Volume 36, Issue 3 - Special Issue: Roots in Context, Guest editors: Noam Faust and Andrew Nevins

The vast majority of roots in English show uniform selectional properties across their various instantiations in verbs, nouns, or adjectives: relyV\textit{rely}_{\textrm{V}} on , reliance N_{\textrm{N}} on , reliant A_{\textrm{A}} on . This paper reports a new discovery: there are more than a hundred roots that display nonuniform selectional behavior. Their selectional class depends on whether the root is realized as a verb, a noun, or an adjective: pride V_{\textrm{V}} oneself on , pride N_{\textrm{N}} in , proud A_{\textrm{A}} of . I argue that this is best modeled if the categorizing node itself determines selection.

This paper argues that a core component of root meaning is the distinction between body parts versus the body conceived as a whole. This distinction is shown to be relevant in the acceptability of motion sentences in English with whole-body roots like dance$\sqrt {\textsc{dance}} $ and body-part roots like smile$\sqrt {\textsc{smile}} $. In keeping with the assumption that roots lack syntactic category, I argue that verbal roots occur freely in syntactic structures but that some root-structure combinations are degraded (or unacceptable), and that this is due to an incompatibility between conceptual root content and interpreted syntactic structure.

Recent work in Distributed Morphology, most prominently Harley (2014), argues for roots being able to take syntactic complements, which opens the door for the possibility of having syntactic features within a root’s representation – something most DM literature rejects (Embick 2015). Upon a closer inspection of the arguments presented in the literature, it is not clear whether the disagreement has an empirical underpinning, or whether it stems from the lack of methodological clarity as far as the identification of the precise nature of what constitutes a syntactic feature. This paper takes this methodological question seriously and investigates a type of derivational behavior that, in our view, provides a decisive argument for the presence of syntactic features on roots. We argue that the presence of a syntactic feature on the root can be conclusively established based on a feature’s impact on specific properties within a larger syntactic structure. Based on empirical evidence form gender agreement phenomena, we introduce a model of grammar that distinguishes roots with syntactic features from those which do not have them. We propose that such a distinction between roots will manifest itself in the timing of root insertion – roots without syntactic features are late inserted, while roots with syntactic features must be early inserted.

We investigate how saturation of different theta-roles by the non-head constituent correlates with derivational suffixes and, in turn, with the event structures compatible with those suffixes. We also investigate XP realisations of themes, causers and instruments in deverbal nominal and participial constructions and which ±agentive and/or ±process/episodic sub-readings allow which type of argument. It turns out that for each theta-role, the contexts that allow an XP realisation are exactly the complement of the contexts that would allow compounding of that same theta-role. We take this complementarity to be an indirect argument in favour of (i) divorcing argument licensing from argument selection and (ii) dissociating argument introduction from event-structure-related heads, which then potentially reaffirms the role of roots in (first phase) syntax.

We explore certain predictions of the theory first presented in Acedo-Matellán and Real-Puigdollers (2014), whereby roots correspond to (lately inserted) Vocabulary Items that phonologically and semantically interpret functional nodes. First, we deal with categorizers ( a, n, v ), a prominent locus for the insertion of roots. Roots such as cat or up are Vocabulary Items that have no context of insertion and are insertable into any categorizer, accounting for the categorial variability of simple words. Derivational affixes like - ation or - al also correspond to roots inserted into little head categorizers ( n and a , respectively, in this case), but their Vocabulary Items have a context of insertion accounting for their categorial rigidity and their c-selection properties. The exploration proposed here focuses, first, on the locality properties of morphemic interactions at the semantic interface, namely, allosemic interactions. We show that our approach makes felicitous predictions seemingly unavailable to those other localist approaches in which roots are distinct nodes in the syntax. Second, we show that the phenomenon of semi-lexicality, as illustrated by classifiers, can be successfully modeled, in our framework, as the insertion of roots into inflectional nodes.

The “causative” template heXYiZ in Hebrew is the morphological form of verbs which are usually transitive. I discuss cases in which specific roots give rise to the labile alternation, otherwise unattested in the language. A straightforward analysis is suggested for the majority of cases, based on causative Voice[+D]$\text{Voice}_{\text{[+D]}}$. This analysis is then extended to account for the labile exceptions, which inform how the idiosyncratic meaning of roots influences syntactic computation.

Working within the framework of Distributed Morphology (Halle and Marantz 1993, 1994), this paper offers a derivational analysis of the range of structures and the types of idiosyncrasy associated with compounding. Building on prior analysis by Harley (2009), compound structures are argued to vary according to the ways in which the head and the non-head of a compound are categorised. Specifically, if the non-head of a compound is acategorial, then the relationship between the compound head and non-head is non-decomposable. Based on data from Hebrew (Borer 2009), it is shown that this also makes the non-head inaccessible to independent syntactic-semantic operations, including coordination, and coreference with a pronoun. It is additionally shown that morphologically-conditioned allomorphy (Bobaljik 2012) may be conditioned between the compound head and a suffix, as constitutes part of a bracketing paradox (Williams 1981). Where categorisation of the head of the compound gives rise to effects of headedness, however, this allomorphy may be ‘blocked’ by the structure associated with exocentricity. The final sections of the paper consider exocentricity, and other interactions between idiosyncratic meanings and phonology, in further detail.

This paper presents a cross-linguistic survey of non-local allomorphy and it develops a formal model that accounts for the observed patterns. The distance between the trigger and target of allomorphy in non-local patterns is much more conservative than expected. A model of Vocabulary Insertion is developed, where the limited distance follows from the basic linear computational properties of the PF-interface.

In this paper I examine certain gradable adjectives in Serbian, whose suppletive comparative forms display unexpected semantic properties. In particular, while these adjectives are ambiguous between intersective and non-intersective readings in the positive form, their suppletive comparative and superlative forms are limited to the non-intersective interpretation. These facts show, I argue, that in a theory like Distributed Morphology either adjectival roots or category-assigning heads they combine with come in semantic subtypes (i.e. are specified for certain semantic properties; Harley 2005, Anagnostopoulou and Samioti 2014). I show how the analysis I propose explains semantic properties of change-of-state verbs derived from these adjectives and why these adjectives are restricted to the intersective interpretation when their positive form takes the long-form (definite) inflection. I also provide an illustration of how Arregi and Nevins’s (2014) analysis of the so-called “disuppletive” roots, such worse/badder , can deal with the facts presented in this paper. Finally, I discuss implications of these facts in the context of Bobaljik’s (2012) approach to suppletive comparative morphology.

Verbal suppletion for participant number has recently attracted theoretical attention (Bobaljik & Harley 2017; Toosarvandani 2016). I show that cross-linguistically, participant number is marked by suppletion and reduplication and is part of a broader phenomenon, verbal number. I argue that reduplication provides evidence that participant number suppletion is not directly triggered by the argument, and I propose a unified account of participant number where there is a verb-internal number node that reflects number and mediates an agreement relation between the argument and the root. The number node, which marks plural arguments and events, may be valued by the closest c-commanding DP, and will be realized as reduplication or will trigger root suppletion. This is supported cross-linguistically by languages which mark both participant and event number by reduplication and suppletion. I propose that this head allows a stricter locality condition on suppletion where the trigger is always a head in the morphological word (complex X⁰). My analysis bears on current issues of locality and suppletive domains since I argue that the topmost complex X⁰ node forms a morphological domain which contains smaller domains that are privileged in the syntax but restrictive in the morphophonology.

In this article we examine patterns of root allomorphy in Greek that involve vowel alternations and propose a Generalized Non-linear Affixation (Bermúdez-Otero 2012) analysis according to which these alternations result from the competition between segments that belong, on the one hand, to the vocabulary items of roots and, on the other, to the exponents of functional heads (Voice/Aspect, n ). More specifically, we claim that phonological entities have a gradient degree of presence in a structure, that is, are specified with a certain activation strength value underlyingly (Smolensky and Goldrick 2016). As a result, the surface realization of roots is determined by the relevant activation level of the exponents of functional heads they are eventually combined with. From all available exponents, the one that optimally complements the strength value of the vocabulary item of a given root will eventually surface. Our analysis is shown to be theoretically advantageous because it develops a strictly phonological account of allomorphy and, moreover, it captures the attested generalizations without resorting to extensive stem/span listing or to the application of phonologically unrestricted readjustment rules.

This paper provides two new arguments for the existence of the root as a level of representation in Semitic languages. The first argument, from the South Arabian language Mehri, presents a case of allomorphy in which the choice of the suffixal exponent is sensitive to the last vowel of its base, but not to its last consonant, even though the latter is closer to the position of the suffix than the former. Only an approach which distinguishes between a root and a template can explain this case by placing the template closer to the suffix than the root, against the linear order of segments. The second argument shows how in two Neo-Aramaic languages (Jewish Urmi and Jewish Arbel), only the consonants of verbs must be stored; all other verbal exponents are predictable on this basis.