The structure of content questions in Cheyenne

Abstract

This article has as its main goal to investigate interrogative sentences in Cheyenne and therefore adds to the diversity of such analyses found in other Algonquian languages. This paper attempts to provide evidence that content questions in this language differ in structure depending on the status of the questioned element as an argument or adjunct and that content questions involving an interrogative pronoun are bi-clausal, as they exhibit a cleft-like structure such that the fronted ‘wh’-word functions as the stative predicate of a copular clause that is generated sentence-initially encoding the focus of the construction. By contrast, content questions requiring an interrogative adverb are mono-clausal and exhibit ‘wh’-movement in the traditional sense, that is through ‘wh’-fronting. Thus, this paper intends to contribute to the long-standing discussion on whether Algonquian ‘wh’-questions have a mono-clausal structure, involving traditional ‘wh’-movement or a bi-clausal structure, exhibiting ‘wh’-clefting.

Keywords:

• Algonquian language family
• Cheyenne
• content questions
• ‘wh’-movement
• ‘wh’-clefting

Notes

1. The examples I use throughout this paper come from two different sources: from published studies, especially Leman et al. (2006), and native speaker elicitation via telephone and email correspondence from 2010 to 2016. Unless accompanied with citations, examples come from my own notes. Financial support for this research has been provided by the Spanish Ministry of Economy and Competitiveness (MINECO) through the projects FFI2014-53788-C3-1-P and EDU2014-54673-R. I would like to express my heart-felt gratitude to my late and dear friend Ralph Redfox for having kindly and patiently shared their beautiful language with me and ‘Richard Littlebear and Eugene Blackbear for his invaluable help. I would also like to thank Wayne Leman, Donald G. Frantz and Andrew Cowell for having helped me with my questions so kindly and patiently. Of course any errors are entirely my own responsibility.

2. For detailed information on Cheyenne word order, see Leman (1999).

3. More accurately, according to Leman (1999: 37), and just like other Algonquian languages (Tomlin & Rhodes 1994, for Ojibwa), Cheyenne appears to have a tendency to place the newsworthy information (any element expressing a contrast, introducing or changing a topic, or representing new information) preverbally and inferable or evoked information postverbally. Although the correlation between focus and newsworthy information is complete, it seems logical to think that ‘wh’-words, which are traditionally assumed to be inherently focal (see Erteschik-Shir (1986) for a discussion), must be placed at the front of a sentence preceding both the main verb and any other more topical participant.

4. Leman (1986: 14) considers the Conjunct prefix tsé-, which tends to occur in relative (Participial mode), complement (Indicative mode) and adverbial clauses (Indicative mode), a realis marker.

5. Abbreviations used in this paper: (1) – first person singular agreement, (2) – second person singular agreement, (3) – third person / proximate singular agreement, (4) – fourth person / proper obviative agreement, (5) fifth person / further obviative agreement, (11) – first person plural agreement, (22) – second person plural agreement, (33) – third person plural agreement, (I) – inanimate singular agreement, (II) – inanimate plural agreement; a – adjunct, agr – agreement, an – animate, anaph – anaphoric, cisl – cislocative - conj – conjunct marker, deic – deictic, dis – distal, ditr – ditransitive, dub – dubitative mode, endoph – epenthetic sound – epe, endophoric,  exoph – exophoric, fai – animate intransitive final stem, fii – inanimate intransitive final stem, fta – animate transitive final, fut – future, i – initial stem; if – Illocutionary Force, imp – imperative illocutionary force, inan – inanimate, ind – indicative mode, int – interrogative illocutionary force, loc – locative, o – object, obv – obviative, part – participial mode, past – past, pl – plural, prox – proximal, s – subject, sing – singular, refl – reflexive, rr – relative root, transl – translocative, v – verb, vai – animate intransitive, vta – transitive animate, vti – transitive inanimate. Cheyenne verbal theme-sign suffixes (Bloomfield 1946) are glossed according to the person features of the arguments they correspond to as ‘more agent-like:more patient-like’. For example, ‘1:2’ would stand for a first person agent and a second person patient. In view of the complex morphophonology of this language, I will treat the agreement suffixes as an unanalyzed whole to keep the glosses as uncluttered as possible.

6. The concept ‘obviative third person’ refers to a grammatical person marking that distinguishes a more salient third person referent (named proximate or third person) from a non-salient third person referent (normally referred to as obviative or fourth person) in a given discourse context. Although it is extremely rare, there could even be a fifth person or further obviative in a specific context. Finally, the term ‘inanimate’ refers to a third person argument referring to an inanimate participant.

7. Building upon the Person – Salience hierarchy, the direct / inverse marker serves to indicate the role that is assigned to every argument of a predicate in a transitive construction so that, while the direct construction is used when the subject or more agent-like participant of the transitive clause outranks the object or more patient-like participant in the hierarchy, the inverse is used when the object outranks the subject.

8. The difference between these two types of Imperative, that is to say, between ‘Immediate’ and ‘Delayed’ Imperatives, lies in the immediacy with which the command is expected to be carried out.

9. This restriction that no interrogative pronoun can occur in embedded questions also applies in Blackfoot (Frantz 1991) and Arapaho and Gros Ventre (Cowell, in personal communication).

10. This marker is used to express the dubitative or inferential mode of the verb in the Independent order.

11. See page 7.

12. This correlation between the choice of verbal order and the type of interrogative element is not observed in the same way in the other Plains Algonquian languages. Blackfoot appears to work similarly as Cheyenne with regard to the formation of content questions with an interrogative pronoun, since many of these questions involve clausal nominalization. Frantz (1991: 120) calls this construction a conjunct nominal because, although the verbal form in this construction carries suffixes similar – but not identical - to those of the conjunctive order, they function as a cleft clause:

Table

However, some content questions with ‘who’ or ‘what’ require suffixes of the independent paradigm:

Table

As regards content questions containing an interrogative adverb, most of them require inflectional suffixes of the independent paradigm:

Table

However, it is also possible to build content questions with ‘where’ or ‘when’ through clausal nominalization:

Table

This variety of constructions may be due to the fact that Blackfoot does not appear to mark the distinction between the Independent and Conjunct orders so clearly as Algonquian languages do (Cook 2014: 311).

Arapaho (Cowell & Moss 2008) behave slightly differently from Cheyenne in this respect. Thus, similarly as Cheyenne, Arapaho content questions including the interrogative adverbs ‘where’ and ‘when’ always require the use of the Independent order:

Table

By contrast, unlike Cheyenne, content questions containing the interrogative pronoun ‘who’ and ‘what’ in Arapaho may take either Independent or Conjunct order inflections:

Table

Very interestingly, the use of Independent or Conjunct order in Arapaho appears to be associated to transitivity so that an intransitive construction requires Independent order inflection and a transitive construction entails Conjunct order inflection. As regards the syntactic structure of the questions involving Conjunct order, Cowell and Moss (2008: 246) argues that the Conjunct order is used to form relative clauses, which require Conjunct order dependent participle inflection, and, consequently, the interrogative pronouns would function like the heads of a relative clause, hence these constructions could be considered basically clefted questions.

Finally, Cowell (p.c.) states that Gros Ventre works analogously as Arapaho.

13. Although I have not been able to find out why, some of my consultants also used the interrogative pronouns táase ‘which’ (INAN.SING) / táasévoóe (AN.SING) or taasévoone ‘which’ (AN.PL) / táasévoonėstse (INAN.PL) or tónėsto- ‘how much / many’ with the Independent order.

14. Although this structure appears to represent an example of a scope marking construction, it should not be considered so, since the scope marker cannot be iterated at the left edge of any of the clauses embedded within the scopal position in examples of long-distance dependency and the variable, represented by the preverbal particle -hése-, can also be found in non-interrogative environments.

15. I gloss the prefixes né(he)- and hén(e)- in (42a) and (42b) as deictic, rather than ‘who’ or ‘what’ respectively because they are very commonly used in elements such demonstratives to distinguish between animate and inanimate referents.

16. If the assumption that this morpheme functions as a copula, then we should argue that some interrogative pronouns, such as hénáa´e ‘what’ or táase ‘which’, a d-linked interrogative pronoun used to ask about inanimate singular participants, show a covertly-expressed copula.

17. This pronominal affix could be used to refer to an abstract referent in an intransitive construction involving a state, perhaps including the interrogative IF particle –he, which would imply that the interrogative word is in situ within the copular sentence – after it has been extracted from the original main clause to form the cleft construction – so the interrogative particle is necessary to distinguish the declarative and the interrogative interpretation of the sentence.

18. Cook (2008: 283) makes the same claim for Plains Cree.

19. Wolfart (1973: 34) also states that in Plains Cree the ‘wh’-word awina ‘who’: (i) may stand alone as a complete utterance; (ii) may occur in an equational sentence; or (iii) may function predicatively ‘with a conjunct clause depending on it’, i.e. a verbal clause with an e- or ka- complementizer or conjunct marker. 

20. These Conjunct Participial order clauses would function as relative clauses complementing an overt- or covertly expressed antecedent and would, therefore, behave as dependent, rather than as main, clauses. In fact, they can only stand on their own when they are interpreted as headless relative clauses (e.g. Tsé-a´xaemėstse = ‘The one who is crying … ’).

21. A and U stand for ‘Actor’ and ‘Patient’ respectively.

22. The possibility to place a constituent before the ‘wh’-word will be addressed below.

23. According to Cheng (1991), wh-in situ languages have the following properties: (1) the wh-word is in A-position (at S-structure), (2) questions are associated with a question particle, and (3) ‘wh’-words are polarity items, i.e. they are ambiguous between an interrogative and an indefinite interpretation.

Examining to what extent these four properties are present in Cheyenne:

(1) The fact that ‘wh’-words strictly occur in sentence-initial position does not necessarily indicate they are not occupying an A-position.

(2) Cheyenne only has an interrogative particle in polar questions. According to Cheng (1991), if there is a Q-particle for yes/no questions, then there must also be a ‘wh’-question particle, so Cheyenne seems to counter-exemplify this claim. However, Cheng allows for the possibility that the ‘wh’-Q particle may be overt or covert.

(4) Cheyenne presents some forms for indefinite pronouns (e.g. neva´esestse, hénáá´énese, hénová´énese) that resemble considerably interrogative pronouns but they do not appear to function as interrogative pronouns. With the exception of these forms, interrogative and indefinite pronouns are clearly distinct forms. In addition, the ‘wh’-environment is distinguished by obligatory initial position and the use of the conjunct verb form. By contrast, indefinites behave like ordinary NPs since they can occur either before or after the verbal complex and free occur with the independent form of the verb.

In sum, Cheyenne does not satisfy all of the ‘wh’- in situ diagnostics listen above and consequently it is not a ‘wh’-in situ language.

24. Although clefted questions tend to involve ‘wh’-fronting, they can also display in-situ ‘wh’-words, as the following example in Japanese shows:

Table

25. Although all my consultants concur on the order VA (46B and 47B) as the unmarked pattern for these two answers, they also agree that the order AV (46B´ and 47B´) is grammatical in this context, thereby highlighting the flexibility of Cheyenne regarding the positioning of words in a sentence.

26. The same could be said to occur in some examples including a deictic element preceding a ‘wh’-word such as:

Table

In these examples, the initial element could be considered a topicalized constituent, hence these questions could be equivalent to others such as the following in languages like Spanish:

Table

27. Agreement appears to be obligatory on NPs, including those whose nucleus is a proper noun:

Table

28. Unlike Cheyenne, interrogative adverbs can also appear in cleft questions in other languages (e.g. ’Where was it that you had dinner?’, in English)

29. Furthermore, Cheyenne also uses tsé- to introduce some adverbial subordinate clauses:

Table

30. For example, according to Blain (1997), in Plains Cree both a ka- complementizer (using the Conjunct order) or an e- complementizer (using the Conjunct order) can be used, and it is even possible to use no complementizer at all (using the Independent order).

31. A morphological division within the Conjunct order is also very typical of Algonquian languages. Thus, under certain circumstances, some Conjunct verb forms undergo a pattern of ablaut of the first vowel of the verb stem known as ‘initial change’ (Lochbihler & Mathieu 2016) so that they are commonly referred to as ‘changed conjunct’, whereas the Conjunct verb forms, which do not undergo this phenomenon, are commonly called ‘unchanged conjunct’ (Richards 2004: 6). Cheyenne does not show any traces of this process involving a shift in vowel quality, since presumably this language has gradually grammaticalized the Initial Change process in the form of an overt complementizer. This complementizer tsé- occurs only in the Conjunct verbal order, which is associated with suffixal person agreement.

32. The term ‘participial’ is used in the sense of verbal forms that modify nouns, which behave analogously to relative clauses. Participial mode is not only found in ‘wh’-interrogatives and relative clauses, but also in focus constructions:

Table

In conclusion, tsé- always occurs in contexts involving the fronting of an operator, that is, in clauses in or through which a ‘wh’-element has moved.

33. It is also of note that participle agreement (Bruening 2004: 259) occurs successive-cyclically on every verb along the path of extraction, that is, it marks every clause a ‘wh’-element has moved through, even in a clause including a predicate of which such a ‘wh’-element is not a semantic argument, as can be clearly seen in long-distance questions such as the following:

Table

It is very interesting to notice the use of participle agreement in the first verbal form, namely tsénêhetatose ‘(my grandfather) told him’, which is not the relative clause modifying the ‘wh’-element.

34. See page 8.

35. The ban on multiple ‘wh’-questions in cleft constructions can be apparently explained by arguing that that the ‘wh’-word is the predicate, rather than an argument, of the nominal clause and there can only be one predicate per predication (Calabrese 1984, 1987). Furthermore, according to Blain (1997: 88), the clearest evidence of the absence of overt wh-movement involves the prohibition of multiple ‘wh’-questions, so she proposes that Plains Cree wh-phrases, like focused NPs, are clefted.

36. This correlation between ‘wh’-clefting and the presence of island effects depends on the condition that the ‘wh’-phrases display movement. In fact, a common analysis of clefts (Chomsky 1977) is that clefts involve base generation of the wh-word which binds a silent operator that undergoes A’-movement. For example, Chang (2000) claims that Tsou questions are indeed cleft constructions but its ‘wh’-words remain in-situ, hence they do not exhibit island effects.

37. Besides, Cheyenne has a rich verbal morphology including a theme morpheme indicating whether the action is direct or inverse and a number of pronominal affixes offering grammatical information (e.g. person, number, animacy, obviation, syntactic relation, etc.) about the obligatory arguments of the predicate.

38. The only possibility for coreference with respect to two participants, for instance between a ‘wh’- operator and a pronominal affix standing for the subject, would be in a reflexive construction such as:

Table

39. Because two co-arguments being co-valued must be marked as reflexive, we should use a bi-clausal sentence such as ‘*Who_i does he_i think that Mary likes t_i ?’ to illustrate strong crossover.

40. I venture the hypothesis that in the past there could have been a distinct marking for further obviative in nouns: in this case it could be something like he-škemo.

41. One of the consultants, however, notes that the example (78b) could also be interpreted as ‘Who_i t_i likes his mother?’, understanding the interrogative pronoun as the further obviative (5), the possessor of the NP as the proximate (3) and the possessed NP as the normal or proper obviative (4). This is due to the lack of distinctive nominal marking between the proper obviative and the further obviative, which does not allow us to establish an unambiguous correspondence between participants and semantic roles.

42. One of my consultants acknowledges that nowadays a less complex option would definitely involve using the same construction as that in (56) but with a non-coreferential reading:

Table

This fact is very significant because it would mean that the obligatory coreference between proximate participants also seems to be relaxing.

43. Taking into account that the possessed NPs of a third person possessor must always be obviative, not proximate (Rhodes 1992; Dryer 1998: 37; among others), in my humble opinion there could be a mistake regarding the marking of the agent of the matrix verb, which should bear obviative marking.

44. This is an instance of long-distance agreement or copying-to-object.

45. I assume that there might formerly have been a different form used to express a further obviative agent acting on a second person patient but that nowadays, owing the rare occurrence of this situation, the form crossreferencing a proper obviative agent is used instead.

46. The difference might lie in the length of the discourse span, which might range from a single clause to a whole text.