Decomposing Notions of Adjectival Transitivity in Navajo1 Elizabeth Bogal-Allbritten [email protected] Abstract Points of variation manifested by adjectives crosslinguistically has received much recent attention in the literature. This paper argues that one way in which adjectives may differ (crosslinguistically or within a single language) is in their projection of a degree argument position in the syntax. Under standard analyses of adjectival meaning, semantic transitivity implies syntactic transitivity. However, the Navajo data presented in this paper suggests that while all Navajo adjectives have a degree argument in their semantics, syntactic projection of the degree argument is only licensed by special morphology on the adjective. 1. Introduction 1.1. Overview While differences in the semantics of particular degree constructions have long been a focus of study, the gradable adjectives occurring in these degree constructions are typically taken to have the same transitive semantic type which is reflected in their syntactic structure. The most widely adopted class of approaches takes gradable adjectives to be relations between individuals and degrees, where the degree argument is syntactically projected like an individual argument (Seuren 1973, Cresswell 1977, von Stechow 1984, Bierwisch 1989, Heim 2001). (1) [[tall]] = λddλxe.x is tall to d The degree argument d can be directly saturated through composition with degree-denoting elements like measure phrases or degree pronouns (2a). The degree argument can also be quantified over, as with comparison of superiority (more…than) (2b). The important point, however, is that all gradable adjectives categorize for a degree argument. (2) a. Susan is {5 feet / that} tall. b. Susan is taller than Sam. More recently, studies have begun to consider whether this syntactic and semantic approach is adequate given types of typological variation among adjectives and degree constructions. One set of approaches has argued that adjectives in certain languages are semantically and syntactically intransitive expressions like (3) (Klein 1980, Kennedy 2007). This style of analysis has been applied to Motu (Beck et al. 2009) and Washo (Bochnak to appear). (3) [[tall]]c = λxe.x is considered tall in c This paper presents a novel analysis of adjectival data from Navajo (Southern Athabaskan). I argue that while all adjectives in Navajo are semantically transitive – they have both a degree and an individual argument – morphology determines whether adjectives are 1 I thank Navajo consultants Ellavina T. Perkins, Louise Kerley, and Irene Silentman. Page numbers from Young and Morgan (1987) are prefixed with ‘d’ for ‘dictionary’ and ‘g’ for grammar. 1 syntactically transitive or intransitive. Navajo permits adjectival stems (the final syllable in the adjective, e.g., -neez) to bear different sets of morphology. The choice of morphology determines whether the adjective must be accompanied by an additional expression. Adjectival stems marked by COMPARATIVE ASPECT (CA) are only grammatical when directly preceded by a member of a set of expressions I will analyze as degree phrases (Heim 2001). The set of degree phrases includes equative (4b) and comparative standards of comparison, high degree adverbs (‘very’), and measure phrases (‘six feet’). (4) a. *Shimá ’áníłnééz. my.mother 3SUBJ.tallCA COMPARATIVE ASPECT b. Shimá [shí=gi] ’áníłnééz. my.mother me=at 3SUBJ.tallCA ‘My mother is as tall as me.’ COMPARATIVE ASPECT The other two sets of adjectival morphology are called ABSOLUTE ASPECT (AA) and PERFECTIVE ASPECT (PA). Although AA- and PA-marked adjectives are morphologically distinct from each other, they behave identically in ways that set them apart from CA-marked adjectives. In particular, AA/PA-marked adjectives do not have to be preceded by a degree phrase: (5) a. Shimá nineez. my.mother 3SUBJ.tallAA ‘My mother is tall.’ b. Kinłání deesdoi. Flagstaff 3SUBJ.hotPA ‘Flagstaff is hot.’ ABSOLUTE ASPECT PERFECTIVE ASPECT However, a vague predicate analysis is not appropriate for AA and PA-marked adjectives. They can be modified by the same set of degree phrases that was available for CA-marked adjectives. These constructions have the semantics that we would expect if the adjective had a degree semantics. The only difference is syntactic. When AA- and PA-marked adjectives are modified by degree phrases, the degree phrase has a special syntax, shown in (6a). The degree phrase is followed by the copula (‘át’é ‘he/she/it is’). The degree phrase copula are then marked by the clausal subordinator =go. This syntax is characteristic of modifiers of clauses in Navajo. This syntax is not permitted for CA-marked adjectives (6b). (6) a. Shimá [shí=gi ‘át’ée=go] nizhóní. my.mother me=at 3SUBJ.be=SUB 3SUBJ.prettyAA ‘My mother is as pretty as I am.’ b. *Shimá [shí=gi ‘át’ée=go] ’áníłnééz my.mother me=at 3SUBJ.be=SUB 3SUBJ.tallCA ABSOLUTE ASPECT COMPARATIVE ASPECT I propose that all Navajo adjectives are semantically transitive. However, only CA-marked adjectives are syntactically transitive. CA morphology [‘á ní ł ] combines with the stem [-nééz] 2 ‘tall’ and projects an extended syntactic structure of the shape in (7). The CA-marked adjective syntactically subcategorizes for both an individual and a degree argument (DegP): their syntax is such that they select for a DegP (7) AP t 3 <et> DegP 3 DP A <edt> 3d CA stem [[ACA]] = λxλd.TALL(x) > d By contrast, AA- and PA-marked adjectives are syntactically intransitive. AA/PA-marked adjectives only syntactically select for an individual. A DegP cannot directly combine with an AA/PA-marked adjective. The AP level of an AA/PA-marked adjective will be of type <dt>. (8) AP <dt> 3 DP A <edt> 3 AA/PA stem [[AAA/PA]] = λxλd.TALL(x) > d An AA/PA-marked adjective can be modified by a DegP only at the clausal level. To close the semantic degree argument of the adjective at the AP level, I propose that a type-shifting operation, CLOSE, applies (9). At application of CLOSE, the adjective’s degree argument (d) is quantified over and a domain variable (D) is introduced. The introduction of domain variable D gives us a way to indirectly manipulate the adjective’s bound degree argument. The analysis of CLOSE and modification of AA/PA-marked adjectives by degree phrases builds on Schwarzschild’s (2010, 2011) work on adverbial modification of adjectives. (9) [[CLOSE]] = λAdtλDdt∀d.D(d) A(d) This paper’s primary claim is that semantic transitivity of adjectives does not guarantee their syntactic transitivity. After introducing at greater length the standard semantic and syntactic analysis of adjectives in §2, I present Navajo data in support of the claims about syntactic (in)transitivity of Navajo adjectives in §3. In §4, I argue that an analysis of AA/PA-marked adjectives as semantically intransitive will not be possible: they, like CA-marked adjectives are both semantically transitive. In §5, I address how we might formally and compositionally implement the proposal for Navajo. I present a compositional analysis of both CA- and AA/PAmarked adjectives used in various degree constructions. In §6, I suggest further directions for work. First, I investigate a decompositional analysis in the style of Bartsch and Vennemann (1972) and Kennedy (1999), where CA, AA, and PA are separated from the adjectival stem. I also place the proposal in the context of recent work on crosslinguistic differences in adjectival meaning. §7 concludes. 3 2. Assumptions about Adjectival Meaning 2.1. Adjectives as transitive expressions The standard analysis of degree constructions assumes that adjectives are transitive expressions denoting a relation between individuals and degrees along some dimension (e.g., height) (Seuren 1973, Cresswell 1977, von Stechow 1984, Bierwisch 1989, Heim 2001, Kennedy and McNally 2005, inter alia). Heim (2001) makes explicit the relationship between the denotation of an adjective and its syntax. Heim analogizes the syntax and semantics of an adjective like tall to transitive verbs. Both expressions project internal and external argument positions. The adjective takes the degree argument as its internal argument and the subject as its external argument.2 The adjective exits the lexicon as a type <det> expression with both its syntax and semantics fully formed: (10) AP 3 subject 3 degree A [[tall]] = λdd.λxe.tall(x) > d Below, I will return to an alternative conceptualization of the lexically supplied meaning of adjectives. First, I will sketch the ways in which the degree argument can be saturated or bound. I assume measure phrases to be degree-denoting expressions that can directly saturate the degree argument position, as in (11).3 (11) a. James is six feet tall. b. James is [AP six feet [A’ tall]] c. [[tall]](James)(six feet) = 1 iff tall(James) > 6 feet An adjective’s degree argument can also be quantified over. Degree constructions in which the adjective’s degree argument is quantified over include the comparison of superiority (more…than), comparison of inferiority (less…than), and equative (as…as). I will exemplify these constructions with comparison of superiority. Different entries have been posited for degree heads (Deg) like MORE. I assume an A-NOT-A analysis of the comparative morpheme (Schwarzschild 2008). MORE takes two properties of degrees (D1, D2) as argument. The first property of degrees, D1, is contributed by the than-phrase. The second property of degrees, D2, is contributed by the main clause adjective. A sentence with MORE is true iff there exists a degree d that is in D2 (the extension of the main clause adjective applied to the subject) but not in D1 (the extension of the than-phrase). 2 This analogy only goes through under a pre-Kratzerian conceptualization of transitive verbs (Kratzer 1996). alternate and widely accepted proposal for measure phrases as predicates of intervals (type <dt,t> expressions) was introduced by Schwarzschild and Wilkinson (2002) and discussed further in Heim (2006). Under the interval analysis, measure phrases would undergo QR like the DegP headed by more. 3 An 4 (12) [[MORE]] = λD1 dtλD2 dt∃d.D2(d) & ¬D1(d) Under this account, all adjectives syntactically select for a DegP to occupy the specifier position of AP. In the subcomparative (13), the type <det> adjectives in the main and embedded clause must first compose with a degree. The degree is then abstracted over. (14a) shows the underlying structure while (14b) shows the corresponding LF. The compositional semantics given the above entry for MORE are shown in (14c). (13) The door is wider than the table is long. (14) a. [IP the door is [AP [DegP MORE [CP than [IP the table is [AP OP long]]]] wide ] ] b. [ [MORE 1 the table d1 long] 2 the door d2 wide ] c. [[(14b)]] = MORE(λd1.the table is long to d1)(λd2.the door is wide to d2) = ∃d.(λd1.the table is long to d1)(d) &¬(λd2.the door is wide to d2)(d) = ∃d.(the table is long to d) & ¬(the door is wide to d) “There exists a degree d to which the table is long and to which the door is not-wide.” Both adjectives wide and long compose first with a degree that is subsequently abstracted over. I have underlined this position in (15a) for both adjectives. This is the type of syntactic position occupied by arguments. Even where it is not explicitly discussed as in Heim (2000), semantic and syntactic notions of transitivity are consistently linked where the analysis of adjectives as type <det> expressions is adopted. 2.2. Adjectives as intransitive expressions The linking of syntactic and semantic transitivity is maintained also by analyses that take adjectives to be semantically intransitive. Klein (1980) and more recent adaptations (e.g., van Rooij 2008) take adjectives to be vague predicates of type <et>. Such adjectives only take an individual as argument and have a denotation like in (15). Whether or not x counts as tall is evaluated with respect to contextual parameter c. (15) [[tall]]c = λx. x counts as tall in c The vague predicate analysis is well-equipped to handle simple predicative uses of adjectives (16). These instances of adjectives are uncontroversially intransitive. (16) Mary is tall. T iff Mary counts as tall in the context c However, a vague predicate analysis is not appropriate for languages for which a degree semantics is motivated; see von Stechow (1984) for arguments for a degree semantics, and Kennedy (2007) and Beck et al. (2009) for diagnostics for a degree semantics. Under the standard analysis, then, intransitive adjectives bear a null POS(ITIVE) morpheme. A representative semantics for POS is given in (17). When applied to the adjective tall in (17), the sentence is true iff there exists a degree d to which Sally is tall and d exceeds the contextual standard of comparison (STND). (17) [[POS]]c = λgdet.λxe.∃d. x is g to d ∧ d > STND (Cresswell 1977) 5 The morpheme POS occupies the Deg head position. The POS Deg head does not select an overt standard phrase as, for instance, MORE selected a than-phrase as shown in (14). Instead, the POS morpheme existentially closes the degree argument of the adjective and relates it to the contextual standard of comparison. The POS head occupies the same position in the syntax that was occupied by measure phrases (11) and degree traces (12). (18) AP t 4 <et> DP 4 e Deg A POS tall <det,et> <det> Through POS, the standard analysis can creating intransitive adjectives from adjectives that are underlyingly both semantically and syntactically transitive. 3. Syntactic (In)Transitivity Under the standard analysis, semantic transitivity entails syntactic transitivity. Navajo does not seem to fit this conceptualization, however. Navajo has three different sets of morphology that can mark adjectives: COMPARATIVE ASPECT (CA), ABSOLUTE ASPECT (AA), and PERFECTIVE ASPECT (PA). I present evidence compiled by Young and Morgan (1987) and collected in original fieldwork that CA-marked adjectives are syntactically transitive while AA- and PA-marked adjectives are syntactically intransitive. The evidence is summarized below. (19) CA-marked adjectives: a. Are obligatorily and directly preceded by a degree phrase. (e.g., comparative clause or measure phrase). b. Bear morphology that indicates syntactic transitivity. AA/PA-marked adjectives: a. Do not have to be modified by a degree phrase. When a degree phrase is present, it must be marked as adverbial. b. Bear morphology that indicates syntactic intransitivity. One might argue that CA-marked adjectives can be analyzed according to the standard analysis, while AA/PA-marked adjectives should be analyzed according to the vague predicate analysis. However, I will argue against this hypothesis. The syntactic transitivity of Navajo adjectives is determined by morphological marking: CA projects a syntax that selects for two arguments (a degree and an individual), while the syntax projected by AA/PA selects only for one argument (an individual). However, all adjectives (whether CA or AA/PA-marked) take an individual and a degree as arguments. More formally, all adjectives (CA-, AA-, and PA-marked) have denotation in (20). This is the denotation from the standard analysis. The novel part of the proposal comes from the syntax. CA projects the syntax in (21a) while AA/PA projects the syntax in (21b). 6 (20) (21) [[ACA/AA/PA]] = λxeλdd.x is A to d a. AP b. AP 3 3 degree 3 subject AAA/PA subject ACA d In the remainder of §3, I present empirical arguments for this proposal. After setting up my assumptions about Navajo in §3.1 and introduce the morphology and distribution of CA, AA, and PA in §3.2, I present a series of empirical arguments for the proposal outlined above. 3.1. Basic description and theoretical assumptions for Navajo Adjectival meaning in Navajo is expressed with structures that Young and Morgan (1987) describe using the same verbal template of stems and prefixes that they adopt for eventive verbs. All Navajo verbs (including adjectives) are traditionally described in terms of templatic morphology. Consider the verb in (22). I will briefly outline the functions of portions of the template relevant to our discussion of adjectives in the language. Prefixes that are not of relevance to our discussion, or which will not be analyzed individually, will be labeled PRFX. (22) yi ø ø mas ⇒ yimas PRFX SUBJECT VALENCE STEM ‘It is rolling.’ (Young and Morgan 1987: g119) All verbs and adjectives have a stem. In (22), the stem is –mas. This stem occurs in verbs in which an object is moving in a rolling motion. Stems do not appear on their own: they require additional morphology. Minimally, a VALENCE (VAL) marker and a subject marker are required to mark both verbal and adjectival stems, even if these markers are not overt. VAL markers are frequently taken to indicate the number of arguments syntactically projected by the verb. Substitution of a different VAL marker changes the meaning of the verb as a whole. The same stem appears in (22) and (23). In (22), the VAL marker is [ø], which is associated with an intransitive syntax.4 In (23), the VAL marker is [ł], which is associated with transitive argument structure. (23) yoo ø ł mas OBJECT SUBJ VAL STEM ⇒ yoołmas ‘S/he/it rolls it.’ (Young and Morgan 1987: g119) The verb also always bears morphological marking of subject (SUBJ) and, for transitive verbs, the object (OBJ), regardless of whether a verb-external noun phrase is also used. The subject marker in (22) and (23) is the null third-person subject. Verb-external nouns are typically used only to introduce new referents in the discourse and are otherwise be freely omitted (Willie and Jelinek 2000). Both (24a) and (24b) are grammatical. (24) a. Jooł yimas b. Yimas 4 Referred to as a ‘classifier’ in the Athabaskan literature. 7 ball 3SUBJ.roll 3SUBJ.roll ‘The ball is rolling.’ ‘It is rolling.’ Optional modifiers can be included specify location, time, etc., as Kinłánídi ‘in Flagstaff’ in (25). I will assert that these modifiers are adverbs. Not only are these modifiers optional (25a) like verb-external nouns are, but their linear position relative to the verb and other verb-external material is flexible (25b,c). They are assumed to be adjoined above the verb: they do not occupy argument positions. (25) a. Naalnish. 3SUBJ.work ‘He/she works.’ b. Biyáázh Kinłání=di naalnish. his/her.son Flagstaff=at 3SUBJ.work ‘His/her son works in Flagstaff.’ c. Kinłání=di biyáázh naalnish. Flagstaff=at his/her.son 3SUBJ.work ‘His/her son works in Flagstaff.’ (Faltz 2000: 38-39) While verb-external nouns like jooł ‘ball’ and locative modifiers like Kinłánídi ‘in Flagstaff’ are both optional, nominal arguments are obligatorily marked on the verb. This has led some authors – proponents of the Pronominal Argument Hypothesis – to argue that the subject and object pronominal morphemes on the verb occupy the argument positions of the verb. Verbexternal nouns are only optional because another element (the subject or object marker) has already satisfied the needs of the verb, e.g., Case assignment. For concrete accounts, see Jelinek Baker 1996, and Willie and Jelinek 2000. By contrast, locative modifiers are optional but also are not marked any on the verb, suggesting they are only adverbial modifiers. Faltz (2000) writes that if the syntactic proposals of the Pronominal Argument Hypothesis are carried over to semantics, then there are no Navajo verbs that, when spelled out, are of type <et> or <eet>. Faltz follows the lead of previous syntactic work by only discussing saturation of nominal argument positions, but I will apply it to other types of arguments as well (namely, degrees).5 I rephrase Faltz’s generalization in terms of a spell-out condition on phrasal categories VP and AP in Navajo. ‘Original semantic arguments’ are introduced by the lexical head (A, V) and not by subsequent morphology or type-shifting operations. These argument positions must be saturated by the level of VP or AP, although saturating material may not be internal to the adjective or verb word. (26) SPELL-OUT CONDITION: AP and VP can only be spelled out when all of the adjective or verb’s original semantic arguments have been saturated or bound. 5 I do not explicitly extend this part of the theory to event arguments. I assume that event arguments will also be bound within the VP by aspectual morphology that marks verbs and verb stems in Navajo (Young and Morgan 1987, Hale 2003). Modification of the event argument by verb-external temporal adverbials could proceed analogously to modification of AA/PA-marked adjectives by subordinated degree phrases. 8 This condition will be invoked in the analysis of AA/PA-marked adjectives in §5. I introduce it here to locate it in the broader context of work on Navajo verbal structure and meaning. 3.2. Morphological and distributional differences Having introduced a starting set of assumptions about Navajo morphology, syntax, and semantics, we can now examine adjectives. I summarize below Young and Morgan’s (1987) discussion of three sets of morphemes and morphophonological changes found on adjectival verbs (henceforth, adjectives): COMPARATIVE ASPECT (CA), ABSOLUTE ASPECT (AA), and PERFECTIVE ASPECT (PA).6 3.2.1 Comparative Aspect We first examine COMPARATIVE ASPECT (CA). The adjective in (27) shows marking the adjectival stem –nééz ‘tall.’ (27) ’á ní Ø ł nééz CA morphology ⇒ ’áníłnééz PRFX PRFX SUBJ VAL STEM adjectives bear the left-edge prefix [‘á] and, to its left [ní].7 Next comes the subject marker, which is null for third person. To the right of the subject marker is the valence marker [ł], always found on CA-marked adjectives. CA-marked adjectives like (30) are exceptional among adjectives in Navajo because one would never find a CA-marked adjective like ’áníłnééz in ‘isolation’ as it is in (27). We will see in §3.3 that CA-marked adjectives must be preceded by some member of a set of expressions that I will call ‘degree phrases’ (e.g., measure phrases, comparative phrases). This requirement by CA-marked adjectives will set them apart from other adjectives. CA-marked 3.2.2. Absolute Aspect We now turn to ABSOLUTE ASPECT (AA). The adjective in (28) shows the adjectival stem –neez ‘tall.’ (28) ni ø ø neez AA morphology marking ⇒ nineez PRFX SUBJ VAL STEM ‘S/he/it is long/tall.’ Morphologically, AA-marked adjectives are somewhat simpler than CA-marked adjectives. AAmarked adjectives lack a prefix where CA-marked adjectives have [‘á-]. A range of other 6 I refer the interested reader to Bogal-Allbritten (2010) for more discussion of CA, AA, and PA, particularly in the broader context of the Athabaskan language family. Bogal-Allbritten (2010) provides lists of Navajo adjectives. Apparently exhaustive lists can be found in Young and Morgan (1987). 7 The term ‘aspect’ appears to be due to earlier work on Athabaskan languages, which invoked the linear proximity of morphological pieces of CA and AA to true aspectual morphemes (Jeff Leer, p.c.). I do not analyze CA, AA, and PA as true aspectual morphemes. 9 lexically selected prefixes can appear at the left edge of some AA-marked adjectives, but for many AA-marked adjectives, e.g., nineez in (28), the prefix [ni-] is leftmost.8 As with CA-marked adjectives, the valence marker comes to the right of the subject morpheme. AA-marked adjectives always bear the [ø] valence marker. AA morphology has a wider distribution than CA. All adjectival stems that can be CAmarked also have an AA-marked counterpart. Although they differ in vowel tone [ ´] indicates high tone), the stems –neez and –nééz ‘tall’ are clearly related. I analyze them as identical stems. (29) nineez 3SUBJ.tallAA ‘áníłnééz. 3SUBJ.tallCA However, there are many stems for which there is an AA-marked form but not a CAmarked form. The set of adjectives bearing only AA-marking includes color terms (red, black, white) and descriptions of texture (bumpy, smooth). (30) łinishgai 1SG.SUBJ.gaiAA ‘I am white-colored.’ *’áníshgai 1SG.SUBJ.gaiCA 3.2.3 Perfective Aspect We finally turn to PERFECTIVE ASPECT (PA). In contrast with AA- and CA-marked adjectives – which, as classes, always had some prefixes in common – PA-marked adjectives do not systematically bear any special prefixes: prefixes, including valence markers, are idiosyncratic. Young and Morgan (1987) classify PA-marked adjectives on the basis of their stem. PA-marked adjectives are related to eventive verbs denoting changes of state (e.g., heat, bend). The stems of PA-marked adjectives have the shape of eventive verb stems marked for perfective aspect. However, there is no sense in which PA-marked adjectives are synchronically marked for perfective aspect: like all other adjectives, they can be used outside of perfective aspect contexts. Recall that all CA-marked adjectives had AA-marked counterparts, but that there were many AA-marked adjectives for which there was no CA-marked form. Adjectives that take PA morphology have neither CA- nor AA-marked counterparts. (31) a. sido 3SUBJ.hotPA ‘It is hot.’ * ’áníłdo 3SUBJ.hotCA *nido 3SUBJ.hotPA b. shiyish 3SUBJ.bentPA ‘It is bent.’ * ’áshiníłyish 3SUBJ.bentCA * niyish 3SUBJ.bentPA 8 In the presence of additional prefixes, the prefix ni- (VI) disappears when the adjective bears the null third-person subject prefix (Young and Morgan 1987). Young and Morgan (1987) also confirm that the prefix [ni-] is distinct from the morpheme [ní-] found on CA-marked adjectives. 10 Although AA- and PA-marked adjectives are not morphologically related, we will see in §3.3. that they pattern as a set in several respects to the exclusion of CA-marked adjectives. 3.3. Evidence for transitivity and intransitivity in the adjectival domain I will argue that CA-marked adjectives are transitive expressions. They syntactically select a degree phrase. By contrast, there is evidence that AA/PA-marked adjectives do not syntactically select for degree phrase. The remainder of this section outlines four pieces of evidence for this conclusion. The evidence presented here is important for the analysis in §4. 3.3.1 Obligatoriness of degree phrases The fragment in (32) is not grammatical: (32) *Shimá ‘áníłnééz. my.mother 3SUBJ.tallCA (Intended meaning: ‘My mother is tall.’) NO DEGREE PHRASE According to Young and Morgan (1987: g192), CA-marked adjectives cannot appear without one of the degree phrases given in TABLE A and exemplified throughout the paper. TABLE A: Degree Phrases in Navajo Construction Navajo and Gloss a. Comparison of Superiority P-lááh Adj P-beyond Adj b. Comparison of Inferiority P-‘oh Adj P-short.of Adj c. Measure Phrase e.g., naaki ’adeeseez Adj two feet Adj d. Equative NP-gi Adj NP=at Adj e. Intensifier ‘ayóo Adj very Adj f. Degree Pronoun kó-Adj that-Adj g. Degree Question haa-Adj WH-Adj i. Excessive doo shó-Adj da NEG EXCESS-Adj NEG Full Translation Subj is more Adj than P Subj is less Adj than P Subj. is MP Adj. Subj is as Adj as NP Subj is very Adj Subj is that Adj How Adj is Subj? Subj is extremely Adj, Subj is too Adj. The standard markers in comparisons of superiority and inferiority are postpositions: in TABLE A, ‘P’ stands for pronominal object inflection on the postposition. The pronominal inflection matches the standard of comparison (the object of the postposition) in features. These same postpositions are used elsewhere in the language to express physical position. The equative construction also features marking of the standard of comparison with a marker used elsewhere with locative meaning (=gi ‘at’). Semantic shift from locative to degree meaning is not 11 uncommon crosslinguistically (Stassen 1985). Although I gloss these morphemes according to their ‘locative’ meanings in examples, I will posit a degree semantics for them to account for their meanings as parts of degree phrases TABLE A.9 The following examples illustrate the syntax and morphology of degree phrases from TABLE A. Note that the degree phrase is obligatory in all examples. Navajo has both normal comparison of superiority constructions (33a) and subcomparative constructions (33b).10 (33) a. Shimá *(shi-lááh) ‘áníłnééz. my.mother 1SG.OBJ-beyond 3SUBJ.tallCA ‘My mother is taller than I am.’ COMPARISON OF SUPERIORITY b. Díí naaltsoos *(‘eii ’áníłtéél=ígíí yilááh) ‘áníłnééz SUBCOMPARATIVE this book that 3SUBJ.wideCA=REL 3OBJ-beyond 3SUBJ.longCA ‘This book is longer than that one is wide.’ b. Shimá *(shí=gi) ‘áníłnééz. my.mother 1SG=at 3SUBJ.tallCA ‘My mother is as tall as I am.’ EQUATIVE c. Shimá *(kó)-níłnééz. my.mother that-3SUBJ.tallCA ‘My mother is that (gesture, pointing) tall.’ The obligatoriness of degree phrases with Navajo eventive verbs (34). (34) CA-marked DEGREE PRONOUN adjectives sets them apart from a. Naalnish. 3SUBJ.work ‘He/she works.’ b. (Biyáázh Kinłání=di) naalnish. his/her.son Flagstaff=at 3SUBJ.work ‘His/her son works in Flagstaff.’ The requirement for a degree expression to be present also sets CA-marked adjectives apart from AA/PA-marked adjectives. The adjectives in (35) are well-formed on their own. (35) a. Nineez b. Deesdoi 9 The constructions in (f), (g), and (i) lack the left-edge [‘á-] prefix that is otherwise part of CA morphology. I assume that the absence of ‘á is due to morphological competition between the degree phrase and [‘á-]. 10 The subcomparative has a CA-marked adjective in both clauses. This is obligatory (§5.3). The embedded CAmarked adjective lacks an overt degree phrase but I argue (§5.1) that this position is filled by a null operator, on analogy with analyses for subcomparatives in other languages. Given the availability of subcomparatives with –lááh I will give a clausal analysis of degree constructions. I recognize the possibility of a phrasal analysis (e.g., Bhatt and Takahashi 2011), but leave further study of motivation for a phrasal analysis of –lááh to future work. 12 3SUBJ.tallAA ‘S/he/it is tall.’ 3SUBJ.hotPA ‘It is hot.’ When not modified, AA- and PA-marked adjectives appear to have the semantics of POSmarked adjectives in English (§2.2). The subject exhibits the property denoted by the adjective to a degree exceeding a contextual norm or standard of comparison. (36) a. My mother is below average height for a woman of her age. # (35a) b. Context: Barstow is below average temperature for this time of year. # (35b) The requirement for a degree expression is the first way in which CA- and AA/PA-marked adjectives pattern differently. Below, we will see that even when AA/PA-marked adjectives occur with a degree expression, they still pattern differently from CA-marked adjectives. These pieces of evidence will be added together to support the view that CA-marked adjectives – but not AA/PA-marked adjectives – syntactically select for a degree expression. 3.3.2 Subordinated degree phrases Although AA/PA-marked adjectives do not have to be modified by degree phrases, they can be. However, a special syntactic configuration is required (Young and Morgan 1987: g193). I defend the constituency I give below in §3.3.3. The degree expression must be marked by the morpheme =go, which has been variably referred to as an adverbializer and as a clausal subordinator (Schauber 1979). I gloss =go as SUB since we will see examples in which it clearly functions to mark a clause as subordinate to the matrix clause. (37) a. Shideezhí [ [shi-lááh ‘át’ée]=go ] nizhóní. my.little.sister 1SGOBJ-beyond 3SUBJ.be=SUB 3SUBJ.prettyAA ‘My little sister is prettier than I am.’ b. Shideezhí [ [ shí=gi ‘át’ée=]go ] nizhóní. my.little.sister 1SGOBJ=at 3SUBJ.be=SUB 3SUBJ.prettyAA ‘My little sister is as pretty as I am.’ c. *Shideezhí [ shi-lááh ] nizhóní. my.little.sister 1SGOBJ-beyond 3SUBJ.tallAA In (37a,b), the degree expression is also followed by the copula ‘át’é ‘s/he/it is.’ Most of the examples from speakers contained ‘át’é. The copula appears to be a ‘dummy’ form, in a sense. It is always marked for a singular third-person subject regardless of the person features marking, e.g., the standard of comparison. Additionally, ‘át’é is optional for some degree phrases (e.g., comparison of superiority –lááh (38a)), and required for others (e.g., equative =gi (38b)).11 (38) a. Shideezhí [ [ shiláah]=go] nizhóní my.little.sister 1SGOBJ-beyond 3SUBJ.prettyAA 11 The basic form of the copula is ‘át’é. When ‘át’é is marked by =go, the vowel lengthens and acquires falling tone. This is a typical effect for ‘át’é (Young and Morgan 1987). 13 ‘My little sister is prettier than I am.’ b. * Shideezhí [ [ shí=gi]=go] nizhóní my.little.sister 1SGOBJ=LOC=SUB 3SUBJ.prettyAA (Intended meaning: ‘My little sister is as pretty as I am.’) The copula is also obligatory for degree expressions from TABLE A that prefixed directly onto the left edge of CA-marked adjectives. The degree question exemplifies this class of construction. As shown, haa is now prefixed onto the left edge of the copula. Haa alone cannot bear =go. (39) Kinłání hait’ée=go Flagstaff WH-3SUBJ.be=SUB ‘How hot is Flagstaff?’ deesdoi? 3SUBJ.hotPA In my analysis, I will not make much of the role of ‘át’é. The crucial piece of morphology appears to be the subordinator =go. As shown, =go adjoins to the copula (where present) or to the degree phrase directly. There are two observations to make about =go. First, CA-marked adjectives cannot be preceded by degree phrases marked by =go: (40) *Shimá my.mother shi-lááh ‘át’ée=go 1SGOBJ-beyond 3SUBJ.be=SUB ‘áníłnééz. 3SUBJ.tallCA Second, marking of material by =go may be indicative that the material is modifying a full clause. As shown by the sentences below, eventive verbs can be modified by full, -marked clauses – if-clauses (a), manner clauses (b), and temporal clauses (c). (41) a. Łééchąą’í yah ’anáánáádzáa=go ch’íhidííłhą́ą́ł dog in 3SUBJ.comes.again=SUB 3OBJ.2SGSUBJ.throw.out ‘[If the dog comes in again], throw him out.’ (Young and Morgan (YM) 1987: d284) b. Shí t’óó ’ááłk’iis yisháął=go shizhé’é tsékooh gódeg ha’asbą́ą́z 1SG beside 1SGSUBJ.walk=SUB my.father canyon up 3OBJ.3SUBJ.drive.out ‘My father drove up out of the canyon [with me walking alongside].’ (YM 1987: d2) c. Hooghan=di nánísdzáa=go shi’éétsoh home=at 1SGSUBJ.returned=SUB my.coat ‘When I got back home, I took my coat.’ ’ádaa diistsooz 3OBJ.1SGSUBJ-take.it.off (YM 1987: d344) I take the data in (41) as evidence that =go-marking is a syntactic requirement when clauses modify other clauses. The marking of degree phrases marked by =go when they modify AA/PAmarked adjectives suggests that AA/PA-marked adjectives are full clauses. By contrast, the inability of degree phrases to be =go-marked when the adjective is CA-marked suggests that in the absence of a degree expression, CA-marked adjectives are not full clauses. 14 3.3.3 Degree phrases modifying from the adjective:12 (42) Syntactic (in)flexibility of degree expressions AA/PA-marked adjectives are permitted to undergo movement away a. [Shí=gi ‘át’ée=go] shideezhí nizhóní. 1SGOBJ=at 3SUBJ.be=SUB my.little.sister 3SUBJ.prettyAA ‘My little sister is as pretty as I am.’ b. K’ad [ hosiyilts’į́į́łígíí bi-lááh ’át’ée=go ] chidí naat’a’í now speed.of.sound 3OBJ-beyond 3SUBJ.be=SUB airplane dadilwo’ jiní. 3PLSUBJ.fastAA REPORT ‘It is said that airplanes are faster than the speed of sound.’ (adapt. YM 1987: d458) The examples in (42) support the constituency that I assumed earlier. The copula and subordinator =go form a constituent with the degree expression rather than with the adjective: the copula and subordinator can be fronted with the degree expression, leaving the AA/PA-marked adjective in sentence-final position. In addition, the sentences in (42) provide another contrast between CA- and AA/PAmarked adjectives. Compare the grammatical (42a) with the ungrammatical (42b). (43) *[Shimá=gi] my.mother=at shideezhí ’áníłnééz my.little.sister 3SUBJ.tallCA The positional flexibility of degree phrases with AA/PA-marked adjectives recalls the positional flexibility of locative and temporal modifiers of eventive verbs (§3.1). Once again, AA/PAmarked adjectives pattern with eventive verbs to the exclusion of CA-marked adjectives, which impose stricter positional restrictions on degree phrases. 3.3.4 Morphological (in)-transitivity of adjectives So far, we have seen that CA-marked adjectives require degree phrases and stand in a tight syntactic relationship with the degree phrase: no movement of the degree phrase is permitted. By contrast, the syntactic relationship between AA/PA-marked adjectives is much looser and they have ‘complete’ meanings even in the absence of a degree phrase. This evidence is suggestive of the theory that CA-marked adjectives syntactically select for a degree phrase while AA/PA-marked adjectives do not. I assume that all adjectives – regardless of their morphological marking – 12 Although Young and Morgan ‘s (1987) dictionary entries contain examples like (46b), they do not discuss or describe the differences in positional flexibility. 15 select for a subject argument. Thus, CA-marked adjectives are posited to be syntactically transitive while AA/PA-marked adjectives are syntactically intransitive. Comparison of AA and CA morphology provides additional support. Recall that every CAmarked adjective has an AA-marked counterpart. (44) gives a minimal pair for comparison. (44) a. ’á ní ø ł nééz ⇒’áníłnééz PRFX PRFX SUBJ VAL STEM b. ni ø ø neez ⇒ nineez PRFX SUBJ VAL STEM ‘S/he/it is tall.’ CA-marked adjectives bear the valence marker [ł] while their AA-marked counterparts bear the null valence marker [ø]. In §3.1, I mentioned that the [ł] marker is often taken to indicate a transitive argument structure, while the [ø] marker is taken to indicate an intransitive argument structure (Young and Morgan 1987; Hale 2003). The minimal pair in (45) illustrate. (45) a. yoo ø ł mas ⇒ yoołmas OBJ SUBJ VAL STEM ‘S/he rolls it.’ b. yi ø PRFX ø mas ⇒ yimas SUBJ VAL STEM ‘It is rolling.’ The fact that CA-marked adjectives bear [ł] while AA-marked adjectives bear [ø] can be added to our current stock of evidence that CA-marked adjectives are transitive, selecting both degree phrase and a subject as argument. By contrast, AA/PA-marked adjectives are intransitive, selecting only a subject as argument. Once it composes with a subject, the AA/PA-marked adjective is a full clause: further modification by a degree phrase is only possible by means of syntactic structure found for clause-level modification. 4. Syntactic Intransitivity Does Not Entail Semantic Intransitivity The evidence collected in §3 supported a particular claim about the syntax of CA- vs. AA/PAmarked adjectives. I argued that CA-marked adjectives are transitive: they syntactically select for both a degree phrase (a degree) and a subject (an individual). AA/PA-marked adjectives are syntactically intransitive and only select for a subject (an individual). My goal is not to stop at this syntactic claim, however, but instead to propose an analysis that accounts for both the syntax and semantics of CA- and AA/PA-marked adjectives. In this section, I begin to address the semantics. One obvious direction to go would be to argue that AA/PA-marked adjectives are semantically intransitive. Following Klein (1980), we could argue that AA/PA-marked adjectives are type <et> vague predicates. (46) [[tall]]c = λx.x counts as tall in c 16 This approach to adjectival meaning has garnered increasing attention in the literature. Kennedy (2007) and Beck et al. (2009) both provide diagnostics or metrics by which one might determine whether a language’s adjectives have a degree semantics or a vague semantics. Three diagnostics for vague predicate meaning are: (47) a. Do not occur with specialized degree morphology (Beck et al. 2009). b. Cannot be used in comparatives in ‘crisp judgment’ scenarios (Kennedy 2007). Both diagnostics hinge on the fact that vague predicates require context manipulation – rather than degree argument manipulation – in order to participate in comparative constructions. These diagnostics are applied to Motu (Malayo-Polynesian) and Washo (Native American isolate) by Beck et al. (2009) and Bochnak (to appear), respectively. The authors convincingly motivate a degreeless analysis of each language. By contrast, as discussed below, Navajo AA/PA-marked adjectives fail both tests for vague predicate meaning. I will argue that this suggests that they have a degree semantics. In other words, all adjectives (whether CA- or AA/PA-marked) take both an individual and a degree as semantic arguments, regardless of the number of arguments that they syntactically select for. First, languages whose adjectives have vague predicates canonically do not occur with specialized degree morphology. For instance, Washo uses a conjunctive strategy of comparison: (48) t’é:liwhu delkáykayiʔ k’éʔi šáwlamhu delkáykayiʔ=é:s k’áʔaš man 3SUBJ.tall is girl 3SUBJ.tall=NEG is ‘The man is taller than the girl.’ Lit: ‘The boy is tall, the girl is not tall.’ (Bochnak, to appear) Both clauses contain the form of the adjective that are found in simple predicative sentences like ‘The man is tall.’ Such sentences in Washo – like their English counterparts – appear to be true only if the subject (‘the man’) exceeds the contextual standard of comparison for tallness. In conjunctive comparison constructions like (48), the first clause asserts that the subject of comparison (‘the man’) meets the standard for tallness in the context. The second clause asserts that the standard of comparison (‘the girl’) does not meet this standard. Kennedy (2007) and Bochnak (to appear) analyze the second clause of a conjunctive comparison as introducing who or what (in addition to the subject) belongs in the context of comparison. In (48), the context of comparison contains both ‘the man’ and ‘the girl.’ The property of being ‘tall’ is evaluated with respect to this reduced context c. The sentence in (48) is true if the boy ‘stands out’ as tall with respect to the girl. As we have already seen, both CA- and AA/PA-marked adjectives occur with the same set of degree phrases listed in TABLE A.13 A minimal pair is presented for comparison. Conjunctive comparison is not utilized by AA/PA-marked adjectives. (49) a. Shimá shi-lááh ‘áníłnééz. my.mother 1SGOBJ-beyond 3SUBJ.tallCA ‘My mother is taller than I am.’ 13 Two exceptions to this generalization appear to be that AA/PA-marked adjectives do not occur in subcomparatives, and they do not occur with measure phrases. I discuss these restrictions in §5.3 and §5.4. 17 b. Shideezhí shi-lááh ‘át’ée=go nizhóní. my.little.sister 1SGOBJ-beyond 3SUBJ.be=SUB 3SUBJ.prettyAA ‘My little sister is prettier than I am.’ Although the standard marker –lááh also can be used in locative phrases, it appears to have been adapted – like locative markers in many languages (Stassen 1985) – to function as specialized degree morphology. If we wanted to argue that AA/PA-marked adjectives had vague predicate meanings, we would be forced either to posit a vague predicate analysis of all adjectives in Navajo, or to argue that the two instances of –lááh have a different semantics. The first option is not tenable and the latter option is not parsimonious. The second strike against a vague predicate analysis for Navajo comes from the absence of crisp judgments in constructions like (49b). I follow Kennedy (2007) in identifying ‘crisp judgment effects’ as the property of a comparison being infelicitous where the difference in the subject and standard of comparison is very small. The source of crisp judgment effects is the entry in (46) and the context manipulation required to produce comparative meanings. If a construction like (48) requires ‘the boy’ to stand relative to ‘the girl’ in terms of height, then the difference in their heights must be relatively significant. In contexts where differences in height are minimal, sentences like (48) and (50) are infelicitous (Bochnak to appear): (50) Context: Two ladders, one only slightly taller than the other. ?? wí:diʔ t’éweʔ dewɡíʔiš k’éʔi wí:diʔ t’éweŋa dewgiʔiš=é:s k’éʔaš this much height is this much height=NEG is (Intended meaning: ‘This one is taller than that one.’ (Bochnak to appear) Navajo does not exhibit crisp judgment effects when adjectives are AA/PA-marked. The comparisons of superiority in (51) and (52) were felicitous even when the difference between the two objects was very small. (51) Context: This bottle of water is 64°F. That bottle of water is 66°F. Díí tó [ [ ‘eii yilááh ‘át’ée]=go] sik’az this water [ that.one 3OBJ-beyond 3SUBJ.be=ADV] 3SUBJ.coldPA ‘This water is colder than that one (water).’ (52) Context: This bottle of water is 66°F. That bottle of water is 64°F. Díí tó [ [ ‘eii yilááh ‘át’ée]=go] deesdoi this water that.one 3OBJ-beyond 3SUBJ.be=SUB 3SUBJ.hotPA ‘This water is warmer (hotter) than that one (water).’ In fact, comparative constructions in Navajo appear to be true in the same types of verifying scenarios more generally, regardless of the morphology that marks the adjective. When used outside of degree constructions, AA/PA-marked adjectives – like adjectives in Washo – require the subject to exceed the contextual standard of comparison. (53) a. Nineez 3SUBJ.prettyAA b. Deesdoi 3SUBJ.hotPA 18 ‘S/he/it is pretty.’ ‘It is hot.’ When AA/PA-marked adjectives appear in the comparison of superiority construction, the subject is no longer required to exceed the contextual standard of comparison. (54) a. (51), Consultant comment: “This sounds fine even though they’re not actually cold.” b. Context: We’re looking at rugs at a trading post. There is a wide range of rugs, from very attractive to very unattractive. You hold up two rugs that are not good-looking: their wool is dyed in strange colors and the weaving is not very straight or complex. One is of slightly better quality than the other. I ask what you think of the two rugs. Díí diyógí [‘eii diyógí yilááh ‘át’ée=go] nizhóní this rug [that rug 3OBJ-beyond 3SUBJ.be=SUB] 3SUBJ.prettyAA ‘This rug is prettier than that rug.’ Consultant comment: “Nizhóní is relative to what you’re comparing. If you have two ugly rugs, then you’re just saying that this one is better than the other one.” The data presented so far suggest that AA/PA-marked adjectives should receive a degree semantics. Like CA-marked adjectives, they take both an individual and a degree as argument. The only analytical difficulty presented by this conclusion presents us is that we have seen evidence that they are syntactically intransitive, in contrast with syntactically transitive CAmarked adjectives. In the next section, I address this discrepancy in the semantic and syntactic argument structure of AA/PA-marked adjectives. 5. Implementing the Analysis 5.1. The analysis of CA In §3, I argued that CA-marked adjectives syntactically projected both a degree argument and an individual argument. I posit the following syntactic frame for CA-marked adjectives. The CAmarked adjective syntactically selects for a degree phrase. The DP subject is assumed to be realized as subject pronominal morphology. I assume that the subject composes first with the stem due to the order of the subject prefix relative to any degree phrase: the subject always comes closer to the stem. (55) AP 4 DegP A 3 DP ACA subject d The denotation of a CA-marked adjective reflects the order of composition shown above:14 14 In this section, I only analyze adjectival down to the level of A, where CA, AA, and PA have already composed with the adjectival stem. I return to a further decompositional analysis in §6.1. 19 (56) [[ACA]] = λxλd.x is A to d The semantics and syntax of a CA-marked adjective are essentially the same as they are under the standard analysis (§2). The only difference is the order of the individual and degree arguments. From the perspective of linear order of prefixes in Navajo, composition of the individual first and the degree second is more appealing: the subject prefix occurs closer to the stem than the degree phrase will. This portion of the analysis could likely revert to more standard assumptions. Given that CA-marked adjectives have the semantics and syntax of adjectives under the standard analysis, it will come as no surprise that the analysis of CA-marked adjectives in degree constructions will be familiar as well. As in English, the internal degree argument can either be saturated by a degree-denoting degree phrase (e.g., a measure phrase), by a trace of a quantificational degree phrase that has undergone quantifier raising, or by a null operator. I analyze the subcomparative as containing a null operator in the embedded clause (Beck 2010). I illustrate the analysis for the subcomparative below: (57) Díí naaltsoos [DegP OP ’áníłtéél=ígíí yilááh ] book 3SUBJ.wideCA=REL 3OBJ.beyond ‘This book is longer than it is wide.’ DET ’áníłnééz 3SUBJ.longCA I assume that Navajo –lááh, like English MORE, is of type <dt<dt,t>>. I continue to adopt an A15 NOT-A analysis of standard markers (Schwarzschild 2008). (58) [[-lááh]] = λD1λD2∃d.D2(d) & ¬D1(d) I assume the following syntax for the subcomparative; a null operator fills the embedded DegP.16 (59) Díí naaltsoos [DegP ‘eii ’áníłtéél=ígíí yilááh ] ‘áníłnééz this book that.one 3SUBJ.wideCA=REL 3OBJ-beyond 3SUBJ.longCA ‘This book is longer than that book is wide.’ [ [DegP [AP that.book d-wideCA] beyond] 1 [AP [DP this.book ] t1 [A longCA] ] ] AP2 qp DegP A qp 3 AP1 Deg DP ACA 4 -lááh this book longCA DegP1 A1 OP 3 DP ACA 15 I assume that comparative deletion applies in cases of regular comparison of superiority. As long as we have an account of how subcomparatives can compose with CA-marked adjectives, I assume that an account of regular comparisons of superiority will follow. 16 The adjective-internal subject positions are actually occupied by subject pronominals, at least in the surface representation. I include the full noun phrases in the derivation for perspicuity. 20 that book wideCA a. [[A1]] = λd.wide(that.book)(d) b. [[DegP1]] = [λD1λD2∃d.D2(d) & ¬D1(d)](λd.wide(that.book)) = λD2∃d.D2(d) & ¬wide(that.book)(d) = λd.long(this.book) c. [[A2]] d. [[CP]] =[λD2∃d.D2(d) & ¬wide(that.book)(d)](λd.long(this.book)) = ∃d.long(this.book.)(d) & ¬wide(that.book)(d) “There exists a degree d to which this book is long and to which that book is not wide.” Outside of the subcomparative, the adjective does not appear in the lower comparative. One way to analyze this construction is to assume that comparative deletion applies to delete the lower copy (Bresnan 1973). In the next section, when we discuss comparison of superiority constructions with AA/PA-marked adjectives, we will see an alternative conceptualization of how adjectives in the lower clause can fail to be pronounced. I have proposed that the DegP must merge directly into the specifier position of AP if the adjective is CA-marked. This requirement explains the tight positional restrictions that we saw CA-marked adjectives impose on degree phrases in §3.3.3. So far, we have phrased the tight relationship between CA-marked adjectives and DegP entirely in terms of the syntactic selectional properties of the adjective: a CA-marked adjective selects for a DegP to be merged into precisely that position. However, a stronger statement might be that in addition to CA-marked adjectives requiring that a DegP occupy their specifier position, the DegP can only (for morphosyntactic reasons) only occur in this syntactic position. A DegP cannot adjoin higher in the syntax. A number of the degree phrases that we have seen are, in fact, even morphologically bound to the left edge of the CA-marked adjective (the degree question and degree pronoun constructions). Given the existence of prefixal DegP, it is a relatively small step to propose that DegP must, as a class, merge within AP. 5.2. Analysis of AA/PA-marked adjectives As argued in §4, AA/PA-marked adjectives have a degree semantics. I posit the denotation in (60) for AA/PA-marked adjectives. This is the same denotation posited for CA-marked adjectives. (60) [[AAA/PA]] = λxλd.x is A to d Unlike CA-marked adjectives, AA/PA-marked adjectives are syntactically intransitive, taking only the subject as argument. (61) AP 3 DP AAA/PA subject d 21 At the end of discussion of CA-marked adjectives above, I proposed that it is not only the case that CA-marked adjectives select for a DegP, but it is also the case that a DegP can only occur within AP. This reasoning permits us to make a prediction. If an adjective lacked a position with AP for DegP to merge into, DegP could not merge above the level of AP, even if the semantic types were compatible. Additional syntactic structure might permit merger of DegP higher in the clausal structure. This seems to be the case for AA/PA-marked adjectives. The AP of an AA/PA-marked adjective is of type <dt> and thus is compatible with the type <dt,t> DegP. However, given the structure in (61), there is no position for DegP to merge into. Navajo has a way to solve this syntactic problem, however. Recall that in §3.3.2, we saw that AA/PA-marked adjectives can be modified by degree phrases if the degree phrase is embedded within the same syntactic structure found for clausal modification for generally. The DegP in (62a) and the temporal clausal modifier in (62b) are both marked by the subordinator =go. (62) a. [[Shí=gi ‘át’ée]=go] shideezhí nizhóní. 1SGOBJ=at 3SUBJ.be=SUB my.little.sister 3SUBJ.prettyAA ‘My little sister is as pretty as I am.’ b. [[Hooghan=di nánísdzáa]=go]] shi’éétsoh home=at 1SGSUBJ.returned=SUB my.coat ‘When I got back home, I took my coat.’ ’ádaa diistsooz 3OBJ.1SGSUBJ-take.it.off (Young and Morgan 1987: d344) At this point, we know that at the AP level, AA/PA-marked adjectives are expressions of type <dt>. They cannot directly combine with the DegP for the reasons outlined above, but a subordinated DegP can adjoin at a higher level. I argue, however, that the AP cannot remain a function of type <dt> (that is, with its degree argument unsaturated) until the point in the clausal syntax at which the subordinated DegP can compose with AP. Recall my claim about the spellout of Navajo adjectives and verbs introduced in §3.1. This claim was based on Faltz’s observation that Navajo verbs are always propositional by the time that they are spelled out. (63) SPELL-OUT CONDITION: AP and VP can only be spelled out when all of the adjective or verb’s original arguments have been saturated or bound. The adjective’s ‘original’ arguments are the degree argument and the individual argument. I define an ‘original’ argument as an argument introduced by the head of a lexical projection. The structure in (61) violates the spell-out condition. The degree argument is unsaturated. But, given the syntax of the AP, a DegP cannot be merged prior to spell out of AP. I propose a type-shifting mechanism similar to Existential Closure (Kamp 1981, Heim 1982, Diesing 1992). The mechanism CLOSE applies to bind the degree argument of the AA/PAmarked adjective. (64) [[CLOSE]] = λAdtλDdt∀d.D(d) A(d) “For all degrees d, if d is in the domain of D (the set of degrees that holds of the overt standard of comparison), then d is also in the set of degrees that holds of the adjective applied to the subject).” 22 The operation CLOSE is closely related to Schwarzschild’s (2010, 2011) proposal for the semantics of comparative morphemes in languages in which standard of comparison phrases (e.g., equivalents to than-phrases) are quantifier domain adverbials.17 As in Schwarzschild’s proposal, the domain variable D can be restricted by degree phrases: this will permit the degree argument of AA/PA-marked adjectives to be manipulated indirectly through domain restriction (von Fintel 1994). Once the type-shifting operation of CLOSE has applied to AP, the requirements of the spell-out condition are met. Both of the original arguments of the AA/PA-marked adjective are now closed. It is true that a new variable (D) had to be introduced in order to accomplish this task, but since D is not a semantic argument of the adjectival stem, no problem arises. In the simplest case – where AA/PA-marked adjectives are not further modified by degree phrases – the domain variable D is contextually set to equal the interval of degrees that surrounds the contextual standard of comparison.18 (65) (66) Shimá nizhóní. my.mother 3SUBJ.prettyAA ‘My mother is pretty.’ (True iff my mother is pretty in excess of some contextual standard of beauty.) AP 3 CLOSE 3 DP A my mother dprettyAA a. [[A]] = λxλd’.x is pretty to d’ b. [[AP]] = λd’.my mother is pretty to d’ c. CLOSE applies: = [λAdtλDdt∀d.D(d) A(d)] (λd’.my mother is pretty to d’) = λDdt∀d.D(d) my mother is pretty to d d. Context fills in D as the standard of comparison = ∃D’. where D’ is the contextual norm for height, ∀d.D’(d) my mother is pretty to d “For all degrees d in the interval corresponding to the contextual standard of beauty in c, my mother is pretty to d.” 17 Schwarzschild (2011) focuses on application of the program to Hebrew, but both Schwarzschild (2010) and (2011) note the applications of the program to Navajo. My application of the analysis to Navajo differs from his in that I (a) separate the contribution of a CLOSE-like morpheme or operation from the semantics of AA/PA itself and (b) do not encode norm-relatedness into the meaning of my morpheme CLOSE (or, into the meaning of AA/PA, which Schwarzschild analyzes as POS). As we saw at the end of §4, we do not want to encode the requirement that the subject always exceed the contextual standard (norm-relatedness; Bierwisch 1989, Rett 2008) into the basic meaning of an AA/PA-marked adjective. 18 I assume here that it is possible to conceive of the ‘contextual standard of comparison’ not as a single degree, as is usually assumed, but as an interval of degrees. There are similar proposals made elsewhere in the lierature. For instance, Bierwisch proposes that gradable adjectives are associated with a comparison class C, which occupies a continuous segment of the scale (Bierwisch 1989: 119). 23 With all of the pieces in place, we now turn to the derivation of a comparative construction. I do not show subcomparative here as I did for CA-marked adjectives because, as discussed in §5.3, subcomparatives are not possible for AA/PA-marked adjectives. In the regular comparison of superiority construction in Navajo as in English, the adjective in the standard of comparison (DegP) is not pronounced. While we can follow much previous work and assume deletion happens through comparative deletion, our system for Navajo actually predicts that we will never see the embedded adjective. This is because given the discussion above about Navajo’s spell out condition, the only way in which an AA/PA-marked adjective can be spelled out is if it is marked by CLOSE. As I show in §5.3 when I return to the unavailability of subcomparatives with AA/PA-marked adjectives, if the embedded adjective is marked by CLOSE a type mismatch between the adjective and the comparative morpheme – defined as before – will result. If the adjective is not CLOSEmarked, it will be of the right type but cannot be spelled out. This is a positive result, since this instance of the adjective is not overt. The derivation in (68) illustrates composition of the sentence in (67). (67) Phoenix [CP3 [CP1 [DegP Flagstaff yilááh ] ‘át’ée=go] [CP2 [AP deesdoi ] ] ] Phoenix Flagstaff 3OBJ.beyond 3SUBJ.be=SUB 3SUBJ.hotPA ‘Phoenix is hotter than Flagstaff.’ There are a few differences between the structure in (68) and the structure I proposed for CAmarked adjectives. Here, clausal structure (CP) mediates between the matrix AP and the subordinated DegP; and the two CPs (the matrix AP and the DegP) combine through Predicate Modification. The denotation of the Deg head [-lááh] is the same as it was when it combined with CA-marked adjectives, however. (68) CP3 q p CP1 CP2 rp 4 VP C AP2 C rp=go 3 ø DegP V CLOSE 3 rp ‘át’é DP APA AP1 <dt> Deg Phoenix deesdoi 4 -lááh DP APA Flagstaff deesdoi a. [[AP1]] = [λxλd.hot(x)(d)](Flagstaff) = λd.hot(Flagstaff)(d) b. [[DegP]] = [λD1λD2∃d.D2(d) & ¬D1(d)](λd.hot(Flagstaff)(d)) = λD2∃d.D2(d) & ¬hot(Flagstaff)(d) c. [[CP1]] = [[VP]] = [[DegP]] 24 d. [[A2]] = [λxλd.hot(x)(d)](Phoenix) = λd.hot(Phoenix)(d) e. [[AP2]] = [[POS]](λd.hot(Phoenix)(d)) = [λAdtλDdt∀d.D(d) A(d)](λd.hot(Phoenix)(d)) = λDdt∀d.D(d) hot(Phoenix)(d) f. [[CP2]] = [[AP2]] g. [[CP3]] = λD’. [ [[CP1]] & [[CP2]] ]D’) = λD’[λD2∃d.D2(d) & ¬hot(Flagstaff)(d)](D’) & [λDdt∀d.D(d) hot(Phoenix)(d)](D’) = λD’[∃d.D’(d) & ¬hot(Flagstaff)(d)] & [∀d.D’(d) hot(Phoenix)(d)] = ∃D’[∃d.D’(d) & ¬hot(Flagstaff)(d)] & [∀d.D’(d) hot(Phoenix)(d)] “There exists an interval of degrees, D’. There exists a degree d in D’ which is a degree to which Flagstaff is not-hot. For all degrees d, if d is in D’, then Phoenix is hot to d.” Rephrased slightly, the interval D’ is such that all degrees in D’ are degrees to which Phoenix is hot. There exists at least one degree in D’ to which Phoenix is hot but Flagstaff is not hot. The situation in (69) would verify these truth conditions: (69) It is 90 degrees in Phoenix today. It is 89 degrees in Flagstaff today. Degrees to which Phoenix is hot: …87, 88, 89, 90] Degrees to which Flagstaff is not hot: [90, 91, 92, … There exists a degree (d=90) to which Phoenix is hot and to which Flagstaff is not-hot. This concludes the introduction and exemplification of the analysis for AA/PA-marked adjectives. The next section demonstrates how the analysis can capture welcome predictions about the unavailability of subcomparatives with AA/PA-marked adjectives. §5 is wrapped up with a discussion of apparently ‘exceptional’ AA-marked adjectives, which are not amenable to the analysis as proposed so far. 5.3. AA/PA-marked adjectives do not permit subcomparatives We have seen that subcomparatives are grammatical when both adjectives are CA-marked. Speakers reported that if either, or both, adjective were AA/PA-marked, the result was ungrammatical: (70) a. Díí naaltsoos [ ‘eii ’áníłtéél=ígíí yilááh ] ‘áníłnééz this book that 3SUBJ.wideCA=REL 3OBJ-beyond 3SUBJ.longCA ‘This book is longer than that one is wide.’ [CA] CA b. * Díí naaltsoos [ [‘eii niteel=ígíí yilááh this book that 3SUBJ.wideAA=REL 3OBJ-beyond [AA] CA 25 ‘át’ée]=go] nineez 3SUBJ.be=SUBJ 3SUBJ.longAA c. * Díí naaltsoos [ ‘eii niteel=ígíí yilááh ] ‘áníłnééz this book that 3SUBJ.wideAA=REL 3OBJ-beyond 3SUBJ.longCA [AA] CA We can account for the ungrammaticality of (70b) and (70c) by combining our account of and the Navajo spell-out condition. As outlined above, if an AA/PA-marked adjective has been spelled out, then it must bear CLOSE. The critical type mismatch comes at the point where Deg lááh composes with AP1. CLOSE (71) CP3 q p CP1 CP2 rp 4 VP C AP2 C rp=go 3 ø DegP V CLOSE 3 rp ‘át’é DP APA AP1 <dt> Deg this book 3 -lááh CLOSE 3 DP APA that book a. [[A]] = [λxλd.wide(x)(d)](this book) = λd.wide(this book)(d) b. [[AP1]] = [[CLOSE]](λd.wide(this book)(d)) = [λAdtλDdt∀d.D(d) A(d)](λd.wide(this book)(d)) = λDdt∀d.D(d) wide(this book)(d) c. [[DegP]] is of type <dt,<dt,t>> and cannot compose with the type <dt,t> AP1. We predict that any configuration in which the AA-marked adjective is required to appear in the standard of comparison will be ungrammatical. However, our account does not have an explanation for the ungrammaticality of structures like (72), where the embedded adjective is CAmarked and the matrix adjective is AA/PA-marked. Our system predicts that (72) should be grammatical from a type-theoretic perspective. (72) *? Díí naaltsoos [ [‘eii this book that ‘át’ée]=go] 3SUBJ.be=SUBJ 5.4. ’áníłtéél=ígíí yilááh 3SUBJ.wideCA=REL 3OBJ-beyond nineez 3SUBJ.longAA [CA] AA Exceptional AA-marked adjectives 26 The source of the unacceptability of (72) may be part of a larger issue for a subset of AA-marked adjectives: AA-marked adjectives for which there exists a CA-marked counterpart that speakers frequently use, e.g., nitsaa ‘s/he/it is big’ (cf. ’áníłtso).19 These ‘exceptional’ AA-marked adjectives are judged odd in a range of constructions that our system predicts should be grammatical, and which are grammatical when they contain ‘unexceptional’ AA/PA-marked adjectives (i.e., AA/PA-marked adjectives for which there is no CA-marked counterpart). These constructions include the degree question (73) and the equative (74). I show both constructions with a CA-marked adjective, an unexceptional AA/PA-marked adjective, and an exceptional AAmarked adjective.20 (73) a. Haa-níłtso? WH-3SUBJ.bigCA ‘How big is s/he/it? What size is s/he/it?’ b. Ha-it’ée=go deesdoi? WH-3SUBJ.be=SUB 3SUBJ.hotPA ‘How hot is it?’ CA Unexceptional AA/PA c. ?? Ha-it’ée=go nitsaa? WH-3SUBJ.be=SUB 3SUBJ-bigAA ‘In what way is s/he/it big?’ (big? wide? tall?) (74) a. Shimá=gi ‘áníłtso my.mother=at 3SUBJ.bigCA ‘I am as big as my mother,’ ‘My mother and I are the same size.’ b. Shideezhí shimá=gi ‘át’ée=go nizhóní my.little.sister my.mother=at 3SUBJ.be=SUB 3SUBJ.prettyAA ‘My little sister is as pretty as my mother.’ Exceptional AA CA Unexceptional AA/PA c. ?? Shideezhí shimá=gi ‘át’ée=go nitsaa Exceptional AA my.little.sister my.mother=at 3SUBJ.be=SUB 3SUBJ.bigAA ‘My little sister is big in the same way that my mother is.’ Consultant Comment: “You’re saying they’re big, and maybe they’re both heavy or tall. They’re big in the same way.’ 19 I add “…that speakers frequently use” here since a few AA-marked adjectives which fall into the ‘unexceptional’ camp are reported by Young and Morgan (1987) to have CA-marked forms (e.g., nizhóní ‘s/he/it is pretty’, ‘ánóoshóní). However, my impression from discussion with speakers is that these CA-marked forms are not frequently used compared to their AA-marked counterparts. In addition, there was disagreement between consultants about what morphological form (if any) is taken by the CA-marked adjective in current speech. 20 Measure phrases are also judged ungrammatical with AA-marked adjectives, while they are grammatical with CAmarked adjectives. We cannot compare the behavior of exceptional AA-marked adjectives to other AA/PA-marked adjectives, however: measure phrases in Navajo only occur with the kind of adjective (e.g., ‘tall’, ‘wide’, ‘heavy’) that is CA-marked. This restriction of measure phrases is quite common crosslinguistically (Schwarzschild 2005, Svenonius and Kennedy 2006). 27 The semantics of ‘degree’ constructions with exceptional AA-marked adjectives merits much more discussion and examination. Very informally, the paraphrases in (73c) and (74c) both seem to invoke some type of comparison of manner: the ‘way’ in which the subject is big is at issue, not the degree to which she is big. For now, however, I will simply note that AA-marked adjectives for which there exists a widely used CA-marked counterpart behave exceptionally. Possible reasons for the exceptional behavior of certain AA-marked adjectives is an interesting question for future research. I tentatively propose a competition analysis. Modification of CA-marked adjectives is more straightforward semantically (degree saturation rather than degree quantification) and requires the generation of less syntactic structure. However, given our system, the same degree constructions with AA/PA- and CA-marked adjectives hold in the same verifying scenarios. When a CA-marked adjective is not used, the Listener may ‘reason’ that the Speaker must have had reasons for not doing so: a meaning that is only available to AA/PA-marked adjectives must have been intended. Such a meaning might involve application of CLOSE and valuation of the D domain variable with the contextual standard, as we saw happen when AA/PA-marked adjectives were not modified by degree phrases. For elaboration of this type of competition theory, and examples of norm-relatedness arising in degree constructions, I refer the reader to Rett (2008) and Krasikova (2009). 6.0 Directions for Further Research 6.1. Further decomposing adjectives To this point, I have avoided decomposing the adjective below the level of ACA/AA/PA, but Navajo seems to motivate further morphological decomposition, with division of semantic labor between individual morphological elements. Two distinct decompositional approaches to adjectival meaning are represented by Bartsch and Vennemann (1972) and Kennedy (1999). I briefly summarize both and point to ways they could be applied to Navajo. Bartsch and Vennemann propose that at its most basic level, an adjective denotes a dimension in the lexicon: an adjective is only made into a function through further morphological machinery (overt or not). Examples of dimensions provided by Bartsch and Vennemann included Tallness, Breadth, Temperature, Speed, Intelligence, and Beauty.21 A measure function fM applies to the dimension. The measure function is a 2-place operator that takes a dimension and an individual and returns a degree. Additional morphology is necessary to introduce the degree argument, if one is motivated. Kennedy (1999) proposes that the basic meaning of an adjective is a measure function, which he defines as a function from individuals to degrees <ed>. Measure functions compose with degree morphology, e.g., MORE, AS, POS, etc. These morphemes introduce comparison with a standard of comparison (e.g., than-phrase) and convert the measure function into a property of individuals (i.e., they are of type <ed,et>). One covert degree morpheme proposed within Kennedy’s framework is particularly interesting given the data from Navajo. Svenonius and Kennedy (2006) propose a null degree head that MEAS takes as argument a measure function and returns a type <det> expression. 21 Bartsch and Vennemann (1972) do not identify the type of a ‘dimension.’ For concreteness, we might propose a dimension to be analogous to a trope (Moltmann 2009), or perhaps a set of degrees <dt>. 28 (75) [[MEAS]] = λged : g is a function from objects to measurable degrees.λdλx.g(x) > d The MEAS syntactically selects for a measure function. MEAS can only compose with adjectives associated with a salient measurement system (Svenonius and Kennedy 2006).22,23 (76) DegP <et> 4 NumP d Deg’ <det> six feet 4 Deg AP <ed> MEAS tall MEAS recalls Navajo CA: it introduces position in the syntax in which a semantic degree argument can be saturated. The proposals differ in several important ways, however. In particular, CA does not occupy the same position as other degree morphology in the language, e.g., -lááh ‘more.’ As a result, we find CA co-occurring (obligatorily, in fact) with degree phrases instead of, as MEAS does, blocking them from occurring with MEAS-marked adjectives. However, a decompositional analysis is clearly appealing for Navajo. As the analysis stands, I have proposed a semantic entry for the adjective once already marked with CA and AA/PA. Navajo adjectives, like eventive verbs, are clearly separable into a stem and additional morphemes. Neither Bartsch and Vennemann nor Kennedy provide examples of languages that express overtly the more basic level of adjectival meaning that their theories posit. Navajo adjectival stems seem like good candidates for overt expression of an adjective before it is a property of individuals. Likewise, CA, AA, and PA seem like they could be the types of morphology that a decompositional analysis would predict to be necessary to create a property of individuals-type meaning. A concrete decompositional analysis for Navajo could begin with the following denotations for CA, AA, and PA, using the framework of Bartsch and Vennemann (1972).24 (77) [[CA/AA/PA]] = λDIMλxλd.fM(DIM)(x) > d The remainder of the proposal (i.e., the differences in the syntactic structure projected by CA vs. AA/PA) would remain the same. 6.2. Crosslinguistic applications 22 As can be seen from (78), Kennedy assumes a different syntax than is assumed under the standard analysis or in the present analysis of Navajo. I refer the reader to Kennedy (1999) for explication of this system. 23 A more recent application of a MEAS-like morpheme is found in Grano and Kennedy (2011). The authors argue that MEAS (renamed µ) is affixed to the adjective and occurs in ‘transitive’ comparative constructions in Mandarin Chinese, where no standard marker introduces the standard of comparison. 24 An anonymous reviewer suggests that Bartsch and Vennemann’s analysis of adjectives as dimensions (not functions) may avoid problems that would arise if Navajo stems are analyzed as measure functions. In particular, if stems were analyzed as measure functions, we might predict (incorrectly) that stems can compose directly with an individual to return a degree. In fact, we never see stems – adjectival or verbal – directly composing with individuals: it is unclear, then, if the inability of adjectival stems to compose with individuals may not be due to their semantics but rather to a Navajo-wide condition that requires stems of all types to compose with additional adjectival or verbal morphology before composing with an individual. The syntactic structure posited in Hale (2003) reflects this idea. 29 The proposal that adjectives may always be semantically transitive but only syntactically transitive under special (here, morphological) circumstances not only challenges the standard analysis as summarized in §2, but also fits into much recent work on the semantics and syntax of adjectives crosslinguistically. Recent work has considered the degree to which the semantics and syntax of adjectives and degree phrases is universal, and possible variation crosslinguistically and within individual languages. Beck et al. (2009) addresses the issue of crosslinguistic variation, examining evidence that while a set of languages may all be analyzed as having syntactically and semantically transitive adjectives (i.e., be analyzed under the standard analysis), some languages differ in not permitting particular syntactic configurations involving the degree argument position in the syntax. Beck et al. examine languages in which the degree argument position of an adjective (specifier of AP) cannot be filled in the overt syntax. Beck et al. identify a number of constructions (subcomparative, measure phrase, degree question) which involve filling this syntactic position in the overt syntax by an operator. Languages may differ minimally from each other in whether this set of constructions is permitted: for instance, English and Russian are amenable to similar syntactic analyses, but Russian lacks these constructions while English has them. In other languages (e.g., Romanian, Gergel 2009), the degree argument position may only be filled in the overt syntax if the adjective bears particular functional morphology. It seems that it will be rewarding to compare Navajo with such languages: in both, a transitive semantics does not guarantee that the syntax will permit all configurations involving the degree argument. A second interesting direction that may be informed by Navajo is the treatment of a degree phrase as an argument vs. an adjunct to the adjective. The proposal for degree phrases modifying AA/PA-marked adjectives developed here was developed from Schwarzschild’s recent (2010, 2011) work on this issue. Interest in this issue can also be traced back to Schwarzschild (2005), where measure phrases are analyzed as adjuncts rather than arguments of the adjective. Navajo is seemingly unique in clearly marking degree phrases as adverbial when they modify syntactically intransitive adjectives. A third way in which the proposal for Navajo connects with strands of crosslinguistically inquiry is the fact that CA morphology – and, by extension, a transitive syntax – is almost entirely restricted to Navajo adjectives that denote dimensions associated with some salient system of measurement. Svenonius and Kennedy (2006) identified the existence of a ‘salient system of measurement’ as the factor determining the distribution of MEAS heads in Northern Norwegian. Even earlier, Bierwisch (1989) held “dimensional” adjectives (again, adjectives associated with salient systems of measurement) as distinct from other gradable adjectives. There is convergence between observations on variation in adjectival meaning and on the special status of ‘dimensional’ adjectives. Navajo promises to be a interesting and informative locus for study. 7.0 Conclusion I have proposed that all adjectives in Navajo are semantically transitive, but syntactic transitivity is only licensed for adjectives marked by CA (‘comparative aspect’) morphology. (78) a. AP 3 degree 3 b. AP 3 subject AAA/PA 30 subject ACA While composition between the degree phrase and the adjective occurs below the level of AP for CA-marked adjectives, composition between AA/PA-marked adjectives and a DegP is only possible at the clausal level after the degree argument of the adjective is been quantified over by the type-shifting operator CLOSE. DegP restricts the domain variable that CLOSE introduces. The proposal integrates a rich base of empirical observations from Navajo with current theories of adjectival meaning. The proposal questions the universality of the standard analysis, which takes adjectives to be both semantically and syntactically transitive. Navajo seems to be a language in which it is necessary to separate notions of syntactic and semantic transitivity of adjectives. The relationship between adjectival syntax and semantics – and the ways in which syntactically intransitive adjectives can still participate in degree constructions – is a potentially rewarding domain for future work. References Baker, Mark. 1996. The polysynthesis parameter. New York: Oxford University Press. Bartsch, Renate and Theo Vennemann. 1972. The grammar of relative adjectives and of comparison. Linguistische Berichte 20: 19-32. Beck, Sigrid, Sveta Krasikova, Daniel Fleischer, Remus Gergel, Stefan Hofstetter, Christiane Savelsberg, John Vanderelst, and Elisabeth Villalta. 2009. Crosslinguistic variation in comparison constructions. In Linguistic variation yearbook 9, ed. Jeroen van Craenenbroeck and Johan Rooryck, 1-66. Amsterdam: John Benjamins. Bhatt, Rajesh and Shoichi Takahashi. 2011. Reduced and unreduced phrasal comparatives. Natural Language and Linguistic Theory 29: 581-620. Bierwisch, Manfred. 1989. The semantics of gradation. In Dimensional adjectives, ed. Manfred Bierwisch and Ewald Lang, 71-262. Berlin: Springer-Verlag. Bochnak, M. Ryan. To appear. The non-universal status of degrees: Evidence from Washo. To appear in Proceedings of NELS 42, ed. Stefan Keine and Shayne Sloggett. Amherst: GLSA Publications. Bogal-Allbritten, Elizabeth. 2010. Distribution and function of comparative aspect in Athabaskan/Dene. In Proceedings of 2009 Dene/Athabaskan Languages Conference, ed. Justin Spencer and Siri Tuttle. Fairbanks, AK: Alaska Native Language Center. Bresnan, Joan. 1973. Syntax of the comparative clause in English. Linguistic Inquiry 4: 275-343. Cresswell, Max J. 1977. The semantics of degree. In Montague grammar, ed. Barbara Partee, 261-292. New York: Academic Press. Diesing, Molly. 1992. Indefinite. Cambridge, MA: MIT Press. Faltz, Leonard. 2000. A semantic basis for Navajo syntactic typology. In The Athabaskan languages: Perspectives on a Native American language family, ed. Theodore Fernald and Paul Platero, 28-50. Oxford: Oxford University Press. von Fintel, Kai. 1994. Restrictions on quantifier domains. Doctoral dissertation, University of Massachusetts Amherst. Gergel, Remus. 2008. Topicalization, doubling, and a little more on adjectives: What de does to degree dependencies in Romanian. Paper presented at the 30th Annual Convention of the German Society of Linguistics (DGfS), Universität Bamberg. Grano, Thomas and Christopher Kennedy. 2011. Mandarin transitive comparatives and the grammar of measurement. Journal of East Asian Linguistics. 31 Hale, Kenneth. 2003. On the significance of Eloise Jelinek’s pronominal argument hypothesis. In Formal approaches to function in grammar: In Honor of Eloise Jelinek, ed. Andrew Carnie, Heidi Harley, and MaryAnn Willie. Linguistik Aktuell 62. Amsterdam: John Benjamins. Heim, Irene. 1982. The semantics of definite and indefinite noun phrases. Doctoral dissertation, University of Massachusetts Amherst. Heim, Irene. 2001. Degree operators and scope. In Audiatur Vox Sapientiae: A festschrift for Arnim von Stechow, ed. Caroline Féry and Wolfgang Sternefeld, 214-239. Berlin: Akademie-Verlag. Jelinek, Eloise. 1984. Empty categories, case, and configurationality. Natural Language and Linguistic Theory, 2: 39-76. Kamp, Hans. 1981. A theory of truth and semantic representation. In Formal Methods in Study of Language, Part 1, ed. Jeroen Groenendijk, Theo Janssen and Marin Stokhoff, 277-322. Amsterdam: Mathematisch Centrum. Kennedy, Christopher. 1999. Projecting the adjective: The syntax and semantics of gradability and comparison. New York: Garland. 1997 UCSC doctoral dissertation. Kennedy, Christopher. 2007. Modes of comparison. In Proceedings of CLS 43, ed. Malcolm Elliot, James Kirby, Osamu Sawada, Eleni Staraki, and Suwon Yoon, 141-165. Chicago: The Chicago Linguistic Society. Kennedy, Christopher and Louise McNally. 2005. Scale structure and the semantic typology of gradable predicates. Language 81: 345-381. Klein, Ewan. 1980. A semantics for positive and comparative adjectives. Linguistics and Philosophy 4: 1-45. Moltmann, Friederike. 2009. Degree structure as trope structure: a trope-based analysis of positive and comparative adjectives. Linguistics and Philosophy 32: 51-94. Rett, Jessica. 2008. Degree modification in natural language. Doctoral dissertation, Rutgers University. van Rooij, Robert. 2011. Implicit versus explicit comparatives. In Vagueness and Language Use, ed. Paul Egré and Nathan Klinedinst. Palgrave MacMillan. Schauber, Ellen. 1979. The syntax and semantics of questions in Navajo. New York: Garland. Schwarzschild, Roger. 2005. Measure phrases as modifiers of adjectives. Recherches Linguistiques de Vincennes 34: 207-228. Schwarzschild, Roger. 2008. The semantics of comparatives and other degree constructions. Language and Linguistics Compass 2: 308-331. Schwarzschild, Roger. 2010. Dimensional adjectives and quantifier-domain adverbials, handout for talk at Workshop on Comparison Constructions Crosslinguistically, Universitët Tübingen, May 2010. Schwarzschild, Roger. 2011. Quantifier domain adverbials: Semantic change and the comparative. Handout for talk at University of Massachusetts Amherst, October 2011. Seuren, Pieter A.M. 1973. The comparative. In Generative grammar in Europe, ed. Ferenc Kiefer and Nicolas Ruwet. Dordrecht: Riedel. Svenonius, Peter and Christopher Kennedy. 2006. Northern Norwegian degree questions and the syntax of measurement. In Phases of Interpretation, Vol. 9 in Studies in Generative Grammar, ed. Maria Frascarelli. Berlin: Mouton de Gruyter. Stassen, Leon. 1985. Comparison and universal grammar: An essay in universal grammar. Oxford: Blackwell. 32 von Stechow, Arnim. 1984. Comparing semantic theories of comparison. Journal of Semantics 3: 1-77. Willie MaryAnne and Eloise Jelinek. 2000. Navajo as a discourse configurational language. In The Athabaskan languages: Perspectives on a Native American language family, ed. Theodore Fernald and Paul Platero, 252-287. Oxford: Oxford University Press. Young, Robert W. and William Morgan. 1987. The Navajo Language: A Grammar and Colloquial Dictionary. Albuquerque, NM: University of New Mexico Press. 33
© Copyright 2026 Paperzz