Grammaticalization and (inter)subjectification in an Iranian modal verb: A paradox resolved by Dutch

ABSTRACT

This paper deals with the grammatical and semantic development of a modal verb in four West Iranian languages: gu/ga in Kahangi, gijabon in Semnani, boGostæn/bogostæn in the Takestan dialect of Tati and goan in Vafsi. Field work data demonstrate that, from the perspective of the grammaticalization and (inter)subjectification literature, this verb in these languages poses a challenge. It occurs as a non-grammaticalized full verb and as a grammaticalized auxiliary. Yet the full verb features an arguably more (inter)subjectivized meaning than the auxiliary: the former expresses volition, the latter dynamic, deontic and epistemic modal meanings. The absence of historical data for these languages does not allow the puzzle to be resolved directly, but a comparison with the modal verb hoeven ‘need’ in Dutch, which in the present-day language has properties similar to those of the Iranian verbs, and which has been investigated diachronically, suggests a solution: the present situation may be the result of a convoluted diachronic evolution, in which the semantic and grammatical developments do not align. The case thus demonstrates that one cannot easily draw conclusions from synchronic observations about diachronic relations between forms and/or meanings.

KEYWORDS:

• Modality
• modal verbs
• grammaticalization
• (inter)subjectification
• West Iranian languages
• Dutch

Acknowledgements

We are grateful to the anonymous reviewers for their useful comments on earlier versions of this paper. All mistakes obviously remain our own. We would also like to thank all the participants who patiently and enthusiastically took part in the questionnaire and in the follow-up interviews. We especially wish to mention Mrs Soghra Ghoreyshi, Mr Hossein Ali Jokar and Mr Ahmad Darabi.

Data availability statement

The data that support the findings of this study are available from the corresponding author, S.K., upon reasonable request.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 Kahang is a village south of Ardestan, a town in the northeastern part of the province of Isfahan in central Iran. Semnan is a city in the province with the same name, situated in the north of Iran, east of Tehran. Takestan is situated in the Qazvin province in the north of Iran, west of Tehran; the dialect spoken there is part of the Southern Tati dialect group (cf. Stilo, 1981). Vafs is a city in the Markazi province, southwest of Tehran.

2 Most authors (see Rezai Baghbidi, 2009; Skjærvø, 1989; Windfuhr, 1989, 2009) divide the Iranian languages into two major groups: roughly, West Iranian includes the languages spoken in Iran and further west, while East Iranian covers all the other languages.

3 A gX verb also exists in other Iranian languages, including Farzandi, Qohrudi, Abyanei, Meimei, Khansari, Yarani, Abuzeidabadi, Joshaghani, Varzanei, Mahallati, Sivandi, Gurani and the Chali dialect of Tati. Other forms with similar semantic and grammatical properties exist in yet other Iranian languages. This includes, at least, Naeini va and Shahmirzadi vænæ. None of these languages was covered by the survey on which the present study is based, however.

4 Glossing is in accordance with the Leipzig Glossing Rules (http://www.eva.mpg.de/lingua/resources/glossing-rules.php) with the addition of DER.PRFX ‘derivational prefix with unclear semantic status’, EZ ‘ezafe’, INFL.PRFX ‘inflectional prefix with arguable grammatical status’ and SINF ‘short infinitive’.

5 For reasons of space, each category is illustrated with examples from only one (randomly chosen) Iranian language. As some examples are ambiguous out of context, they may receive one or more other readings beyond the one presented. (We indicate this when it applies.) While this may be somewhat confusing, it is inevitable given the polysemous nature of modals cross-linguistically.

6 Example (5b) could also be used deontically (when ‘I’ is under the moral obligation to be in Isfahan next year, e.g. for the sake of his/her needy mother) or as a directive (e.g. when ‘I’s boss has ordered him/her to be there; see below on these notions). In the participant-imposed dynamic reading, the speaker means to express that there are practical circumstances forcing him/her to be there.

7 The examples in (9) are ambiguous: both may also be used to express participant-imposed dynamic modality, and (9b) could also be directive.

8 The examples in (13) can also be understood deontically, as concerning moral considerations (whether it is good or bad to do it).

9 The semantic notions in this hierarchy that are relevant for the discussion later in this paper were defined in §3.1 above. For the sake of clarity, here is a summary definition of the other notions: evidentiality concerns the marking of the information source; time involves the situation of the state of affairs in time; quantitative aspect marks the frequency of the state of affairs (iterative, habitual, etc.); phasal aspect marks the state of deployment of the state of affairs (inchoative, progressive, egressive, etc.).

10 In Kahangi, unlike the other three languages, there is no infinitival form for this verb, which is common for quite a few verbs in some of the West Iranian languages. The Takestan Tati form has two alternative pronunciations, as indicated; we do not repeat the alternative in the remainder of this paper.

11 In these languages one might expect a transitive verb to only have an oblique or clitic subject in the past tense; however, there are some verbs, including this volitional verb, which select a clitic subject in both present and past tense. See further below for more detail.

12 The Iranian languages feature three different be- prefixes: one marks the subjunctive mood, one marks perfective aspect and the third marks the imperative mood. Some linguists (cf. Dabir-Moghaddam, 2013) assume that the subjunctive and imperative markers are actually the same, and only the perfective marker is distinct. The situation in Kahangi is more complex though. While for most verbs in this language the subjunctive and perfective is marked with be-, for the volitional main verb the subjunctive form of the verb ‘be’ (bo) is used, as in ɡu=m bo ‘if I want’ and ɡɑ=m bo ‘if I wanted’.

13 The diachronic evolution of the split ergativity system in the Iranian languages is unclear and is not directly traceable. As a consequence, there is disagreement among linguists regarding whether there is an ergativity system in these languages at all. For some, including Bynon (1980) and Jügel (2009), the main characteristic of languages featuring an ergative system is the agreement between the object and the verb. Since in the Iranian languages there is no such agreement, we cannot call them ergative according to these scholars. Others, including Dabir-moghaddam (2013, 2018) and Haig (2008), consider the distinct agreement system for transitive verbs in the past tense to be a remnant or a type of ergativity. We follow Haig and Dabir-Moghaddam: we consider the four languages covered in this study as featuring a type of ergativity, or, as Dabir-Moghaddam (2013) puts it, as involving a ‘tripartite type of language in the past tense domain’.

14 These Iranian languages are pro-drop languages. The pronoun can be dropped easily since the subject is still coded, with a clitic on the gX verb in the normal main verbal use (cf. (24b)), and with the subject endings on the embedded main verb (cf. –ume in bæxɑn-ume).

15 We should mention that no other modals in these languages show the kind of complexity in the syntax–semantics interface characterizing the gX verb. Hence, we may assume that they have all evolved through a normal grammaticalization and (inter)subjectification process. This includes ɢɑde ‘should’, bebu ‘be possible’, tiɢ vontemon ‘be able’ and ʃɑ ‘can’ in Kahangi, mæzonɑn/bozonijon ‘be able’ or ‘know’ and mæbu ‘be possible’ in Semnani, botonæstæn ‘be able’ and megærdæ ‘be possible’ in Tati, and tʃuɑn ‘be able’ and ærbu ‘be possible’ in Vafsi.

16 One may wonder how the impersonal main verbal use in (24c) in §4 diachronically relates to the regular main verbal use in (24a–b) and the auxiliary use in (24d–e). One hypothesis might be that the impersonal use has been the ‘mediator’ between the regular main verbal use and the auxiliary use, or in other words, that the evolution from the main verbal use in (24a–b) to the auxiliary use in (24d–e) went through the intermediary step in (24c). This fits perfectly into a scenario of gradual grammaticalization. An alternative hypothesis emerges if we compare our present case with data from Middle Persian, also a split ergative language. In Middle Persian all transitive verbs in a past form show agreement with the object on an auxiliary or another element preceding the main verb, and the main verb itself is non-finite. The intransitive verbs, however, have a nominative-accusative case marking system in which the subject is marked on the main verb with affixes. In New Persian the system of the intransitive verbs is generalized, and all verbs, including the transitive ones, mark agreement with the subject through affixes on the main verb. In this perspective one might postulate that the main verbal use in (24c), which lacks agreement marking, is the original form, and that the main verbal use in (24a–b), which does feature agreement marking, has emerged from it. The auxiliary use in (24d–e) would then be an even later development. The absence of historical data for our Iranian languages prevents us from empirically checking the plausibility of these hypotheses. The issue is of no further import for our present paradox, however: in either scenario there is a development from a main verb with a volitional meaning to an auxiliary expressing a range of modal and related meanings.

17 One might remark that for English will, Bybee and Pagliuca (1987) and Bybee et al. (1991, 1994) have suggested that volition (among others) is the source for the future meaning. This would be an evolution which runs counter to the assumption in Figure 1. And if one assumes that the epistemic meaning of will has evolved out of the future meaning, this would then also be a case where volition has been the (indirect) source for a modal meaning. Yet Bybee and colleagues’ evidence for their claim is weak. Both volition and future occur side-by-side as meanings of will from Old English onwards, and on the basis of the few incidentally selected examples they quote it is not possible to determine which of these meanings has been the source for the other.

18 The following Dutch examples are all based on (strongly simplified versions of) corpus instances used in Nuyts et al. (2018).

19 In PDD this modal is actually developing a new main verbal use, which has evolved out of the auxiliary use and has grammatical and semantic properties which differ from the original main verbal use (e.g. it maintains the (inter)subjective meanings of the auxiliary use). However, this is of no further relevance for the present story and is disregarded in the scheme in (31) below.

20 Legend: MV = main verb, AUX = auxiliary, MD = Middle Dutch, END = Early New Dutch, PDD = Present-Day Dutch, (I)S = (inter)subjective meanings.

21 The meaning of this form is a matter of dispute. Alternative views are offered in Durkin-Meisterernst (2004, p. 29) and Humbach (1974, p. 198). Humbach postulates ‘be faulty’, a meaning which is also mentioned by Cheung, and which may be a second or maybe even an older sense of the original form (one can easily imagine a meaning change from ‘be faulty’, or ‘lack’, to ‘need’). Durkin-Meisterernst postulates that the source form already had the meaning of ‘desire’ or ‘wish’ (i.e. volition). The question is, however, whether he is then referring to the oldest stages of the evolution. The analyses in Cheung and Humbach suggest not. Descendants of the form are postulated to code both ‘need’ and ‘desire/wish’ in Parthian (Boyce, 1975, p. 42; Cheung, 2007, p. 95; Rezai Baghbidi, 2006, p. 172), Khotanese, Sogdian (Cheung, 2007, p. 95; Zarshenas, 2017, p. 247), Choresmian, Bactarian (Cheung, 2007, p. 95) and Middle Iranian. Kellens and Pirart (1990, p. 238), finally, consider the reconstructed form as unanalyzable.

Additional information

Notes on contributors

Sepideh Koohkan

Sepideh Koohkan obtained her PhD from Tarbiat Modares University and the University of Antwerp in 2019. She has been a teacher of EFL since 2001, and a lecturer of English and general linguistics at the Imam Khomeini International University since 2012. Her fields of interests are modality, Iranian languages, typology and language documentation.

Jan Nuyts

Jan Nuyts (PhD 1988, Habilitation 1994) is a Professor in the Linguistics Department at the University of Antwerp (Belgium). His major research area is cognitive-functional semantics. His current focus of attention concerns the cognitive and functional structure of ‘time–aspect–modality–evidentiality’ or ‘qualificational’ categories – and the modal and evidential categories in particular – and their linguistic expressions, synchronically and diachronically, and what one can learn from them regarding the ‘language and thought’ issue. His most important book publications include Aspects of a cognitive-pragmatic theory of language (Amsterdam, J. Benjamins, 1992), Epistemic modality, language and conceptualization (Amsterdam, J. Benjamins, 2001) and The Oxford handbook of modality and mood (Oxford, Oxford University Press, 2016; edited jointly with Johan van der Auwera).