The Grammar of PP-like Free Relatives: Evidence from Subordinate wo-Clauses in German

This paper addresses the grammar of PP-like free relatives (= PP-FRs) such as the locative clause in Bello schläft, wo Grace schläft ‘Bello is sleeping where Grace is sleeping’. The most prominent compositional analysis of PP-FRs is developed by Caponigro (2004), Caponigro and Pearl (2009), and Hall and Caponigro (2010). They derive PP-FRs from a stacked CP-structure that is based on silent prepositions and a nominal treatment of wh-words such as where. I first take issue with this analysis by means of a detailed discussion of subordinate clauses introduced by wo ‘where’ in German. Specifically, I show that wo-clauses do not involve silent prepositions, and that both wo and wo-clauses have a prepositional instead of a nominal nature. Second, I sketch a surface-oriented alternative approach to PP-FRs that dispenses with silent prepositions and assigns the prepositional nature of PP-like wh-words a key role in the derivation. Specifically, I propose that the fronting of PP-like wh-words in free relatives licenses an abstraction over predicates (instead of entities) and a corresponding type shifting. The new analysis is related to analyses of corresponding subordinate headed relative and interrogative clauses. Furthermore, it is confronted with potential general constraints on higher-order abstraction as argued for, in particular, by Poole (2017).

The puzzling nuanced status of who free relative clauses in English: a follow-up to Patterson and Caponigro (2015)

This squib challenges Patterson & Caponigro's (2015, this journal) claim that there are few acceptable free relative clauses with who. We show that free relatives with who are generally acceptable when they are ‘transparent’ free relatives or complements of a copula, and add further nuance to their findings concerning how the degree of acceptability of free relatives with who varies according to positional factors.

Headless Relative Clauses in Ch'ol

This chapter surveys headless relative clauses in Ch’ol, a Mayan language spoken in the state of Chiapas in southern Mexico. Ch’ol is rare among Mayan languages in possessing a special morpheme found with relativized nouns, the second position clitic = bä. While this morpheme is required for relativized argument nouns, it is not present in free relatives, which suggests a different derivation for this class of construction. Maximal (definite) and existential (indefinite) free relatives are described. They both make use of a fronted wh-expression and lack the morpheme = bä. Maximal and existential free relatives in Ch’ol appear identical to one another in structure. Following existing studies on other languages, it is argued that the different interpretations of these clauses are a result of the environments in which they appear. Finally, Ch’ol has two different types of constructions in which a determiner element is followed by a headless relative: one corresponding to the = bä structure and one corresponding to the free relative structure. The former is proposed to be a regular headed relative clause with an unpronounced head, as has elsewhere been argued for Yucatec. The latter, on the other hand, corresponds to a free relative structure with an added determiner element.

Headless Relative Clauses in Pesh

The aim of this study is to describe the two main kinds of headless relative clauses that are attested in Pesh, a Chibchan language spoken in Honduras: free relative clauses, which use a wh-word that functions as a relative pronoun at their left edge and a subordinator at their right edge, and headless relative clauses, which lack a wh- word but show a case marker or the topic marker at the right edge of the clause. The first type is less frequently attested in the natural corpus this study relies on, although the corpus does contain various instances of maximal, existential, and free-choice free relative clauses. Each of the constructions is distinguished by features of the wh-word and/or by certain restrictions regarding the tense of the verb in headless relative clauses or the type of verb in matrix clauses. The second type of headless relative clause, the ones that do not use a wh-expression, are much more frequent in the corpus and behave like headed relative clauses that lack a wh-expression. They are like noun phrases marked by a phrase-final case marker or the topic maker. The case or topic markers are used for light-headed relative clauses and for almost all types of maximal headless relative clause that have neither a light head nor a wh-expression, in contrast to maximal free relatives, in which only locative wh-words occur.

Headless Relative Clauses in Tseltalan

This chapter investigates headless relative clauses in Tseltal and Tsotsil, languages which make up the Tseltalan branch of Western Mayan. Headless relatives introduced by wh- interrogative expressions (free relatives) are associated with two interpretations: maximal and existential. There is no distinct free-choice free relative construction, but free-choice interpretations arise as possible readings of maximal free relatives. There are other headless relative clause constructions in Tseltalan which involve an overt determiner combined with the wh-pronoun or which lack an overt wh-pronoun. The authors argue that some of these are derived from headed relative clauses, with discourse-conditioned elision of the head noun, while others are based on free relatives in which the wh-pronoun is augmented by a determiner.

Whatever His Arguments, Whatever Relatives Are Not Free Relatives: A Reply to Caponigro’s Reply

This is a reply to Caponigro 2019, which argues that the phrase structure theory proposed by Donati and Cecchetto (2011; 2015) falls short of accounting for the attested patterns of free relative clauses. Caponigro questions the reliability of the data supporting D&C’s hypothesis that ever-relatives introduced by a phrase ( ever+NP relatives) should not be assimilated to free relatives. This paper reports the results of 4 controlled experiments in English and Italian and discusses five properties that set apart free relatives from full relatives (occurrence with a complementizer, occurrence with a relative pronoun, infinitival use, absolute use, adverbial use). Crucially, ever+NP relatives do not pattern like free relatives in any of these five domains, either in Italian or in English. This clearly shows that ever-relatives are not a counterexample against D&C’s phrase structure theory. Another potential counterexample, Romanian free relatives, is also discussed. As for the analysis of ever+NP relatives, in Italian they are shown to be garden variety headed relatives, while in English they are headed relatives that involve a D to D movement which is responsible for the syntactic formation of the complex determiner what+ ever.

Case and preposition stranding in Old English free relatives

Taylor (2014) observes that some of the factual claims made in Allen (1980), the most thorough examination of free relatives in Old English to date, are not entirely correct. Taylor presents some examples that Allen’s analysis of Old English free relatives does not account for and proposes an alternative analysis in which the relative pronoun can be internal to the relative clause and the case of the pronoun is determined by the case hierarchy proposed by Harbert (2007) for Gothic. This corpus-based study supplies new data showing that while Taylor’s relative-internal analysis is needed for some examples, the evidence does not support the suggested case hierarchy except in regulating optional case attraction. Latin influence may account for examples that do not fit the usual patterns.