Review of Relative Clauses in Languages of the Americas
This edited volume stems from a series of seminars on linguistic complexity at the Department of Linguistics, University of Sonora (Hermosillo, Mexico), with a focus on the indigenous languages of the Americas. The volume assembles 13 contributions on the topic of relative clauses (RCs) by linguists associated with the seminars. Besides the 13 main chapters, the book also has an introduction by the editors, a map of the languages cited, and language, author, and subject indices. The main body of the book is organized into three parts: a historical/typological/theoretical part with 3 papers; a second part devoted to Uto-Aztecan languages with 4 papers; and a final part on languages elsewhere in the Americas with six papers.
In the first contribution, ''Toward a diachronic typology of relative clause” (pp. 3-25), Talmy Givón proposes that there are two mega-pathways for the creation of RCs: the clause-chaining pathway and the nominalized “V + Complement” pathway. The first pathway, in fact, encompasses two sub-paths, of which only one, I think (see the evaluation section), may be a genuine source of RCs. In this pathway, at the initial stage, a clause stands in a paratactic relation with a non-restrictive RC. The latter becomes a typical embedded restrictive RC once the two clauses merge their intonation contour (plus any language specific adaptations; cf. Geman: Ich kenne die Frau, DIE hat dem Mann das Buch gegeben ‘I know the woman, the one gave the book to the man’ > Ich kenne die Frau die dem Mann das Buch gegeben hat ‘I know the woman who gave the book to the man’). The author also cites a related example in Hebrew, where the paratactic structure literally meaning “the man, the one I met yesterday...” merges its intonation contours to become the equivalent of “the man I met yesterday...”. The second major pathway to RCs involves a nominalized clause standing in a paratactic relation with another clause. At the restrictive RC stage, the two clauses merge their intonational contours (e.g. paraphrasing Ute, “the woman, [that] putter of rock on the table”... > “the woman putter of rock on the table”, i.e., ‘the woman that put a rock on the table’). The author also claims that the same paratactic to syntactic change explains the emergence of cleft and wh-question constructions in some languages.
The second contribution, ''The evolution of language and elaborateness of grammar: The case of relative clauses in creole languages” (pp. 27-46), by Tania Kuteva and Bernard Comrie, argues that language structure is simple before it evolves to become more complex. To ground their claim, the authors focus on the sharp contrast between non-creole languages and creole languages in their way of marking RCs. Non-creole languages can simultaneously mark their RC with zero, one, two, three, or even five markers, as in the Ngemba language (Bantoid, Cameroon), and can also increase the number of relative markers following language contact. By contrast, creole languages in the authors’ sample (from all the over the world) use just one marker on their relative clause (no more, no less), regardless of the RC properties of the languages that gave rise to the creole language. The authors propose that creole languages simply did not have enough time to develop complex grammars beyond the basic one form/ one meaning correspondence.
In the third contribution, ''Some issues in the linking between syntax and semantics in relative clauses'' (pp. 47-64), Robert D. Van Valin Jr. proposes a structural representation of the two main types of RCs (external and internal-head RCs), as well as the rules for their interpretation. The author reminds readers that internally-headed RCs (IHRCs) were not paid due service in the Generativist framework, where they were analyzed as externally-headed RCs (EHRCs), with covert movement and a null head. Using the Role and Reference Grammar model of clause structure (the Layered Structure of the Clause), the author proposes two different representations for the two types of RCs. The EHRC appears at a Noun Phrase’s (NP’s) periphery, modifying the noun. By contrast, the IHRC, so to speak, wraps around the head noun and the whole clause is an argument of the main clause. Nonetheless, the EHRCs and the IHRCs present a symmetrical challenge to the speaker/hearer: s/he must recover the function of the external head noun in the EHRC, or recover the function fulfilled in the main clause by the internal head noun of the IHRC. This enables the author to posit very similar rules of interpretation linking the syntax and semantics of the two types of RCs.
The fourth contribution, which is the first dealing with language-specific issues, is titled ''Relative clauses and nominalizations in Yaqui'' (pp. 67-95), by Albert Álvarez González. The paper describes in detail the five main relative strategies used in Yaqui (Uto-Aztecan, Mexico, USA), which depend on the syntactic function of the relativized noun: subject, direct object, indirect object, oblique, and locative. All five strategies use nominalization to turn a clause into a NP which, according to the author, stands in apposition with the head noun (p. 89). The apposition is then interpreted as a modification in the typical sense of a RC. The three nominalizing affixes used in the five types of RCs also function as clause nominalizers elsewhere in the language. Despite their formal complexity, RCs in Yaqui are shown to respect the NP Accessibility Hierarchy in relativization (Keenan and Comrie 1977). Nonetheless, the author, in his final discussion, claims that Yaqui has no true relative clauses, thus reducing relativization to nominalization.
In contrast to the Álvarez Gonzalez’s contribution, the fifth paper ''On relative clause and related constructions in Yaqui'' (pp. 97-126), by Lilián Guerrero, claims that this language truly has RCs that are distinct from non-relative nominalized subordinate clauses. The author further complements the data by showing that Yaqui can have internally-headed RCs, although these are not very frequent. A major contribution of the paper is its study of the discourse functions of RCs. For example, intransitive subject RCs serve to introduce new protagonists in a story, which helps explain the high frequency of this type of RC, its high degree of nominalization, and its high degree of attachment to head nouns (as compared to RCs relativizing other syntactic roles, which are less frequent, less nominalized, and more detached from their heads). The author also meticulously describes the differences between true RCs and related non-relative subordinate clauses.
In the sixth contribution, ''From demonstrative to relative marker to clause linker: Relative clause formation in Pima Bajo'' (pp. 127-146), Zarina Estrada-Fernández claims that in this language, only the subject and the object can be relativized in post-nominal clauses whose verb is suffixed with an element -kig (which also functions as a demonstrative in the language), as seen on p. 134 in the following examples: hig a’an [gii-kig] vig ‘det. feather [fall-rel.] red’, i.e., ‘the feather that fell is red’; ig okosi [in-niir-kig] ig gi’id ‘det. woman [I-see-rel.] det. big’, i.e., ‘the woman I saw is big’. The author claims that NPs in other syntactic functions cannot be relativized. They certainly appear next to clauses that also contain the element -kig, but in these cases, the element is attached to the postposition rather than to the verb (see the following example containing a relativized oblique nominal: aap timitim maa ik okosi-vui-ta-kig in-tikpan ‘you tortillas give det. woman-to-TA-link. I-work’, which the author directly translates as ‘you gave tortillas to the woman I worked for’, p. 142). In these clauses, the element -kig is not taken to be a relativizer, but rather a clause subordinator/linker. The author justifies her straight translation of these examples with English RCs by saying that the constructions, although not RCs, are functionally equivalent to RCs. It should be noted, however, that subjects, too, can appear in non-relative constructions where the element -kig is attached to a relative pronoun (so labeled by the author) that is based on an interrogative pronoun, as in the following example on p. 143: ig kil mua gogos aita-kig in-kiik ‘det. man kill dog rel.pro.-link. I-bite’, translated as ‘the man killed the dog that bit me’, where -kig is taken to be a non-relative clause linker. The author fails to give examples concerning direct objects in the non-relative -but translated as relative- constructions. The author concludes that Pima Bajo RCs conform to the NP Accessibility Hierarchy of Keenan and Comrie (1977).
In the seventh contribution, ''Functional underpinnings of diachrony in relative clause formation: The nominalization-relativization connection in Northern Paiute'' (pp. 147-170), Tim Thornes shows that relativization and nominalization are so intimately connected, and due to this, he wonders whether the former should be reduced to the latter. Nonetheless, despite the author’s insistence on their similarities, throughout the paper he maintains a terminological distinction between the two phenomena. Northern Paiute can relativize subjects (with the nominalizer -dia), objects, and obliques (with the nominalizer -na). Besides nominalization, RCs display other nominal features (such as a possessive subject inside them). These features distinguish RCs from complement and adverbial clauses, which take the nominalizer –na, but have no other nominal feature. The author suggests that this may confirm the claim by some that headless RCs may give rise to other types of subordinate clauses through “refinitization” (see Epps 2007 and this volume).
In the eighth contribution, ''Clauses as noun modifiers in Toba (Guaycuruan)'' (pp. 173-190) by María Belén Carpio and Marisa Censabella is the first paper of Part III of the volume, devoted to languages in areas other than the Mexico-US border area. Toba (Guaycuruan, Argentina) has three post-nominal relativization strategies that do not cater to the usual accessibility hierarchy, but rather code the informational nature of RCs’ contents. One marker, maʒe (related to a 3rd person pronoun and used only in relativization), introduces typical restrictive RCs that contain known or construable information. Maʒe RCs can relativize nominals in all syntactic functions, down to constituents of possessive constructions. Toba, however, also has a series of six demonstratives that code direction, position, or the perceptual or cognitive status of referents. These demonstratives can be used as such to introduce RCs that code new information that will be relevant down a storyline (the author used a set of nine narrative texts as a corpus). The same demonstratives can be combined with a topic marker to introduce RCs modifying topicalized nominals. Finally, the demonstratives can be used to build noun complement clauses, which share with RCs the status of being embedded inside a NP, but have different semantics (no coreference with a head noun).
In the ninth contribution, ''Between headed and headless relative clauses'' (pp. 191-211), Patience Epps discusses the headed/non-headed continuum in the RCs of Hup (Madahup, Brazil). Indeed, according to the author, the traditional two-way or the revised three-way distinction between headed, light-headed, and headless RCs cannot account for the RCs in Hup, which are nominalized clauses that are best conceived as being in apposition with a head element. This element can be: (i) a lexical noun; (ii) a semi-generic bound noun (such as a noun “-egg” when it refers back to a more specific referent, such as “chicken-egg”); (iii) a more generic classifier (referring to abstract properties of objects); and, finally, (iv) a plural marker, which is the weakest “nominal” to head a RC. Hup also has headless RCs. According to the author, Hup’s unmarked and non-restrictive headless RCs are the source of adverbial clauses, which are structurally similar to the RCs. However, the author adduces no evidence to support this particular claim (which goes against Givón’s paratax to syntax general pathway for RC formation). The lack of evidence for this claim contrasts with the meticulous and convincing argumentation in favor of four distinct points along the continuum between headed and headless RCs in Hup.
In the tenth contribution, ''Relative clauses in Seri'' (pp. 213-241), Stephen A. Marlett claims that Seri (isolate, Mexico) has nominalized, yet internally-headed RCs that have no relative pronoun and no special subordination marker. RCs in Seri occur less frequently than one would expect. For example, three folktales that contain 24, 60, and 132 clauses have, respectively, only 3, 8, and 29 RCs. In Seri, subjects, objects, and obliques can be relativized using different nominalization strategies: subject-oriented strategies (with 4 nominalizers); object-oriented strategies (with 5 nominalizers); and oblique-oriented strategies (with 4 nominalizers). If an object or oblique is relativized, the RC subject appears as a possessor of the nominalized verb. Possible (internal) heads are: nouns, pronouns, proper names, and null elements. If two NPs are in the RC, the head is easily identified, since it cannot be modified by a determiner.
In the eleventh contribution, ''Relative clauses in Gavião of Rondônia'' (pp. 243-252), Denny Moore shows that Gavião (Tupi, Brazil) only has nominalized clauses that can receive RC interpretation in appropriate referential contexts. Terminologically, the author essentially describes “nominalized” clauses throughout the paper and only in a final discussion section does he look back to see which of these clauses can fit the definition of a RC. Gavião has two discourse pronouns, mát and mére, that appear at the beginning of sentences and refer back to elements in previous sentences. The two pronouns can also be used as Verb Phrase (VP) or clause nominalizers. The pronoun mát is used for concrete nominalizations (substantives, places, actions), while mére is used for abstract nominalization (facts, reasons, manners; see p. 245). Contrary to discourse pronouns, nominalizers appear at the end of their clause. The development from pronouns to nominalizers is given as follows (slightly adapted by this reviewer): “Previous discourse + [mát/mére + S]” => “[VP/clause + mát/mére] + S”. The nominalized clause can have its subject, object, or possessor as an internal head, or it may contain a null head. It can also have an internal and an external head in the main clause in a correlative-like construction. Sometimes, the nominalized clause may be placed in apposition with a noun. According to the author, when these apposition clauses have an internal head, they can then be interpreted as RCs.
In the twelfth contribution, ''Relative clauses in Yucatec Maya: Light heads vs. Null domain'' (pp.255-268), Rodrigo Gutiérrez-Bravo discusses the question of headed versus headless RCs. Yucatec (Mayan, Mexico) has finite post-nominal RCs that are very similar to regular clauses (they have no subordinator). The language uses two RC strategies. Some RCs use a relative pronoun (identical with the interrogative form) that is shifted to the RC’s left boundary. These RCs can relativize nouns in nearly all syntactic functions (with corresponding gaps in the RC – except for the head-marking of core arguments on the verb). RCs with relative pronouns, however, cannot modify lexical heads unless they are obliques. In the second strategy, called the “gap” strategy by the author, the RCs have no relative pronoun and are comparable to English’s “the book I read...”. This strategy can also relativize most syntactic functions: subject, object, indirect object, temporal expressions, possessors, and prepositional objects (with preposition stranding). RCs of both strategies can be headless (in fact, this is almost the rule for the relative pronoun strategy). The author claims that when a RC appears seemingly without a head, one is, in reality, dealing with a null head RC, not a light-headed RC, nor a headless RC, nor a free RC. That is, Yucatec only has headed RCs, whether the head is overt or null. One argument in support of this claim is the fact that the null head can be modified by determiners, quantifiers, etc., that is, the same elements that can modify an overt noun. The author also shows that the null head notion is, in any case, independently needed elsewhere in the grammar of Yucatec.
In the thirteenth and last contribution, ''Questionable relatives'' (pp. 269-300), Marianne Mithun charts the development of the use of interrogative pronoun-based relative pronouns in the recent history of Tuscarora (Iroquoian, USA). The author finds that over a period of about a century, Tuscarora interrogative pronouns have expended their contexts and occurrence in RCs. This development, although very likely spurred by contact with English, follows the sequence of steps as defined by Heine and Kuteva (2006). The scheme, as adapted slightly by the author (cf. p. 284), has five stages: 1) simple questions (as in: Who came?); 2) indefinite complement clauses (e.g. I don’t know who came); 3) definite complement clauses (e.g. I know who came); 4) structures interpretable as headless RCs (e.g. I know the one who came); and 5) headed RCs (e.g. I know the student who came). In any language, interrogative pronouns can stand at various stages, although the overwhelming majority of languages at stages 4 or 5 are concentrated in Europe. In Tuscarora, the pronouns for ‘what, who, where’, and ‘when’ have today reached stage 4 (headless RC), although none of them had such a use in the 19th century. Also, no pronoun has reached the final stage. According to the author, one factor that may influence the pace of expansion of individual pronouns is the availability (or lack thereof) of constructions competing for the same function.
This is an interesting and informative book, with contributions from well-known linguists, all of whom show a deep level of knowledge of the languages on which they report, even beyond RCs. This allows for a comparison between RCs and related constructions. The papers deal with a number of typologically relevant aspects of RCs: headed versus headless RCs; externally-headed versus internally-headed RCs; nominalized versus finite RCs; the use of (interrogative-based) relative pronouns; restrictive versus non-restrictive RCs; the accessibility hierarchy; embedded versus extraposed RCs; the function of RCs in discourse; and the diachronic development and avatars of RCs. The findings and analyses in the papers are innovative and will be of interest to anyone dealing with RCs and the development of language structure.
However, one issue that most papers evoke, about which there seems to be no agreement, is the question of what is the relevant constituent for the embedding of RCs; in particular, externally-headed non-nominalized RCs. Some papers assume that RCs are embedded foremost in the main clause (Givón, Guerrero (pp. 97f), Estrada-Fernández (p. 127)), while others say that the relevant embedding is inside the NP (Van Valin (p. 56), Thornes (p. 155), Carpio and Censabella (p. 178), Gutiérrez-Bravo (p. 256) and, maybe, Moore (p. 244)). Álvarez González explicitly discusses both options (p. 69). Givón (2001: 175), as cited by Álvarez González (p. 69), does claim that the RC is embedded within the NP, but in his contribution to this volume, he focuses on the main clause as the relevant embedding constituent. Maybe for this reason, Givón considers that constructions containing a head noun and a RC would still be at the pre-RC stage if they are not embedded in the main clause (for example, if they are separated from the main clause by a pause; cf. the Bambara examples (3a-e), pp. 5f). If one considers the NP as the relevant embedding constituent, then these constructions would not contain a “would be” RC, but rather a typical noun-modifying RC.
There are also specific issues with some of the papers. In Lilián Guerrero’s contribution, on p. 100, there are two comments on inadequate (non-matching) data and a comment on a non-existing example (3c). The discussion section is sometimes difficult to follow, such as when the author claims - in an attempt to explain why non-relative subordinate clauses have no gap - that direct perception always involves a state of affairs and its participants, which then must all be always expressed. However, on p. 120, example (32b) shows that one can perceive a state of affairs long after main participants are gone (also, on p. 136, Estrada-Fernández shows that Equi clauses have gaps without being RCs). The contribution by Marlett is sometimes difficult to read, maybe due to the complexity of the language discussed. Also, some RCs he labels “stacked” are in fact simple coordinated RCs (see data (22, 23, 30, 40-42)) and some others he labels “center-embedding” are probably right-embedding (see data (50)). Some minor issues are also found in Estrada-Fernández’s contribution, such as the claim that a RC provides an assertion about the relativized noun (cf. p. 127). Some typos can be found on p. 129, 135n5, 137, and 142, where a verb meaning ‘cut’ is glossed as ‘lend’.
Givón, T. (2001). Syntax: A introduction. Vol. II. Amsterdam: John Benjamins.
Heine, Berndt and Tania Kuteva (2006). The changing languages of Europe. Oxford: OUP.
Epps, Patience (2007). Escape from the noun phrase: The adventures of a relative clause. Paper presented at the Seminario de Complejidad Sintáctica, Hermosillo, Sonora.
Keenan, Edward L. and Bernard B. Comrie (1977). Noun phrase accessibility and universal grammar. Linguistics Inquiry 8(1):63-69.
| ABOUT THE REVIEWER:
Mahamane L. Abdoulaye teaches linguistics at the Abdou Moumouni University, Niamey. His main research focuses on Hausa and Zarma Chiine morphology, syntax, and semantics.