Inflectional periphrasis in Persian

Modern Persian conjugation makes use of five periphrastic constructions. We contrast the properties of these five constructions and argue that they call for different analyses. We propose contrasting analyses relying on the combination of an HPSG approach to feature geometry and syntactic combination, and an approach to paradigm organization and morphological exponence based on Paradigm Function Morphology. This combination of analytic tools allows us to treat the whole array of periphrastic constructions as lexical in origin—no phrasal construction or multi-word lexical entry of any kind is required. Grammars of Persian (e.g. Lazard et al., 2006) distinguish five conjugational periphrastic construction types. The passive construction is based on an inflected form of šodan ‘become’ preceded by a perfect participle (1). So-called ‘perfect’ forms are based on an inflected form of budan ‘be’ preceded by a perfect participle (2). The auxiliary is a full word (2a) or a clitic, (2b) depending on tense and mood, and giving rise to different syntactic and semantic properties. The future is formed with a special present tense form of xâstan ‘want’ followed by a bare stem (3). Finally, the progressive is based on an inflected form of dâštan ‘have’ followed by a finite form (4).1 (1) In this tâblo painting foruxte sold mi-šav-ad. UNBD-become.S1-3SG ‘This painting is sold.’ (2) a. Maryam Maryam in this tâblo=râ painting=DDO foruxte sold bud. be.S2.3SG ‘Maryam had sold this painting.’ b. Maryam Maryam in this tâblo=râ painting=DDO foruxe=ast. sold=be.PRS.3SG ‘Maryam has sold this painting.’ (3) Maryam Maryam in this tâblo=râ painting=DDO xâh-ad want.S1-3SG foruxt. sell.S2 ‘Maryam will sell the painting’ Aspects of this work have been presented at the Décembrettes 6 International Morphology Conference (December, 2008), at the HPSG Seminar at U. Paris Diderot (March, 2009), at a Morphology Meeting in Surrey (April, 2009), and at the HPSG 2009 Conference (Göttingen, July 2009). We thank for their comments and suggestions the audiences at these events, and in particular Anne Abeillé, Gilles Boyé, Dunstan Brown, Patricia Cabredo Hofherr, Greville Corbett, Berthold Crysmann, Gerald Gazdar, Stefan Müller, Ivan A. Sag, Gregory Stump, Jesse Tseng, and Gert Webelhuth. This work was supported by a grant from Agence Nationale de la Recherche and Deutsche Forschungsgemeinschaft to the Franco-German project ‘PER-GRAM: Theory and Implementation of a Head-driven Phrase Structure Grammar for Persian’. The glosses use the following abbreviations. BD: bounded aspect; DDO: definite direct object; EZ: Ezafe; NEG: negation; PAF: pronominal affix; PRF: perfect; PRS: present; PST: past; S1: first stem (a.k.a. the present stem); S2: second stem (a.k.a. the past stem); SBJV: subjunctive; UNBD: unbounded aspect. 27 (4) Maryam Maryma dâr-ad have.PRS-3SG in this tâblo=râ painting=DDO mi-foruš-ad. UNBD-sell.S1-3SG ‘Maryam is selling the painting.’ The differing properties of these five types of periphrasis stem from different origins as finite, infinitival or participial complements, and different degrees of grammaticalization, going from the quasi-analytic passive to the recently morphologized present perfect, through truly periphrastic forms that need to be integrated into inflectional paradigms despite being multi-word expressions. We assume that the different properties call for different analyses. We propose five contrasting analyses relying on the combination of an HPSG approach to feature geometry and syntactic combination, and an approach to paradigm organization and morphological exponence based on Paradigm Function Morphology (PFM; Stump, 2001). Interestingly, this combination of analytic tools allows us to treat the whole array of periphrastic constructions as lexical in origin—no phrasal construction or multi-word lexical entry of any kind is required. 1 Synthetic conjugation in HPSG/PFM Before we address the analysis of periphrastic forms, we start with an account of synthetic conjugation. (5) lists the synthetic subparadigms of the lexeme xaridan ‘buy’, using the positive 2SG form as an illustration. (5) a. Finite forms: i. Simple present: mi-xar-i ii. Simple bounded past: xarid-i iii. Simple unbounded past: mi-xarid-i iv. Simple subjunctive: be-xar-i v. Imperative: be-xar b. Nonfinite forms: i. Infinitive: xarid-an ii. Present participle: xar-ande iii. Perfect participle: xarid-e iv. Gerund: xar-ân Persian verbs exhibit a morphomic stem alternation (here xar vs. xarid). Neither stem is predictable from the other in general, and both stems are used in a combination of contexts which do not form a natural class. Affixal exponents realize unbounded aspect in the indicative (mi-), irrealis mood (be-), negation (naor ne-, not illustrated here), type of nonfinite form (-e vs. -ande vs. -an vs. -ân), and subject agreement for finite forms. Within Paradigm Function Morphology, this rather simple position class system can be accounted for using the series of rule 28 III II I IV V namistem-selection -e -am ne-ande -i/∅ bean -ad/∅ -im -id -and Table 1: Rule blocks for Persian synthetic conjugation blocks outlined in table 1. Remember that in PFM, realization rules are organized in successive blocks. When attempting to realize a given set of morphosyntactic feature, the most specific applicable rule within the block is chosen. (6) are sample rules from block V, written in an attribute-value matrix format.2: while (6a) asks that finite verbs with a 2SG subject take the suffix -i, the more specific (6b) indicates that the suffix is dropped in the imperative. (6) a. [ PHON X LID Y ] , σ : [ PER 2 NB sg ] −→ [ PHON X⊕i LID Y ] (block V) b. [ PHON X LID Y ]