 1983, 1987a, 1990, 2002, 2012; Pustejovsky, 1995; Pinker, 2007; plus closely related work broadly in the Cognitive Grammar tradition by Lakoff, 1987; Langacker, 1987; Talmy, 2000; and many others). The theory represents a fundamental break from traditional formal semantics, and it has consequences not only for philosophy of language, but also for philosophy of mind, metaphysics, and possibly even ethics. The present chapter sketches a few of the many issues treated within this theoretical worldview, in the hope of capturing the spirit of the enterprise.