This dissertation is broadly concerned with word-formation in Generative Grammar. Particularly, it investigates a class of modal adjectives in Brazilian Portuguese (BP) ending in -vel, as exemplified below in (1):

    1. 1.
    1. (a)
    1. O
    2. The
    1. vaso
    2. vase
    1. de
    2. of
    1. porcelana
    2. porcelain
    1. é
    2. is
    1. quebrável.
    2. break.MOD[-able]
    1. The porcelain vase is breakable” = the porcelain vase can be broken
    1. (b)
    1. Nina
    2. Nina
    1. é
    2. is
    1. amável.
    2. love.MOD[-able]
    1. Nina is lovable” = Nina inspires love and affection

Adjectives like those in (1a) (e.g., audível ‘audible’, decifrável ‘decipherable’, recuperável ‘retrievable’) express a disposition, and are compatible with a possibility reading. Adjectives like those in (1b) (e.g., admirável ‘admirable’, adorável ‘adorable’, deplorável ‘deplorable’) lack the possibility interpretation and express subjectivity (i.e., the “worthy of” reading; notice that admirável ‘admirable’, for instance, can be paraphrased by ‘worthy of being admired’). This variation in modal interpretation is not immediately obvious, and certain adjectives seem to convey both meanings (e.g., English readable: 1. able to be read, 2. enjoyable/interesting to read).

The investigation begins with a question posed by Halle (1973: 11) regarding “how a grammar is to reflect the fact that a derived word quite commonly shares semantic and syntactic properties with the word from which it is derived”. Existing theories of modal adjectives in different languages, from a lexicalist perspective, posit a (passive) lexical rule to derive them (Wasow 1977). From a syntactic perspective, (1a) and (1b) reflect the height of attachment of the suffix (i.e., compositional word derivation or idiosyncratic root derivation; see Oltra-Massuet 2010; 2014).

This dissertation provides a syntactic account in which -vel adjectives are uniformly derived from a passive operation, as defined by Bruening (2013: 35): “[t]he passive is a morphosyntactic operation that prevents the realization of the external argument as an argument.” The analysis adopts the system proposed in Ramchand (2008), in which verbal predication can be (maximally) decomposed into three heads: init (initiation), proc (process), and res (result). The adjectival suffix, conceived as a dynamic can expressing susceptibility or disposition, attaches to an unsaturated Initiation projection (i.e., little v). Thus, deverbal adjectives of type in (1a) and (1b) reflect subevental complexity.

The dissertation contains six chapters and one appendix. Chapter one is introductory. It presents lexicalist and non-lexicalist approaches to word formation, particularly focusing on adjectives (i.e., modal adjectives and adjectival passives). It discusses the relationship between modal adjectives and passives, mediopassives, and generic sentences, and also the range of modality that -vel expresses (i.e., possibility or subjectivity).

Chapter two isolates the main properties that modal adjectives share with the aforementioned constructions, provides a description of the data, discusses the preference for verbal bases and presents examples of newly coined adjectives in BP (e.g., bicicletável facebookável, instagramável, which are derived from the following nouns: bicicleta ‘bicycle’, facebook ‘facebook’ and instagram ‘instagram’, respectively).

Chapter three compares previous theories of modal adjectives and adjectival passives, both from a lexicalist perspective (in terms of lexical rules) and from a syntactic perspective (in terms of height of attachment), and evaluates their predictions and challenges. It is shown that positing a lexical rule to derive modal adjectives, as in Wasow (1977), misses important generalizations, particularly because an identical rule is also proposed to derive adjectival passives. Additionally, a passive transformation is also necessary according to this analysis, in order to derive verbal passives. Therefore, it is not possible to capture the relationship between them. Regarding Oltra-Massuet’s (2010; 2014) analysis, it is shown that the proposed correlation between low attachment of the suffix and idiosyncratic meanings, on one hand, and high attachment of the suffix and predictable meanings, on the other, does not always hold. This chapter also presents Ramchand’s (2008) framework of verbal decomposition and discusses its implications for the formation of deverbal adjectives in BP.

Chapter four focuses on the study of modality. Given the importance of this category in understanding the nature of -vel, this chapter focuses on the range of modality that adjectives formed with -vel express. Based on previous work on dynamic modality (Brennan 1993), which is concerned with abilities and dispositions, -vel is analyzed as a dispositional modal (i.e., a dynamic can) relativized to bundles of property-denoting expressions introduced by “in virtue of” phrases, following Brennan (1993). The possibility reading is correlated with dynamicity and involves the properties of a Resultee argument. The subjective reading is correlated with stativity and involves properties of a rhematic complement, which co-describes a state. This chapter explores a distinction between objective properties and subjective properties. In (1a), for example, the property-denoting expressions to which the modal is applied describe objective, physical properties of the entity in question, the vase (i.e., concerning its material, composition, fragility, etc.). Contrariwise, (1b) involves subjective properties (i.e., concerning Nina’s temperament, personality, etc.), which are subjectively evaluated by some kind of “judge”, in the sense of Lasersohn (2005).

Chapter five develops the idea that modal adjectives are uniformly derived from a passive operation (Bruening 2013). According to this view, the suffix always targets an Initiation projection. This chapter also explores the consequences of this analysis for adjectival passives in BP, and reevaluates syntactic tests with modal adjectives and adjectival passives, showing that both constructions allow by-phrases and adverbial modification in BP, contra previous analyses. Finally, chapter six discusses the main conclusions of the dissertation. The appendix gathers all the relevant data used in the dissertation.

Essentially, this dissertation proposes that modal adjectives ending in -vel involve the modalization of different verbal structures, which consist of richer structures that encode [init, proc, res] or [init, proc] (i.e., dynamic eventualities), or of simpler structures that encode [init] (i.e., stative eventualities). Instead of attributing apparently idiosyncratic meanings to root derivation, the analysis connects, to a large extent, the subjective interpretation with the lexical semantics of the verb – see, for instance, adjectives derived from psych verbs (e.g., adorável ‘adorable’) or judging verbs (e.g., deplorável ‘deplorable’). Regarding psych verbs, note, moreover, that the derived adjective is sensitive to the Subj.Exp/Obj.Exp distinction (cf. detestável ‘detestable’ and agradável ‘agreeable’). Thus, the subjective interpretation involves the presence of an Experiencer or some kind of sentient Agent (an undergoer of the process or event, as in readable) serving as “judge”. Ultimately, this dissertation provides evidence for a syntactic view of word-formation.