Mirative expressions, which mark surprising information (DeLancey 1997), are often expressed through linguistic forms that are also used to encode other, seemingly unrelated, meanings – e.g. evidential markers that mark lack of direct evidence (Turkish: Slobin & Aksu 1982, Peterson 2010; Cheyenne: Rett & Murray 2013; Cuzco Quechua: Faller 2002; Ostyak: Nikolaeva 1999; among others). In this paper, we show that the English particle like features a parallel polysemy between a mirative use and its better-known hedging use, which expresses weakened commitment to the strict denotation of a linguistic expression. After presenting several diagnostics that point to a genuine empirical difference between the hedging and mirative functions of like, we propose that both uses widen the size of a contextually restricted set, admitting elements that were previously excluded. More specifically, hedging like expands the set of ‘similar enough’ interpretations that we can apply to a linguistic expression in a context, including interpretations that we would normally consider to be too different from the context at hand. Mirative like, on the other hand, expands the set of worlds that we are willing to consider as candidates for the actual world in the conversation, resulting in the inclusion of worlds that interlocutors have previously ruled out due to perceived outlandishness. We therefore suggest that the two uses are best treated as sharing a common semantic kernel, deriving hedging and mirativity as effects of the particular type of object to which like applies.

The syntax literature has overwhelmingly adopted the view that Condition C reconstruction takes place in wh-chains for R-expressions contained within arguments, but not within adjuncts of fronted wh-phrases. At the same time, this empirical picture has been questioned by various authors. We undertake a series of grammaticality surveys using Amazon Mechanical Turk in an attempt to clarify the empirical picture regarding reconstruction for Binding Condition C. We find absolutely no evidence of an argument–adjunct distinction in reconstruction for Binding Condition C. Neither arguments nor adjuncts reconstruct for Condition C. We suggest that those speakers who report such a contrast (linguists, primarily) are following a pragmatic bias, and not Condition C. While we do not find reconstruction of dependents of fronted NPs for Binding Condition C, we do find reconstruction of fronted PPs. That is, the NP complement of a fronted P must reconstruct for Binding Condition C. The literature also finds reconstruction of NP complements of verbs and adjectives. This means that fronted Ns are special in not requiring reconstruction of their arguments and adjuncts. We suggest that, syntactically, arguments of Ns are treated as adjuncts: semantic arguments simply adjoin in the same manner as true adjuncts. Syntactic adjuncts can be left out of lower copies in chains, something that we suggest follows from a left-to-right syntactic derivation plus an economy condition on copying.

This paper addresses one of the most contested issues in phonology: unnatural alternations. First, non-natural phonological processes are subdivided into unmotivated and unnatural. The central topic of the paper is an unnatural process: post-nasal devoicing (PND). I collect thirteen cases of PND and argue that in all reported cases, PND does not derive from a single unnatural sound change (as claimed in some individual accounts of the data), but rather from a combination of three sound changes, each of which is phonetically motivated. I present new evidence showing that the three stages are directly historically attested in the pre-history of Yaghnobi. Based on several discussed cases, I propose a new diachronic model for explaining unnatural phenomena called the Blurring Process and point to its advantages over competing approaches (hypercorrection, perceptual enhancement, and phonetic motivation). The Blurring Process establishes general diachronic conditions for unnatural synchronic processes and can be employed to explain unnatural processes beyond PND. Additionally, I provide a proof establishing the minimal sound changes required for an unmotivated/unnatural alternation to arise. The Blurring Process and Minimal Sound Change Requirement have implications for models of typology within the Channel Bias approach. This paper thus presents a first step toward the ultimate goal of quantifying the influences of Channel Bias on phonological typology.

This work provides evidence that Subject Island violation effects vanish if subject-embedded gaps are made as frequent and pragmatically felicitous as non-island counterpart controls. We argue that Subject Island effects are caused by the fact that subject-embedded gaps are pragmatically unusual – as the informational focus does not usually correspond to a dependant of the subject phrase – and therefore are highly contrary to comprehenders’ expectations about the distribution of filler–gap dependencies (Chaves 2013, Hofmeister, Casasanto & Sag 2013). This not only explains why sentences with subject-embedded gaps often become more acceptable ‘parasitically’, in the presence of a second gap outside the island, but also explains why some Subject Island violations fail to exhibit any amelioration with repetition (Sprouse 2009, Crawford 2011, Goodall 2011); some ameliorate marginally (Snyder 2000, 2017) or moderately (Hiramatsu 2000, Clausen 2011, Chaves & Dery 2014), and others become fully acceptable, as in our case. This conclusion extends to self-paced reading Subject Island studies (Stowe 1986, Kurtzman & Crawford 1991, Pickering, Barton & Shillcock 1994, Phillips 2006), which sometimes find evidence of gap filling and sometimes do not.

How do people answer polar questions? In this fourteen-language study of answers to questions in conversation, we compare the two main strategies; first, interjection-type answers such as uh-huh (or equivalents yes, mm, head nods, etc.), and second, repetition-type answers that repeat some or all of the question. We find that all languages offer both options, but that there is a strong asymmetry in their frequency of use, with a global preference for interjection-type answers. We propose that this preference is motivated by the fact that the two options are not equivalent in meaning. We argue that interjection-type answers are intrinsically suited to be the pragmatically unmarked, and thus more frequent, strategy for confirming polar questions, regardless of the language spoken. Our analysis is based on the semantic-pragmatic profile of the interjection-type and repetition-type answer strategies, in the context of certain asymmetries inherent to the dialogic speech act structure of question–answer sequences, including sequential agency and thematic agency. This allows us to see possible explanations for the outlier distributions found in ǂĀkhoe Haiǁom and Tzeltal.

This paper addresses the meaning and use of clefted wh-interrogatives (I-clefts) in Swedish. It is shown that I-clefts always relate immediately to the topic under discussion and serve to clarify a matter in relation to this topic. They are never used in out-of-the-blue contexts. I argue that I-clefts have the same information structure as typically assumed for declarative clefts: the clefted clause expresses an existential presupposition and the cleft phrase is the identificational focus of the utterance. I further argue that the implication of existence commonly associated with canonical argument questions is weaker (a conversational implicature) than the existential presupposition associated with clefts. The results from an extensive corpus survey show that argument I-clefts (who, what) constitute no less than 98% of the total number of I-clefts in my material. This frequency is linked to the presuppositional status of the cleft construction: in contexts where the denoted event is presupposed as part of the common ground, the clefted variety is the more effective choice, due to its clear partitioning of focus and ground. The ‘cost’ of using a more complex syntactic structure (the cleft) is thus counterbalanced by the benefit of being able to pose a question adjusted to the contextual requirements. As non-argument questions are typically presuppositional irrespective of syntactic form, the gain of using a cleft is less obvious – hence their infrequency in the material.

Passivization played a central role in shaping both linguistic theory and psycholinguistic approaches to sentence processing, language acquisition and impairment. We present the results of two experiments that simultaneously test online processing (self-paced reading) and offline comprehension (through comprehension questions) of passives in German while also manipulating the event structure of the predicates used. In contrast to English, German passives are unambiguously verbal, allowing for the study of passivization independent of a confound in the degree of interpretive ambiguity (verbal/adjectival). In English, this ambiguity interacts with event structure, with passives of stative predicates naturally receiving an adjectival interpretation. In a recent study, Paolazzi et al. (2015, 2016) showed that in contrast to the mainstream theoretical perspective, passive sentences are not inherently harder to process than actives. Complexity of passivization in English is tied to the aspectual class of the verbal predicate passivized: with eventive predicates, passives are read faster (as hinted at in previous literature) and generate no comprehension difficulties (in contrast to previous findings with mixed predicates). Complexity effects with passivization, in turn, are only found with stative predicates. The asymmetry is claimed to stem from the temporary adjectival/verbal ambiguity of stative passives in English. We predict that the observed difficulty with English stative passives disappears in German, given that in this language the passive construction under investigation is unambiguously verbal. The results support this prediction: both offline and online there was no difficulty with passivization, under either eventive or stative predicates. In fact, passives and their rich morphology eased parsing across both types of predicates.

This paper offers an in-depth look at roots and verb stem morphology in Chuj (Mayan) in order to address a larger question: when it comes to the formation of verb stems, what information is contributed by the root, and what is contributed by the functional heads? I show first that roots in Chuj are not acategorical in the strict sense (cf. Borer 2005), but must be grouped into classes based on their stem-forming possibilities. Root class does not map directly to surface lexical category, but does determine which functional heads (i.e. valence morphology) may merge with the root. Second, I show that while the introduction of the external argument, along with clausal licensing and agreement generally, are all governed by higher functional heads, the presence or absence of an internal argument is dictated by the root. Specifically, I show that transitive roots in Chuj always combine with an internal argument, whether it be (i) a full DP, (ii) a bare pseudo-incorporated NP, or (iii) an implicit object in an antipassive. In the spirit of work such as Levinson (2007, 2014), I connect this to the semantic type of the root; root class reflects semantic type, and semantic type affects the root’s combinatorial properties. This work also contributes to the discussion of how valence morphology operates. In line with works such as Alexiadou, Anagnostopoulou & Schäfer (2006), I argue that valence morphology applies directly to roots, rather than to some ‘inherent valence’ of a verb.

This paper offers a comprehensive discussion of the cardinal numeral system of Italian Sign Language. At the lexical level, we present the different formational strategies used to generate cardinal numerals and we provide evidence that in the younger generations of signers, the sign one has lost the function of indefinite determiner and is now used as a cardinal only. At the syntactic level, we show that the attested variation in the ordering between the cardinal and the noun is in part due to definiteness and contrastive focus. We account for this variation within the cartographic approach to syntax. Finally, we offer a principled explanation for the reason why cardinals inside Measure Phrases are not subject to word order variation, but always precede the measure noun.

This study aims to show the dynamic aspect of word-formation paradigms in autonomous morphology by examining the compound marker in Turkish Noun–Noun compounds, as in buz paten-i ‘ice-skate (ice skate-cm)’, and its relation to derivational suffixes. The study proposes a process-based morphological paradigm structure which involves compounding and derivational operations. In this system, the compound marker has a formal paradigmatic function: it creates correct lexeme forms based on bare Noun–Noun compounds, which would otherwise serve as input to certain derivational operations. The current system thus accounts for both permitted and unpermitted suffix combinations involving compounding and the optionality in certain combinations, such as buz paten-ci (-si) ‘a/the ice skater (ice skate-agt-cm)’, where the compound marker may (not) appear in combination with the (derivational) agentive -CI. The study also presents a survey which implies that a group of derivational affixes is in a paradigmatic relation with the compound marker, and all of these affixations constitute alternative paths in a dynamic paradigm structure. The findings of the study are considered to contribute to the understanding of the nature of the autonomous morphological operations and paradigms, which cannot be restricted to the lexicon or manipulated by syntax.

This research investigates the morpho-syntactic behaviour of the Arabic complementizer ʔinn in a range of Arabic varieties (Modern Standard Arabic, Jordanian Arabic, and Lebanese Arabic). It essentially argues that this complementizer shares (not donates or keeps, pace Ouali 2008, 2011) its unvalued $\unicode[STIX]{x1D719}$-features with its complement $\text{T}^{0}$, something that makes ʔinn and $\text{T}^{0}$ separate agreeing heads. An inflectional suffix attached to ʔinn is treated as a PF reflex (i.e. an overt morphological realization) of valuation of ʔinn’s unvalued $\unicode[STIX]{x1D719}$-features or lack thereof. This research also argues that the occurrence of such an inflectional suffix is ruled by the postulated Agree Chain Record, an interface condition that demands an Agree relation to have a PF reflex, called a Record (i.e. an overt Case marking on the goal or, if not, a $\unicode[STIX]{x1D719}$-affix on the probe). This way, we account for the complementary distribution of overt Case and $\unicode[STIX]{x1D719}$-Agree in Arabic. We also show how a host of other phenomena, including word order agreement asymmetries in Modern Standard Arabic and lack of such asymmetries in Arabic vernaculars, fares well with this view.

In this paper, we present an analysis of so-called determinerless PPs in German, i.e. prepositional phrases that allow singular count nouns to occur without an accompanying determiner, despite other rules in the grammar requiring the presence of the determiner. The analysis is based on annotated corpus data, which are fed into a statistical classifier (applying logistic regression). Superficially, the syntax of bare prepositional phrases is difficult to capture, and intuitions cannot be easily elicited. The analysis is based on data sets for two pairs of German prepositions: mit ‘with’ and ohne ‘without’, and über ‘over, above’ and unter ‘under, below’. The results of the classifiers applied to annotated data indicate which syntactic, morphological and semantic features are responsible for determiner omission. We are able to detect common properties of all four prepositions, as well as preposition-specific, and idiosyncratic properties. The apparently unsystematic conditions for determiner omission can be discerned by tracing the interaction of these properties.

This paper brings novel data to bear on whether nominal concord relationships are formed in the narrow syntax or post-syntactically. In Guébie, a Kru language spoken in Côte d’Ivoire, nominal concord marking on non-human pronouns and adjectives is determined not by syntactic or semantic features of the concord-triggering noun, but by the phonological form of the noun. Specifically, concord marking on pronouns and adjectives surfaces as a vowel with the same backness features as the vowels of the head noun. Assuming that syntax is phonology-free (Pullum & Zwicky 1986, 1988), the fact that we see phonological features conditioning nominal concord in Guébie means that nominal concord must take place in the post-syntax. I expand on post-syntactic models of nominal concord in Distributed Morphology (Kramer 2010, Norris 2014, Baier 2015) showing that when combined with a constraint-based phonology, such an approach can account for both phonologically and syntactico-semantically determined concord systems. Additionally, the proposed analysis includes a formal account of ellipsis via constraints during the phonological component.

In Tamil, coronals are licensed in onsets and initial syllables, exemplifying what Jesney (2011b) calls Licensing in Multiple Contexts (LMC). Jesney shows that while only positional faithfulness produces LMC in Optimality Theory, positional licensing provides a competing analysis of LMC in Harmonic Grammar (HG). This suggests that positional faithfulness may not be necessary in HG. We argue, though, that positional faithfulness remains essential. First, other facts in Tamil are incompatible with the positional licensing approach to LMC, rendering the positional faithfulness alternative the only viable analysis. Second, only with positional faithfulness can certain typological generalizations concerning assimilation between consonants be captured.

This article presents an analysis of two constructions in the Eastern Algonquian language Passamaquoddy in which the position of the object of a verb of cognition (‘know’, ‘believe’, ‘remember’, ‘wonder about’, ‘suspect’) is linked, either by apparent raising or by apparent long-distance agreement, to a position within a clausal complement to the verb. The latter position may be arbitrarily deeply embedded. The analysis developed here, formulated in the framework of Head-driven Phrase Structure Grammar, demonstrates that the two constructions in fact represent alternative realizations of identical argument structures for the verbs in question and that the apparent long-distance dependencies involved can be accounted for in terms of a purely local principle of argument selection.

It is known that consonants can act as boundary markers when they are located at the left edge of a prosodic domain, helping listeners to parse incoming speech. To achieve maximum efficiency in marking out boundaries, those markers should be acoustically salient. In Element Theory, domain markers are represented by the elements |H| and |ʔ|. Being inherently voiceless, these elements stand apart from the other elements, which are spontaneously voiced. Most languages show a preference for incorporating |H| or |ʔ| into segmental structures which stand at the left edge of domains. This paper challenges the universality of this view by analysing data from Kaqchikel, a K’iche’an language with a bias for marking the right edge of domains rather than the left. The marker in question is intense/prolonged noise which, in Element Theory is represented by |H|. The |H| element is present in fortis fricatives and aspirates, and in Kaqchikel it regularly appears in segments at the right edge of prosodic domains where it serves as a domain boundary marker. Boundary marking in Kaqchikel is analysed here using a Precedence-free Phonology approach to melodic structure (Nasukawa 2016) in which the linear ordering of segments is determined by the hierarchical organization of melodic units (elements) within a unified melodic–prosodic structure.

This paper investigates the cross-linguistic distribution of argument ellipsis (AE) with an emphasis on Chinese, an Asian language well-known for its lack of overt morphological agreement. It is observed in the literature that Japanese permits AE in both null subject and null object positions whereas Chinese permits it in null object positions, but not in null subject positions. Adopting Saito’s (2007) hypothesis that the presence of $\unicode[STIX]{x03C6}$-feature agreement associated with v or T blocks AE, Miyagawa (2013) and Takahashi (2014) argue that the absence of subject AE in Chinese follows from abstract subject agreement. After presenting three empirical arguments against this analysis from the Chinese literature, I propose that the distribution of AE is better predicted by topichood and link this proposal to Saito’s (2017) recent analysis of AE developed for Japanese, whereby AE, analyzed as LF Copy, cannot apply to an operator–variable configuration. My analysis is supported by the novel observation that the null subject position in Chinese actually allows AE when it is not linked to the topic position, as in hanging topics, relative clauses and conditional clauses.

In this paper, we consider two kinds of vP-fronting constructions in English and argue that they receive quite different analyses. First, we show that English vP-preposing does not have the properties that would be expected of a movement-derived dependency. Evidence for this conclusion is adduced from the licensing conditions on its occurrence, from the availability of morphological mismatches, and from reconstruction facts. By contrast, we show that English participle preposing is a well-behaved case of vP-movement, contrasting with vP-preposing with respect to reconstruction properties in particular. We propose that the differences between the two constructions follow from the interaction of two constraints: the excluded middle constraint (EMC), which rules out derivations involving spellout of linearly intermediate copies only, and the N-only constraint, which restricts movement to occurring where the trace position would license a nominal. The EMC rules out deriving vP-fronting by true movement and instead necessitates a base-generation analysis, while the N-only constraint ensures that participle preposing is only possible in limited circumstances.

In this squib, I provide evidence of a novel type in favor of the existence of the DP in previously undiscussed articleless languages: I show that a comitative preposition in the Ossetic languages cannot attach to nominal expressions that would be analyzed as DPs in better studied languages. On the other hand, nominal expressions that it can attach to are of the kinds that would be analyzed as bare NumPs.

In this paper we provide a novel and unified formal analysis of the stress microvariation found in verb–enclitic groupings in Barcelona Catalan, with stress stability, in Formentera Catalan, with stress shift to the penultimate and the last syllable of the whole sequence, and in Mallorca and Menorca Catalan, with stress shift to the last syllable. We argue that stress shift in Formentera Catalan conforms directly to the unmarked pattern of nominal stress in Catalan, that is, a right-aligned moraic trochee, with the constraint All-Feet-Right undominated, leading to final and penultimate stress. In Barcelona Catalan, the effects of this constraint are inhibited by an anti-alignment constraint prohibiting the right edge of a foot from coinciding with the right edge of a clitic. In Mallorca and Menorca Catalan, stress shift is also understood as a strategy to meet the unmarked nominal stress pattern of Catalan, although in these dialects the anti-alignment constraint is also undominated, leading to catalexis, in this case a melodically empty mora, and thus to final stress.