‘Religion’ in Late Antique Zoroastrianism and Manichaeism Developing a Term in Counterpoint

This article evaluates the development of a generic term for ‘religion’ in late antique Manichaeism and Zoroastrianism. It examines linguistic indications of the use of dēn/δēn as a generic term in the Manichaean Middle Iranian corpora, i.e. Middle Persian, Parthian, and Sogdian, as well as in the corpus of Zoroastrian Middle Persian. The paper considers declination in the plural, the attribution of universal quantifiers or demonstrative adjectives, comparison, and selection as they occur in the above corpora to be indica-tors of generic concepts. Acknowledging that third-century Manichaeism shaped the term for ‘religion’ in the Persian Empire, the paper scrutinizes the reflections of this formative process in Sasanian and also early Islamic Zoroastrianism. The resulting analysis of the linguistic evidence indicates that the newly coined Manichaean concept of ‘religion’ did not find considerable echoes in late antique Zoroastrianism. Furthermore, an investigation of the term daēnā - in the Avestan sources provides earlier evidence for the formation of the term ‘religion’ in pre-Sasanian Zoroastrianism. Finally, the paper highlights the significance of religious contact for the formation of a generic concept of religion. generic concept of religion, Manichaeism, Zoroastrianism, Sasanian period, early Islamic period, daēnā/dēn generic concept religion as what development term of religion in Manichaeism restricted to this religion and what has the generated abstract term found its way into other religions which were in contact with it in Sasanian Iran? In this article, I examine the case of Zoroastrianism to answer the question whether Zoroastrian authors used the substantive dēn in Middle Persian texts as a generic term nearly equivalent to Manichaean ‘religion.’ religion of nobles’ ( ēr-dēn ), ‘the religion’ ( rāst-dēn ), and of ( ) to ‘bad’ religion, presented as ‘bad/worse religion’ ( ag-dēn , wad-dēn , wattar-dēn ), ‘evil religion’ ( duš-dēn , duj-dēn ), ‘anti-religion’ ( jud-dēn ), ‘doctrine’ ( kēš ), and ‘non-religion’ (


Introduction 'Religion' as a Generic Term
In his essential contribution "Mani and the Crystallization of the Concept of 'Religion' in Third [1] Century Iran" in Mani at the Court of the Persian Kings, Jason BeDuhn (2015) analyzes the development of a generic term in Manichaeism for the concept that we now call 'religion.' In his discussion of the conditions necessary for this development, he asserts: "Religions emerged in antiquity when particular sets of religious practices no longer carried exclusive identification with such a native land, but belonged to a community that carried its own disembedded cultic identity" (2015,248). He aptly shows that Mani and Manichaeism fulfilled these conditions in Sasanian Iran for the first time: "Manichaeans did not think of themselves as a 'new race', but as adherents of a new religion comparable to other disembedded systems of cultic practice that through the course of time had crossed ethnic and cultural boundaries to a greater or lesser degree" (2015,270). The corresponding Iranian term which Mani and Manichaeans used in their text in Iranian languages is dēn, a word that designated different concepts in pre-Manichaean Zoroastrianism. The long conceptual history of the term started in the Old Avestan texts, around the twelfth century C.E., and led to the contemporary New Persian term dīn, for 'religion,' on the Iranian scientific metalanguage level 1 of Religious Studies. The formation of the generic concept of religion 2 seems to have depended on religious contacts, as has been shown to be the case in its development in other religious fields. 3 Thus, BeDuhn's study raises the following question: To what extent was the development of the generic term of religion in Manichaeism restricted to this religion and to what extent has the generated abstract term found its way into other religions which were in contact with it in Sasanian Iran? In this article, I examine the case of Zoroastrianism to answer the question whether Zoroastrian authors used the substantive dēn in Middle Persian texts as a generic term nearly equivalent to Manichaean 'religion. ' Despite years of discussions, we have to acknowledge that there is no general consensus on [2] a definition of 'religion.' The postcolonial study of religion even denies the existence of the notion 'religion' in pre-modern societies. 4 As a rough orientation for that which follows, we need a basic definition-like localization of 'religion' among socio-cultural entities. I assume that the following working definition, advanced by Volkhard Krech (2018, 10), founded on his assumptions for an analysis of religious evolution and against the deconstructive approach to religion, presents such a basis: […] religion is a societal communication system, which intrinsically emerges, re- [3] produces, and further develops. Based on specific sign processes within societal differentiation, it is responsible for ultimately coping with undetermined contingency by the means of the code transcendent/immanent.
To highlight the components of this definition that are more relevant to my discussion, I will [4] 1 In Religious Studies, following Linguistics, scholars distinguish two different but related levels: the level of religions under investigation is known as object language level. Scholars, however, attempt to use a more formal language to describe and analyze religions, which differs from the object language. This level is called metalanguage level. For this differentiation in Linguistics, see Allan (2006). 2 In the field of Metaphor Studies, linguists differentiate between conceptual metaphors, which are more or less general to all languages, and their instances, linguistic metaphors, i.e. the realization of the conceptual metaphors in a particular language. To demonstrate this difference, they render the conceptual metaphors in small capitals; see e. g. the leading reference Metaphors We Live by (Lakoff and Johnson 1980). Following this, I render the abstract concepts in this article in small capitals to clearly differentiate between them and the linguistic expressions of the concepts. 3 In the context of the Käte Hamburger Kolleg Dynamics in the History of Religions in Asia and Europe at the Center for Religious Studies, Ruhr-Universität Bochum, Reinhold Glei and Stefan Reichmuth (2012) studied the process of the development of a generic term of religio in Latin by concentrating on the Latin translations of the Koran. In doing so, they study the semantic development of the Arabic term dīn as well.
Another research from this context deals with material from the medieval period: Knut Stünkel (2013) addresses a similar question in his book Una sit religio: Religionsbegriffe und Begriffstopologien bei Cusanus, Llull und Maimonides. He points out that contact situations are an important factor for the genesis of the term 'religion. ' 4 This issue has been intensively discussed in the last decades, starting with Wilfred Cantwell Smith (1962) and continued, to mention only some of the prominent publications, by Asad (1993), McCutcheon (1997), J. Z. Smith (1998), Ford Campany (2003), Fitzgerald (2007), Nongbri (2013), Schalk (2013), and Barton and Boyarin (2016). adopt from the above that religion a) is a communication system, b) has a social dimension, c) helps coping with undetermined contingency, and d) uses for this the transcendent-immanent differentiation. In the following, I will start my engagement with the representation of the linguistic indications of the use of a noun as a generic term (next section). The proposed working definition on the metalanguage level will be contrasted by the pre-Sasanian semantics of the lexeme dēn-on the object language level. For this, representing the prehistory of the development of the generic concept of religion in Sasanian Iran, I will examine the semantics of Avestan daēnā-(section "Daēnā in Antique Zoroastrianism"). This section highlights the semantics of the term that developed into 'religion' in its pre-Sasanian Zoroastrian context and demonstrates with which components of 'religion' pre-Sasanian Zoroastrianism provides Manichaeism. Afterwards, I will retrieve the use of the linguistic means in the Middle Iranian Manichaean Corpora, Middle Persian, Parthian, and Sogdian, 5 for the use of the noun dēn as a generic term for religion (section "The Existence of the Term 'Religion' in Manichaeism"). I will divide the attestations in three categories: morphological and syntactical evidence, semantic evidence, and pragmatic evidence. This section aims to zoom in on the historical development of the term 'religion' in Manichaeism and will provide more evidence for the crystallization of the Manichaean concept of religion than BeDuhn aptly summarizes in one passage. Subsequently, I will examine the Zoroastrian Middle Persian corpus to demonstrate the rare use of the lexeme dēn as a generic term (section "The Term 'Religion' in Sasanian and post-Sasanian Zoroastrianism"). The corpus linguistic approach, in comparison to the philological study of religion or other historical approaches to religion, offers the considerable advantage of being able to quantitatively compare the use of the term dēn as a generic concept in both religions. The article, moreover, attempts to point out the textual departure point of the development of the term 'religion' in Sasanian Iran (section "The Departure Point of the Development of the Term religion in Sasanian Iran") and to show that the process was at its very beginning in Mani's lifetime. The centuries that followed, the article tries to demonstrate, witnessed further development of this concept (section "Conclusion"). A comparison between two instances is possible only when the scholar doing the compari-[5] son presupposes an abstract concept which would include both instances. In the absence of a more abstract concept of 'religion,' it is not possible to compare Manichaeism and Zoroastrianism. This operation, the comparison of different religions, would elude theologians of a particular religion if they did not have a generic concept of religion at their disposal. Similarly, the same kind of elusion appears in the case of scholars of religious studies: A scientific comparative study of religions can take place only with a scholarly agreement on a generic concept of religion. Therefore, such an abstract notion is necessary on the object language level as well as the metalanguage level. The study of the development of the generic concept of religion on the object language level, therefore, can contribute to an inclusive definition of religion on the metalanguage level. This article will also attempt to do this.

Linguistic Indications for Generic Terms
The use of a substantive as a generic term can be linguistically expressed in different ways. [6] Two such ways seem to be the most frequent ones: Firstly, a lexeme can be used in the singular so that it does not denote a concrete entity but designates a generic concept. Secondly, a lexeme can be used in the plural to designate a group of entities. Although the entities in this 5 In addition to these three languages, a Bactrian fragment in Manichaean script is known as well; see Sims-Williams (2009). group could be concrete entities, their grouping together implies the existence of a generic concept which groups the entities together. Therefore, the use of a lexeme in the plural could signify the implicit existence of a generic concept. Correspondingly, the examination of the plural declination of a substantive on the object language level that designates a similar concept to our scientific notion of 'religion' is reliable evidence for the existence of the generic term religion in the religious field that chronologically and geographically corresponds to the searched corpus. In terms of a research question which requires the examination of the whole occurrences of a lexeme in a corpus, focusing on formulations in the plural seems to be a reasonable delimitation to reassess the existence of a generic term. This method appears appropriate especially when the singular form of a lexeme is much more frequently attested than its plural form. This is actually the case for the Middle Persian lexeme dēn/δēn in Manichaean and Zoroastrian corpora.
Regarding the plural form, it is important to notice a Western Middle Iranian syntactical [7] rule. One should remember that, in Middle Persian and Parthian, a morphologically singular noun can be syntactically used with a plural verb as a plural subject. Therefore, lexical searches either in lexicons or in transcriptions can provide only morphologically plural words. Because of the absence of an adequate instrument, the search for plural forms should for the moment be restricted to morphologically plural forms, i. e. dēnān or dēnīhā. Without searching for morphological as well as syntactical plurals, a definitive statement about the uses of dēn in the plural is not possible. Such comprehensive search options are possible only by producing and investigating corpora that are morphologically annotated with parts of speech tags as well as a minimum of syntactical annotation. 6 These still remain a desideratum for Middle Iranian texts at the current stage. 7 Another form of expression which points to the use of a substantive as a generic term is [8] its formulation with a universal quantifier, such as 'every, ' 'each,' and 'all' in English. In the languages that this article deals with, the above quantifiers include MP/Pa. wisp and har(w) as well as Sog. wisp-. Another linguistic formulation that might indicate the generic use of the substantive 'religion' is its formulation in comparison. This could take place in a comparative phrase, such as "That religion is superior," or "Which religion is better?" 8 or, more generally, through ascribing a differentiating attribute to the substantive 'religion.' 9 The use for an annotated corpus is thus quite indispensable both when searching for such expressions and also when searching for the forms of syntactical plurals we pointed out above. Therefore, while this article does not claim being exhaustive in its enterprise, its author hopes to have found out some of the most significant instances for the use of dēn as a generic term. 6 In a corpus including parts of speech tags and syntactical annotation that determines the relation of subject and verb in a sentence, one can search for the syntactical plural by searching for the singular noun as subject of a plural verb. 7 I hope that we can provide it as a first step for Zoroastrian Middle Persian texts through the project Zoroastrian Middle Persian Digital Corpus and Dictionary (MPCD) in the near future. This is a long-term project which will make the texts of Middle Persian literature accessible to philologists and historians in digital ways through their manuscripts, transliterations, transcriptions, as well as their Middle Persian-English dictionary built upon the components of the corpus. As a digital corpus, it will not only provide the texts in electronic form but also include tagging and parsing information which provide the possibility for such queries. 8 See e.g. AWM 185 in paragraph 76 below. 9 As in the case of the syntactical plural, the lexeme dēn/δēn used with comparative (or superlative) adjectives can be easily searched for in an annotated corpus. The examples discussed in this article are not results of a systematic search.

'Religion' before 'Religion': Daēnā in Antique Zoroastrianism
The Middle Persian lexeme dēn-, which, according to BeDuhn (2015), designates religion, [9] derives from Av. daēnā-, from the root di 'to see.' 10 It is worth noting that daēnā-, as many other Avestan lexemes, is highly polysemous (see Ēmētān 1979, xxx-xxxi;Skjaervø 2011, 334f.). The exhaustive and precise study of Firouz-Thomas Lankarany, Daēnā im Avesta: eine semantische Untersuchung (1985) investigates its semantics. In this study, written in 1985, he assigns too easily the meaning religion to the term in the Young Avestan texts, however. 11 Nevertheless, his semantic analysis can be used without the consequential step of identifying the meaning 'religion' in my study. I will reorder the semantics advanced by Lankarany within another scheme, especially because I think that two meanings of the term, '(traditional) text' and '(textual) tradition', did not receive due attention in his study.

Daēnā as Vision: Old Avestan texts
The representation of two different aspects by the same term is significant for the semantics [10] of daēnā-in Old Avestan texts, as is the case for many other Old Avestan terms. 12 According to this, the term designates an aspect of both human beings as well as non-human entities. Lankarany (1985, 20) calls these two aspects the 'subjective' and the 'objective' notion. By subjective notion, he means an entity which is related to human beings themselves, belonging to their mental sphere, such as vision, view, knowledge, perception, insight, self, the state of the self, nature, character, conscience, spirit, sense, and attitude. The objective notion, in contrast, designates a (socio-cultural) entity outside the human being, as, for instance, law, teaching, formula, doctrine, and, last but not least, religion. In his illuminating article, Jean Kellens extends this semantics to a trilateral relationship with the act of viewing-active, passive and causative: daēnā-sees, is seen, and allows seeing (Kellens 1995, 51). The Old Avestan texts ascribe inter alia two different souls to the human being: uruuan-[11] and daēnā-. 13 The latter designates a soul component connected to viewing. According to this etymological meaning and the textual usages of the word, Old Avestan philologists translate daēnā-as 'vision-soul,' uruuan-as 'breath-soul.' 14 The breath-soul and the vision-soul are tightly connected with each other in the Old Avestan conception of communication with transcendent beings. According to Old Avestan texts, the aim of the ritual-as much as we can speak of an aim for ritual-seems to be the priests' encounter with transcendent beings. Ahura Madzā's House, which is called 'house of song (reception)' (garō dǝmāna-), constitutes 10 In the last century, philologists discussed the etymological relationship of Av. daēnā-to Vedic dhénā-.
the most important place of encounter. According to this concept, the priests send not only their songs and offerings to this house, but also their breath-souls from the ritual surface. The Old Avestan texts represent the breath-soul as responsible for the transport of hymns and sacrifice to the gods. In contrast to this, the function of the vision-soul is to lead the breath-soul to the destination of the ritual journey. The Old Avestan texts represent the ritual course (aduuan-) between priests and gods as consisting of more paths and turns, whereas the encounter with Ahura Mazdā takes place at the last turn of this course (apǝma-uruuaēsa-). At this point, the soul components arrive at the bridge of mason (cinuuaṇt-pǝrǝtu-), where they find out whether their ritual has reached the right gods or not. If the vision-or breath-soul strays from straight paths, then the breath-soul fails and goes to the 'house of deceit' (drujō dǝmāna-), and therefore no encounter with Ahura Mazdā will take place. If they stay on straight paths, they traverse the bridge of mason and reach Ahura Mazdā's dwelling. After traversing the bridge, the mission of the vision-soul ends and the breath-soul alone reaches Ahura Mazdā's house. 15 The last chapter of the Old Avestan texts, the Vahištōišti Gatha (Y. 53), envisages-as Jean [12] Kellens (1995, 38-54) convincingly shows-the meeting of breath-soul and vision-soul as an incest marriage with the consequence that daēnā-(active) lets uruuan-see (causative) the daēnā-'vision' (passive). 16 According to Alberto Cantera (2013, 115-35), this union is the moment of consultation (hǝm.paršti;-Y. 33.6) with Ahura Mazdā. He points out that the passive meaning of daēnā-, 'vision,' has a crucial role in its later semantic development to 'religion': "this vision includes in itself the contents of the consultation with god […] The consultation itself is part of the Vision obtained through the Vision" (Cantera 2013, 130). 17 Moreover, he points out that "the contents of the Vision obtained in the sacrifice constitute a corpus of texts" (2013, 130), namely some ritual texts, which the priests intercalate in their main liturgy. Through this, daēnā-appears as a synonym of dāta-'prescription,' t ̰kaēša-'teaching,' or srauuah-'text, hymn.' The double meaning of the lexeme daēnā-, the capacity for consulting Ahura Mazdā, as well as the contents of said consultation is, according to Cantera (2013, 135), the reason why the term took the meaning religion.
According to Jean Kellens andEric Pirart (1988-1991, II:252), a passive meaning 'vision' [13] for daēnā-might be attested in some Old Avestan passages. 18 In some, 19 daēnā-is declined in the plural, 20 but only in Y. 46.6 the designation of an objective meaning, 'vision,' cannot be ruled out. As Lankarany (1985, 62) demonstrates, the lexeme does not designate religion in Old Avestan texts. 21 He asserts a relationship between daēnā-and religiosity, however. 22

22
"In the Gāthās, daēnā-does not designate religion as an objective term because there is no relation between it and terminological terms. In contrast, it designates religion as practiced and lived by a person, as his or her 'religiosity.' The most significant form of this form of practiced religion is realized in the Gathas in In this regard, the passages that denote daēnā's relationship with the Zoroastrian moral triad good thought, good word, and good deed are notable (Lankarany 1985, 43). It is worth mentioning here that Old Avestan texts already attest the negative connotation [14] of the term tkaēša-with the meaning '(wrong) choice' or '(wrong) teaching' (Y. 49.2f.). 23 As we will see in the following, the term is of relevance for the development of the abstract notion of religion in the Iranian religious field.

Cistā as a (Ritual) Guide in the Young Avestan Texts
As Benveniste and Renou (1934, 56-64) convincingly show, Cistā is a goddess of path and [15] trip. The epithets ascribed to her in the first verse of the hymn indicate her function as a guide (Kellens 1994, 281f.): hupaθmainiiā-'provider of good paths,' huuaiβitacinā-'provider of good tours,' huuāiiauna-'provider of good ways,' as well as barat ̰.zaoθrā-'carrier of libation.' Her epithet razišta-'the straightest' correlates a transgressive function with herself: in her company, one can reach his/her destination without indirection. The etymological meaning of the name, 'considered' or 'observed,' should be connected to a function of hers as well: followers must consider their guide and therefore intently observe her. One should add to Cistā's general transgressive functions a special one: In collaboration with Vaiiu and daēnā-, she transmits priests' offerings to the gods. She seems to be responsible for leading people through mountains, woods, and lakes in their immanent daily life as well as for the transmission of offerings to transcendent gods.
The Young Avestan lexeme daēnā-designates a personified goddess whose functions are [16] derived from its passive meaning 'vision.' In the first verse of Yt. 16.1, Cistā is identified with this goddess. Moreover, Cistā's hymn, placed as the sixteenth hymn in the collection of Avestan hymns, has the title Dēn Yašt. The overlapping functions of both goddesses, Cistā and Daēnā, seem to have triggered their identification. 24 This process might have intensified the personification of the Old Avestan daēnā-. Anyhow, daēnā-designates a personified goddess in the Young Avestan corpus who is responsible for guidance to the transcendent world. Eric Pirart (2012, 127) aptly points to a frequently used metaphorical mapping in this regard in the history of religions: religion is a wa . Therefore, the active meaning of daēnā-'leading to vision,' i.e. the function of Daēnā as psychopomp, and its semantical relationship with spatial concepts in Old Avestan could be constitutive for its semantic development to religion.
phrases clearly point to the ritual semantic of the term. 25 The active and causative meanings, which were dominant semantics of daēnā-in the Old Avestan period, are present in the Young Avestan corpus as well. Here, daēnā-designates an anthropological component listed along with ahu-'being (?),' baoδah-'consciousness (?),' uruuan-'breath-soul,' and frauuaṣǐ-'(divine) soul.' 26 In Old Avestan texts, the vision-soul is contextualized in ritual communication with tran-[18] scendence and is presumably unconnected to both postmortem life and to an ethic value system. The components of Old Avestan ritual communication were transformed in the Young Avestan period into individual eschatology. According to Fritz Stolz (2000, 706f.), the eschatological dimension appears when such representations are linked to ethical values and projected into the future. I previously showed that these conditions seem to have been fulfilled in Zoroastrianism no earlier than the Young Avestan period (Rezania 2017a, 242-49).
A significant example of the eschatological reinterpretation of ritual transcendence can be [19] read in V. 19, in which the destiny of the breath-soul after death is represented. At dawn following the third night after death, the breath-soul is led to the bridge of mason. This soul and the consciousness of the deceased are asked about their contribution to the world during their material existence. Afterwards, Daēnā appears with her dogs. She throws the breath-soul of the deceitful one into the darkness and lets the breath-souls of the orderly man traverse the bridge of mason. The second Hādōxt Nask, a Young Avestan text that represents the destiny of the breath-soul after death, starts with Zarathustra's question from Ahura Mazdā, where the breath-soul of an orderly man will stay on the first night after his death. Ahura Mazdā answers that the breath-soul sits near his head. It enjoys as much pleasure as in its entire material life. After repeating the question for the second and third night and stating the same answer, the text depicts the events at dawn after the third night: it seems to the breath-soul that it passes through fragrant plants. Furthermore, the vision-soul of the dead person seems to advance (Hintze 2017) in the form of-to quote the Avestan passage HN 2.9-"a maiden, beautiful, bright, with white arms, strong, well-shaped, well grown, tall, with high (standing) breasts, with a body from song, noble, from a brilliant lineage, fifteen years old in look, in form much more beautiful than the most beautiful creatures." 27 Verse 11 of this text develops the relation of the OAv. daēnā-with the Zoroastrian ethic triad to their identification. The represented Daēnā 28 to the breath-soul is his/her own thought, word, and deed.
The semantic link of daēnā-to memorization (Lankarany 1985, 133, 140f.) attests its re-[20] lation to ritual as well as (ritual) text. Cantera considers the function of the ritual priest obtaining, after his consultation with Ahura Mazdā, the vision " 'to utter and memorize' (mar-) the Vision and to bring it to the sacrificial community in the material world" (Cantera 2013, 127f.), whereas the transmission (bar-) and the preservation (mar-) of vision are alluded to 25 For the meanings connect to ritual, also see Lankarany (1985, 156). 26 See Lankarany (1985, 162-64); for a discussion of Zoroastrian anthropology, see Gignoux (2001, 11-16). 27 For an exhaustive study of Zoroastrian daēnā-departing from this Avestan text, which considers many Zoroastrian and Manichaean Middle Persian texts as well as Arabic sources, see Widengren (1983). 28 The literary presentations of this notion have given rise to some iconographic representations in the Sasanian period. Gherardo Gnoli (1993) identified a female figure with a flower in her right hand on some Sasanian seals as Dēn. He points out that the flower represents the fragrance and perfume which the breath-soul of a righteous person will smell during his/her journey to the hereafter. Azarpay (1976) identifies an allegory to Dēn and 'house of song (reception) in an artefact.' Daēnā's iconography is known from a Sogdian painting from Dunhuang (Grenet and Guangda 1996) as well; for a reassessment and verification, see Hintze (2016). More recently, Yutaka Yoshida (2009) identified Daēnā in a Manichaean painting preserved in the Museum Yamato Bunkakan, Nara, Japan.

Summary
To sum up this section, the Avestan lexeme daēnā-presents the three semantic fields-active, [22] causative, and passive-of the act of viewing from the very beginning. Whereas in the Old Avestan texts the term occurs more as an anthropological component, the passive meanings of the term prevail in the Young Avestan corpus. In the latter, the semantic field of daēnā-intersects with the following fields: ritual, guiding in ritual and postmortem life (psychopomp), ritual or traditional text, tradition, law, conduct of life, as well as community. daēnā-cannot be reduced to one of these semantic fields, and that is true for religion as well: It is not only ritual, conduct of life, community, or even transcendence alone. It cannot be reduced to any of these fields, since its semantics intersects with all these fields. In regard to the proposed working definition of religion above, the function of daēnā-as guide is of great relevance. On the one hand, it binds the divine and the human world in ritual or in postmortem life, which evidences its reference to the transcendence-immanence distinction. On the other, its guide to the transcendent world attempts to cope with the contingency of failure in ritual or postmortem life. daēnā-is responsible for leading the breath-soul and sacrifices to the right gods and leading the breath-soul to the best existence after death. Its intersections with semantic fields such as text and community, as well as the presence of daēnā-in the ritual communication with gods, cover the part of the definition that considers religion a societal communication system. Cantera's analysis, moreover, discovers the semantic of daēnā-as a vision that the ritual priest obtained from Ahura Mazdā and must share with the community. Therefore, the use of daēnā-in the Avestan corpus lets us conclude that whereas the term does not designate religion, and such an abstraction cannot be attested in the Avestan corpus, a potential for the development of this concept already evolved in the Avestan period. The Bahuvrīhi compounds aγa.daēna-as attribute of Aži Dahāka (Az. 3) and duž.daēna-'the one with bad daēnā-' (Lankarany 1985, 122) evidence an abstract understanding of daēnā-, using 29 On this topic, see also Cantera (2012, 45-47) and Panaino (2015, 102-11).
it for good and bad alike. Moreover, daēnā-occurs as the object of the verb var-'to choose.' 30 As represented above, daēnā-can simply mean 'vision' in the often occurring Young Avestan phrases vaŋᵛhī-daēnā-māzdaiiasni-and daēnā-ahūiri-zaraθuštri-. Subsequently, we cannot attest the meaning religion even in these phrases.

The Existence of the Term 'Religion' in Manichaeism
For the purpose of searching the Manichaean Middle Persian and Parthian corpora, I will [23] primarily take advantage of the Dictionary of Manichaean Middle Persian and Parthian (Durkin-Meisterernst 2004), which covers the texts published until 2004. For the Manichaean Sogdian corpus, I will use two dictionaries, Gharib (1995) and Sims-Williams/Durkin-Meisterernst (2012), as the search instrument. For more recently published texts, I will consult the publications directly. The glossaries of these text editions list the attestations of the lemma dēn/δēn in the texts and more or less provide the meanings 'religion; church; religious community' for it. However, they specify these meanings as the potential meanings of the noun. This does not automatically mean that each of these meanings is attested in the corresponding text edition. 31 Therefore, I will check every text passage based on the provided working definition, whether it roughly features the abstract meaning religion or not. Moreover, it is sometimes the case that the substantive has been used in the text to designate Manichaeism as 'the religion.' This rendering is reflected in the translation as well; however, this does not mean that the author used the substantive in the generic meaning religion. Therefore, these glossaries are not of great advantage for the study of the semantic field of the substantive dēn/δēn, and a more careful reassessment of the passages will be necessary.

Syntactical Evidence dēn in the Plural
In Manichaean Middle Persian texts, I was able to identify three attestations of the morpho- [24] logical plural of the noun dēn. 32 The first passage comes from a famous fragment in which the author enumerates the superiorities of the Manichaean tradition 33 to its predecessors, Buddhism, Zoroastrianism, and Christianity.  (Lankarany 1985, 149-41). 31 For an exception, see Sundermann (1992a, 147). 32 See Durkin-Meisterernst (2004, 3:151, 2014. 33 I use the alternative terms 'religious tradition' and 'religious field' to designate a stage in the formation process of religion prior to the one that the material designates as 'religion.' 34 For more convenience, I render the quoted Middle Iranian texts in transcription. The transcriptions are mine, whereas the editions represent the texts only in transliteration.

35
M 5794 I V 20 (Boyce 1975b). A more recent edition of this text (Lieu 2006, 525), which reproduces the recto side of the text completely for the first time, leaves out the last phrase of the verso side, dēn ī man. 36 If not indicated differently, all translations are by the author.
As alreadly highlighted by BeDuhn (2015, 269), Mani speaks of older religions in the plural in [26] this passage: pēšēnagān dēnān. Moreover, he uses the same term dēn for the designation of his religion and of other religions, which was not a matter of course in late Antiquity. The second Middle Persian passage including the plural form of dēn is about the missionary activities of Manichaean disciples in the Roman Empire: (2) šud hēnd ō hrōm (3) dīd was hammōg pahikār (4) abāg dēnān. 37 [27] "They went to the Roman Empire (and) observed much doctrinal battle with religions." Here again we see a clear example of the use of the substantive dēn in the plural. The plural [28] declination of the word implies the existence of a generic concept religion, which is signified by the term dēn in Manichaean Middle Persian. The last attestation of the plural form of dēn in Manichaean Middle Persian known to me comes from a hymn in praise of god with the title 'Syriac Melody,' which remains altogether unclear because of its frequent lacunae: (6) [29] āfurīhēnd ud (7) yōǰdahrīhēnd pad (8) nām yōǰdahr (9) wispān dēnān (10) yōǰdahrān kē-t (11) nām […]. "And from the first and for ever they were joined to you. / They are being praised and they are being blessed and they are being hallowed in/with the sacred name of (?) all sacred religions, which [honour / deserve?] your name." 38 This passage represents an interesting example of the generic use of dēn in the plural, at- [30] tributed with an adjective in the plural (yōǰdahrān) as well as with the universal quantifier wisp, again in the plural.
After reviewing the attestations of dēn in the plural in Manichaean Middle Persian, I would [31] like to turn to the Manichaean Parthian corpus. This corpus comprises 12 attestations of dēn in the plural, 39 which I will discuss in the following. In the first example, religions other than Manichaeism are described as 'deceived.' They are treated as independent entities, however. To designate them, the author has used the term dēn in the plural: 40 (157) ud hawīn kē nē aδ (158) haw ham šud ud nē až paš dēnān wiδeftagān (159) [32] āhend. "They who neither went together with him nor after him are deceived religions." 41 The topic of 'deceiving religions' or of 'the deception of religions' occurs in two other [33] Manichaean Parthian passages: (1) lōg nāz ud iskēm āwaržōg ud šahr (2)  See Durkin-Meisterernst (2004, 3:150, 2014. 40 For the significance of the formulation in the plural instead of an alternative formulation in the singular, see paragraphs 71-74 below. 41 M 44 R7 (Colditz 1987, 301). 42 M 77 R 1-5 (Andreas and Henning 1934, 41).
"The pleasure of the world, the desire of the form and the things of the earth resemble a sweet food with which poison has been mixed. Keep your souls away from these nets! The beings who are deceived by religions are terrified. They will not find the way out […] and they will not recognize the wisdom." ō larz bid až hušk goxan ud dēnān wiδeftagīft.
[37] "and they give life … … the religions and all … … deeds" 44 Another Parthian fragment with dēn in the plural is a text about Manichaean missionaries in [38] the Roman Empire, as in the case of Middle Persian M 2 I R above. According to the text, Mani sent some missionaries, among them Addā, the bishop, one of the earliest disciples of Mani (Sundermann 1983), there from Weh-Ardašīr. 45 According to the text, he instituted many monasteries in the Roman Empire and wrote brilliant treatises. In describing his other activities in the Roman Empire, the author uses the substantive dēn in the plural twice to designate active religious traditions in the Empire. He moreover uses a synonym for the designation of 'religion,' ammōg 'teaching,' which he sets by using a plural verb in the plural as well. 46
Another example of such use of dēn can be found in Gabryab's oath; Gabryab was of the twelve [40] most distinguished disciples of Mani: (2) w(āxt) kū abestāwagān (3) (hem pad) ha(rw)[ī](n) dēn(ān) drōγ (a)mm(ōg) (4) [41] (u)d (paδ) ēw wāwa(r)īft paδ tō d(ēn) (5) hamwadām čē tō ay baγ ud (6) (a)nǰīwag čē (man) gyān. 48 […] said: 'I swear off the deceitful teaching of all religions, and believe in your religion as the only faith because thou art lord and savior of my soul' In this passage, one finds not only the noun dēn in the plural for the designation of 'deceitful' [42] religions, but also its similar use for designating the religions of others and also one's own religion. All religions, one's own as well as others', are designated in this passage by the same These missionary activities are dated between 244 and 262 B.C.E. (Schaeder 1934, 71). 46 This is a sample of the syntactical plural; see paragraph 7 above. 47 M 216c + M 1750 V 9-13 (Sundermann 1981, 26 49 (I say to you) righteously that his soul will lastly find the life higher than all dēnān who stay on deception, who lie against god, (and) deny his light … One can question whether dēn really refers to 'religion' in this passage. It seems that it offers [44] another meaning of the noun dēn: dēnān are not compared with a religion here but with the soul (gyān) of a-from Mani's point of view-righteous man. Presumably, the meaning of the noun dēn in this passage concurs with another meaning of it in Zoroastrian texts, namely 'vision soul,' as presented in paragraphs 11-13 above.
Another morphological plural form of the lexeme dēn in the Manichaean Parthian cor-[45] pus can be found in M 216b V. The passage is again about the missionary activities of a Manichaean apostle, or Mani himself, in the territory of a king named Waruzān / Waručān. 50 dēnān ammōg pad wxēbēh bazag andrenǰād. 51 [46] He defeated the teaching of the religions through their own sins.
In this passage, dēn in the plural is governed by the singular form of āmmōg 'teaching,' which [47] implies that religions, although containing the same single teaching, are perceived as distinct entities. The next example attests the plurality of religions in a higher degree: (2) āzārēd hō gyān kē (3) pad im ābēn (4) niguržēd pad čē dēnān (5)
The significance of this passage is its representation of different entities designated as dēn as [53] comparable to each other without further attribution. Especially the absence of attributes for such entities formulated in the plural suggests that a generic term dēn forms the basis of this formulation. Not only does the author conceive dēn as a generic concept, but he/she compares it to fruit. The metaphorical expression "the fruits are the religions" attests the conceptual metaphor religion is a fruit. 55 For this conceptual metaphor, the Manichaean author did not pick fruit by chance. This resides at the center of Manichaean redemption theory. As the concept fruit exists alongside the expression 'fruits,' so the author must have had a concept of religion besides the expression 'religions.' Therefore, it is permissible to conclude that, from the author's point of view, religion is as generic as fruit.
After the Parthian corpus, I will next investigate the Manichaean Sogdian one. In this corpus, [54] the noun δyn is attested four times in the plural (see Sims-Williams and Durkin-Meisterernst 2012, 76; Gharib 1995, 148;Sundermann 1992aSundermann , 1997. The fragment M5266 (recto, line 10) 56 contains the phrase δēnān 57 axšēδ 'the prince of the religions,' which designates one of the six gods or lords 58 of Manichaean Theogony, namely Jesus. Therefore, the plural form in this passage does not designate a group of different religions. This leaves us with three attestations of the 'religions' in Manichaean Sogdian: I will turn to Āzandnāmē, §45-50, in paragraph 95 below. The Sermon of Light-Nous, §94b, the Sogdian translation of the Parthian passage §94a mentioned above, reads as follows:
The Sogdian term δēnīk in this passage means adheres to a (foreign) religion with the plural [56] form δēnīkt. Theoretically, this plural substantive can designate 'the followers of a heretic religion' as well as 'the followers of heretic religions.' In light of the parallel Parthian passage, however, we might assume that it expresses the second meaning. Therefore, the passage attests a generic concept of religion as its Parthian origin.
The final evidence comes from the Manichaean Sermon of the Soul. The passage represents [57] the five gods as the fundamental basis of the world, without which the world cannot exist (GW, §112). This basis consists of only these five gods (βaγān), and there is no other god beside them (GW, §113). Consequently, the text compares the five light elements (mrδaspand) with military officers in a place threatened by an enemy. If the officers leave the place, the enemy will destroy it. Correspondingly, the world would be demolished without the five light elements. 60

rt(ms ZK) ʿBY ʾʾ(δβγ )[ZY ZK] /11/ [pn]cw βyʾn prw wyspw z-wrnʾkw ZY prw(h)
[58] 55 Cognitive semantics understands metaphor as a cognitive (in contrast to linguistic) process by which one semantic domain can be understood in terms of another. Conceptual metaphor theory identifies the mapping of two conceptual fields to each other as a conceptual metaphor (love is a journe ), which can be unfolded in different linguistic metaphors (Look how far we've come; We're at a crossroads; We've got to go our separate ways). For a classic work on conceptual metaphor theory, see Lakoff and Johnson (1980, examples here are quoted from p. 44). 56 For the text, see Henning (1948, 314). 57 δēnān is the older genitive plural form; see Gershevitch (1954, sec. 1230). 58 These are the Third Ambassador, the Living Spirit, the Mother of Life, The First Man, Jesus, and the Maiden of Light (Henning 1948, 314). 59 Sundermann (1992a, 61, 75); LN §94b. 60 See GW, §114-7; In §119, the five light elements presumably correspond to five Avestan Gathas (γʾδh) (Sundermann 1997, 87 and 141).
Interestingly, the passage sets religion in parallel to language: Just as a diversity of lan- [59] guages exists in the world, there exists a diversity of religions. Just as a generic concept of language exists for the author which allows him to write 'in all languages,' there exists a generic concept of religion which allows the Manichaean Sogdian monk to express 'in all religions.'

dēn with a Universal Quantifier or Demonstrative Adjective
In many Manichaean passages, the substantive dēn/δēn is used with a universal quantifier.
[60] As I pointed out in paragraph 8 above, such expressions might indicate a generic use of the term 'religion.' The use of a universal quantifier, however, might express just one complete entity or all entities classified as such. This difference can be seen in these two sentences, for example: 'She read the whole book' and 'He put all books on the shelves.' What we are interested in is the second expressional form, in which different entities are classified together and designated by a generic term which designates a generic concept, in our example book. The linguistic expression formulated with a universal quantifier differs in these two cases when the qualified substantive is countable in English as well as in Middle Persian, Parthian, and Sogdian. 62 As only the second expressional form mentioned above refers to a generic concept, and the substantive dēn/δēn is countable, the reader might think that the search for the plural forms of dēn/δēn might already have uncovered all these occurrences. This would be true if the plural form in these languages were restricted to the morphological plural. The existence of the syntactical plural, however, requires us to search for any marks of the plural form in cases in which we cannot consult a digital corpus. The syntactical plural form, i.e. the morphological singular noun with a plural verb, can [61] occur only in the rectus as the subject of the verb. Therefore, all phrases with a universal quantifier and substantive dēn/δēn in the oblique case in the singular designate just one religion and are not relevant for our study. Such examples are eżāfe-construction (ī hamāg dēn), 63 constructions with prepositions (abar/az/ō/pad hamāg dēn), 64 or dative complements. 65 Excluding these attestations leaves us with formulations in which dēn has been further specified with an attribute, primarily with yōǰdahr/yōždahr, to designate Manichaeism. 66 These phrases designate 'the whole Manichaean religion/community/church.' Therefore, we cannot find any attestation of a generic concept of religion formulated with universal quantifiers.
In the Manichaean Middle Iranian corpora, we find few constitutes with a demonstrative [62] 61 Sundermann (1997, 86f.); So 18248 II (GW §118). 62 Consider the singular form in the former case ('book') versus the plural form in the latter ('books'). 63 See e.g. M 291c R7f. (Leurini 2017, 78), Otani6156+/A/10 (Kudara, Sundermann, and Yoshida 1997), BBB 448 (Henning 1937, 31). adjective and dēn. The Parthian fragments M 434/B/ii/4 and M6650/V/7, in which the phrase ʾō im dēn 'to this religion' occurs, show that in the author's cognitive map of the religious field there is 'this religion' and 'that religion.' The evidence subsequently demonstrates that there are different socio-cultural entities which have been indifferently called 'religion' on the object language level. 67 These cannot be anything else but instances of a generic concept religion.

Semantic Evidence
Besides morphological and syntactical indications for the existence of a generic concept of re- [63] ligion in the Manichaean Middle Iranian corpora, discussed in paragraphs 24-30 above, there are also semantic indications for the existence of such a concept. One of them is comparison. When one compares two entities with each other, whether on the object language or metalanguage level, one presupposes the existence of an abstract conceptual class of which the two compared entities are two instances. If we do not consider the existence of such an abstract concept as a necessary presupposition for comparison, then we can inquire whether it emerges as a result of the act of comparison itself. Therefore, when we encounter a Manichaean comparison between 'religions,' we can assume that the Manichaeans used a generic concept of religion. The famous M 5794 I V 20 presents the most significant comparison of this type, with which I started my discussion in paragraph 25 above and to which I will return in paragraph 90 below. Such a comparison is expressed in the fragment M 738/R/6-8 as well: u-šān paywennē pad / wuzurg rāstīh az dēn ī / ērdar ō dēn ī abardar.
[64] With the great righteousness You bind them from the lower religion to the higher religion.
This text designates two entities as 'lower religion' and 'higher religion,' respectively. We will [65] leave aside here the question of what the Manichaean representation of these two entities is. The significant point is setting the two entities in relation. By the attribution as 'lower' and 'higher,' the author of the text compares two entities and demonstrates that she possesses an abstract concept of religion, of which one instance can be the lower one and the other the higher. 68 A diachronic comparison of religions can be seen in M539d/A/2, which includes the phrase dēn ī-t pēš[ēn?] … 'religion(s?) before you …' (Durkin-Meisterernst 2014, 252f.). 69 Through comparing different religious traditions in history, that is, previous traditions and the current Manichaeism, the author reveals that he considers them, as Manichaeism, instances of the generic concept of religion.

Pragmatic Evidence
Making religions comparable provides the possibility of their selection. Whether this poten-[66] tiality is realized or not, and under which circumstances this takes place, is not the topic of 67 Such an interpretation, however, cannot be extended to M20/II/R/3-4, S 7/V/i/3/ or S 9/V/ii/17, in which dēn has been specified with the demonstrative adjective ēn 'this,' because dēn has been particularized in these passages with an attribute, yōǰdahr or ardāyīh, as the designation of Manichaeism. The fragmentary passage Otani6178/A/3/does not present an attestation either because of the lacuna after the word dēn in the phrase, which does not allow us to decide whether dēn has been used in this passage with an attribute or not. 68 This argument does not hold if the author considers 'lower' and 'higher' as designating different stages of the same religion. 69 As the verb of the sentence lies in the lacuna, we cannot ascertain whether dēn is formulated in the singular or in the plural. the current discussion. Instead, significance lies in the fact that the possibility of selecting among religions lets us infer the existence of a concept of religion: Without a generic concept no comparison is possible, and without comparison of religions no selection among them. A Parthian hymn expresses this latter possibility: ud wxēbēh dāhwān kirbag […] pad wxēbēh dēn wižīdag ispurr karēh.
[67] "and you make your pious gift in your chosen religion full." 70 In this passage, as well as in the Sogdian fragment SO 14381 B5, 71 dēn has been qualified with [68] the verbal adjective wižīdag, with the meaning 'chosen,' from the infinitive wižīdan "to choose, select, prefer, discriminate" (Durkin-Meisterernst 2004, 3:337). This attribution shows that, in the religious field of Sasanian Iran, dēn could be 'chosen,' at least from the Manichaean point of view. Another Parthian passage which sets dēn in relationship to 'choice' reads as follows:

až haw dard kū trixsād bawēh awēšān až istaft lōg čē dušmenīn abar haw awištād
[69] pawāžēh ud āzād karēh kū wxēbēh dēn až aδar wižīnāh ud wxēbēh handām hawīn kē-šān čihrag āzād kird amwardāh. " […] from that pain where you are being oppressed(?) and you purify them and make them free from the hard world, on which enemies, stood; so that you choose your religion/religious community from below/downwards(?) and you gather your members/limbs, those whose form/nature you/they(?) freed" 72 As it is reflected in the translation, we cannot ascertain in this passage whether dēn has been [70] used in the meaning 'religion,' 'religious community,' or any other semantic fold of the pre-Sasanian notion of daēnā-. I tend to assume that dēn means 'religious community' here and the passage expresses the process of putting community together. If we, however, assume the presence of the meaning 'religion' here, we again have an attestation of the 'choice of religion.' Subsequently, we could infer that, according to the Manichaean author, one religion can be preferred to another. This assumption leads to the hypothesis that, according to the same author, one could choose from a religious field which included different religions. The presented evidence strengths BeDuhn's hypothesis, namely the existence of a generic concept of religion in the Sasanian religious field of Iran.

The Term 'Religion' in Sasanian and post-Sasanian Zoroastrianism
To scrutinize the development of a generic concept of religion in Sasanian Iran further, we [71] can concentrate on the use of the same term, dēn, in Zoroastrian Middle Persian texts. As BeDuhn (2015) hints to the third century C.E. as the departure point for this development, I will start with a Zoroastrian text from this period. Despite the sparse primary sources for Sasanian Zoroastrianism, we are lucky to have an inscription from a high Zoroastrian priest from the third century C.E.: Kerdīr, the Zoroastrian antagonist of Mani and the state priest of four Sasanian kings. 73 He applied a categorization of religions, known to us through more or less contemporary Christian sources: true religion (vera religio) and false religion(s) (falsae religiones). 74 For Kerdīr, there were only two categories of entities in the religious field: Zoroastrianism and the demonic tradition (in the singular). Significantly, Kerdīr reserves in all of his inscriptions-as Jason BeDuhn (2015, 264f.) aptly points out-the noun dēn for Zoroastrianism and applies another substantive, kēš 'teaching, doctrine,' to designate 'false religion,' the doctrine of Ahreman and the demons. 75 As I mentioned in paragraph 14 above, this latter term was loaded with negative connotations even in Old Avestan texts. However, it should be noted that Kerdīr used the noun kēš to designate Zoroastrianism as well: kēš ī dēwān 'doctrine of demons' as opposed to kēš ī yazadān 'doctrine of gods.' 76 As a result, we can conclude that from Kerdīr's viewpoint dēn designated only Zoroastrianism, whereas it was possible to denote true as well as false religions with the noun kēš. In accordance with this understanding of the religious field, Kerdīr contrasted 'good' (Zoroastrian) priests with heretics. 77 Illuminating for the construction of the religious field in third-century Iran is Kerdīr's enu-[72] meration of non-Zoroastrian doctrines. He listed seven different groups: Jews (yhwdy), Śramaṇas (šmny), Brāhmanas (blmny), Nazarenes or Nazoreans (nʾclʾy), Christians (klystydʾn), Baptists (mktky) and Manichaeans (zndyky). 78 Significantly, he did not designate them as 'false religions' in the plural, but as 'the doctrine of Ahreman and demons' in the singular (kēš ī Ahreman ud dēwān). Apparently, from Kerdīr's point of view, non-Zoroastrian religions are only different realizations or exemplifications of one and the same demonic doctrine. It should be mentioned that-as Jason BeDuhn (2015, 265) points out-Kerdīr described the actualizations of the demonic doctrine using non-ethnical categories, and in doing so he introduced a new categorical division to Sasanian society in the third century. He does not refer to, for example, 'Indians,' but to Śramaṇas and Brāhmanas. He reserved the noun dēn for Zoroastrianism, 79 however, and even used it without any further attribution for its designation: §17: […] ud was dēn ōšmurd gōnag gōnag ud anī-z kerdagān ī yazadān was abzūd ud [73] abardar būd […]. "[…] and the religion was much studied 80 in various ways, and also the rites of the gods were much increased and became more important […]." 81 It is worthwhile to compare this Zoroastrian use of the substantive dēn with the one in the [74] Manichaean text M 312 R 6, presented in paragraph 52 above. In the Manichaean text, dēn See §11 (KSM 14 = KKZ 9 = KNRm 29); for the identification of the listed religions, see de Menasce (1945, 206f.), Widengren (1965, 277f.), Bailey (1980), Sundermann (1987Sundermann ( , 56n109, 2001, De Blois (2002, 1-6), and Skjaervø (2012). 79 It is worth mentioning that Kerdīr used the lexeme dēn in the meaning 'psychopomp' as well; see KSM 29 / KNRm 57. 80 The verb used here is ōšmurdan 'to enumerate;' on the term dēn ōšmurdan, see Vevaina (2010). 81 MacKenzie (1989,55,59).
is used in the plural without any further attribution to denote (all) religions; in this Zoroastrian inscription, dēn is used similarly, without any adjective, but in the singular to signify Zoroastrianism. Similar to the Manichaean Middle Iranian texts explored above, a Zoroastrian Middle Per-[75] sian text, presumably from the late Sasanian period, could be taken as evidence for a comparison between religions. A passage in the Memorandum of Wuzurg-mihr asks the question: AWM 185f.: dēn kadār weh. ān kē yazdīh ī yazdān dēwīh ī dēwān kerbag mizd wināh [76] puhl aziš paydāgtar ud rāh ud ristag ī frārōntar kerbag pad-dādtar jast estēd. Which religion is better? The one from which the divinity of the gods (yazdān) and the demonic nature of the demons as well as reward for virtues and punishment for sins more immanently emerge, and from which the way and manner of more righteous virtue has occurred as more lawful.
The passage indicates how Sasanian Zoroastrianism dealt with the question of superiority of [77] one religion over others. This question attests the comparability of religions in late Sasanian Zoroastrianism and, as a result, the existence of a generic concept of religion, which, as we will see in the following, is not completely detached from the dualistic division of religions in true and false.
The Zoroastrian dualistic categorization of religions is not restricted to the beginnings of [78] the Sasanian period. The same perspective on religion can be observed in Pahlavi literature from early Islamic times. The most exhaustive Pahlavi book, the Dēnkard, which is entirely dedicated to inter-religious debates 82 in the context of Abbasid Islam, 83 regards religion in the same way. It opposes 'good' religion, i.e. Zoroastrianism-which is designated with the phrases 'the better/good religion' (dēn ī weh, weh-dēn, hu-dēn), 'the noble religion' or 'the religion of nobles' (ēr-dēn), 'the right religion' (rāst-dēn), and 'the religion of Ohrmazd' (ohrmazd dēn) to 'bad' religion, presented as 'bad/worse religion' (ag-dēn, wad-dēn, wattar-dēn), 'evil religion' (duš-dēn, duj-dēn), 'anti-religion' (jud-dēn), 'doctrine' (kēš), and 'non-religion' (adēn). My search for the morphological plural forms of dēn in the Dēnkard, as far as possible with the available tools, yielded only one certain occurrence. 84 Dk. IX, 38.7 ud ēn-iz kū ēd ī tō dēn zardušt pahlom ast az dēnān. ēd ī tō dēn zardušt [79] gēhān frāxwēnēd ī ahlāyīh ud bowandag-menišnīh. This (applies) as well that this religion of yours, o Zarathustra, is the most excellent among the religions. This religion of yours, which is righteousness and rightmindedness, flourishes in the world, o Zarathustra. This passage is probably the only example in all of late antique Zoroastrian literature in [80] which the author explicitly expresses the plurality of religions and does not categorize them 82 For the inter-religiosity of the Dēnkard and the revision of the encyclopedic view on it, see Rezania (2017b). 83 For the interaction of the authors of the Dēnkard with Abbasid's caliphate, see Rezania (forthcoming). 84 The noun dēn in the plural might appear in Dk. III, 225.16: u-š dēn ud waxšwar ā druwandān pad dastwar ī +dēnān abar-estišn. This chapter of the Dēnkard corresponds to the lacuna in manuscript B (the only older manuscript of the text) after page 192. Therefore, the chapter can be read only from younger manuscripts. From de Menasce's (1973, 237) translation of the corresponding phrase, it emerges that he reads the word as dēnān. Zaehner (1939, III/900) reads the phrase "ē druwandān pat dastaβar ī dēvān apar ēstišn;" see also different alternative suggestions by Fażīlat (1396, 1093n29). As the phrase dastwar dēw can be found in Dk. III, 288.1 as well, and as it fits better semantically in this phrase, I prefer the reading dēwān between the two homographs.
in a dualistic scheme. The passage presents semantic evidence for the existence of a generic concept of religion as well: Ādurbād ī Ēmēdān, the author of the ninth book of the Dēnkard, compares Zoroastrianism with other religions. We can therefore assert a significant discrimination between Kerdīr's viewpoint in his inscription from the third century and this Dēnkard passage from the ninth century. Whereas Zoroastrianism was not comparable with other religious traditions for Kerdīr, who designated them differently, as dēn vs. kēš, Ādurbād considers Zoroastrianism and religious traditions as instances of a generic concept of religion. 85 Extending the search for the morphological plural of dēn more or less to the whole Zoroas-[81] trian Middle Persian corpus results in only one other passage, from Dādestānīhā ī Dēnīg, 86 likewise a post-Sasanian Pahlavi book from the ninth century: ud ēk ān kē-š ahlomōγ-dēnīhā kāmist ō dād ī stūd, ēg pad frēftārīh ī wardēnīd abestāg [82] ud zand az xwēš wimand […]. 87 One is that person, who wished the heretic religions to be the praised law, afterwards turned the Avesta and Zand according to his own * limit(ation) (?) through deception.
Although Manuščihr, the head of the priest of Fars and Kerman, followed the Zoroastrian [83] model of the division between true and false religions, he spoke of 'false' religions in the plural and not in the singular like Kerdīr, for example. Despite subscribing to the distinction between true and false categories, Manuščihr accepts the plurality of religions. At the same time, he does not designate religions as heresies different from his own 'religion,' as Kerdīr did by calling them ahlamōγ. Instead, Manuščihr ascribes the term dēn to them, ahlomōγ-dēnīhā. However and significantly, the same author reduces the meaning of dēn to Zoroastrianism in a passage in his Epistles: 88 dēn ī dēn mazdēsn ārāyišn ud wirāyišn ud wābarīgānīh abāz rasād. 89 [84] May the religion, which is the Mazda-worshiping religion, become again prepared, arranged, and believable.
Manuščihr acknowledges only Zoroastrianism as a religion. Yet, according to the positions I [85] have advanced above, he uses the term dēn on two different semantic levels. The first one signifies religion on the conceptual level, whilst the second one, employing the attribute 'Mazda-worshiping,' designates an instance of that concept, namely Zoroastrianism. In order to express that 'Zoroastrianism is the only religion,' Manuščihr was compelled to advance a generic concept of religion. If one reads these two passages together, one could infer that, on one hand, Manuščihr believes in the plurality of heresies and accepts them as special instances of religion (ahlomōγ-dēn), and, on the other hand, acknowledges only Zoroastrianism as (true) religion. 85 It should be highlighted that Zoroastrianism was in agile contact with other religions in the Sasanian and early Islamic period. These exchanges were certainly formative for the (restricted) development of the abstract notion of religion in Zoroastrianism. 86 On this text, see Shaki (1993). 87 Dd. 71.9 König (2010, 360, 362). 88 For a short review of his biography and his Epistles, see Rezania (2020). 89 NM 1, 11.9 (Kanga 1988).

The Manichaean Designation of One's Own Religion and Other Religions
The Manichaeans referred to their religion mainly by the same linguistic means as the Zoroas- [86] trians. The latter designated their religion by modifying dēn with an adjective, such as 'good, ' 'better,' or 'right What is more, they do not differ from the Zoroastrian emic designations for Zoroastrianism.
To be more precise, late antique Manichaeans employed the same word, dēn, to designate both Zoroastrianism and Manichaeism, and modify both religions with further attributes. The following is one of the most significant passages which names Manichaeism and Zoroastrianism side by side:

nēw bād ud pirār ī pērōz gar(dāg)ān īg dēn māzdēs / frazendān ī wāxš yōǰdahr u-tān
[87] mād ī abēzag padiš wifrāyīhād. "… may good … be and the omen of the victor to those returning (from?) the Mazda-worshiping religion! / Children of the holy religion and may your pure mother be helped through him/it!" 95 Strikingly, the Manichaeans, in opposite to Zoroastrians, used the same term, dēn, for the des- [88] ignation of Zoroastrianism (dēn māzdēs and weh-dēn) in both text fragments above. In the latter passage, however, the author did not use the term dēn for the designation of Manichaeism, dēn yōǰdahr, but replaced dēn with wāxš to make a point of difference between the designation of Zoroastrianism and the label of Manichaeism. The choice of wāxš as an alternative to express religion is highly striking. In Middle Persian, waxš/wāxš means 'spirit' (MacKenzie 1971, 88;Durkin-Meisterernst 2004, 3:336). In Sogdian, the same phonetic form, wāxš, means 'word, spirit, thing' (Gharib 1995, 400). Taking into consideration substantives like Pa. wāxšβar / MP wāxšwar, we can assume that waxš/wāxš in Middle Persian possessed the 90 This term is very common in Manichaean texts; for some examples, see BBB 27, 126, 131f., 170f., 254, 454f., 469 (Henning 1937 See BBB e18-20 (Henning 1937, 48) and M8280/R/I/3-4 (Sundermann 1973, 76). 94 See M 543/R/3-5. 95 Durkin-Meisterernst (2014, 233); M 68a/I/R. meaning of 'word' as well. It seems that the author has chosen the term waxš/wāxš in this passage as an alternative for dēn because of its semantic parallelism to dēn. It is known that Manichaeism absorbed the Zoroastrian function of the goddess Dēn as a psychopomp. 96 This passage evidences the presence of the meaning 'traditional text' of the Zoroastrian daēnā-, presented in paragraph 20 above, in Manichaean texts as well. Therefore, waxš/wāxš must have served as a good substitute for the substantive dēn in this passage. Although this Manichaean phrase uses two different terms for itself and its rival religion, Zoroastrianism, its phrasing reveals a significant difference: Zoroastrianism designates itself as dēn and other religions as 'doctrine' (kēš) or 'false religion(s).' In doing so, it regards other religions as being inferior.
Manichaean writers also labeled their own religion as well as other religions with the term dēn. At the same time, they placed Manichaeism above other religions by naming it wāxš yōǰdahr. In both cases, these two religions are situated in a certain hierarchy and in a certain relation to other religions. There is a considerable difference between the ways in which Manichaeism and Zoroastrianism envisaged the above, however. Manichaeism accepted the presence of other entities in the religious field as 'religions,' thus acknowledging the plurality of religions in third-century Iran. Third-century Zoroastrianism discredited other entities and presented itself alone as religion, a position that could be accounted for, to some degree, up to the end of the first millennium C.E.

Making It Possible to Say 'Religions'
Let us now return to the Manichaean texts and revisit the Manichaean evidence of the noun [89] dēn in the plural. As we saw in paragraphs 24-59 above, this is attested in all three Middle Iranian Manichaean corpora, Middle Persian, Parthian, and Sogdian. However, its frequency differs in these three linguistic fields. The Dictionary of Manichaean Middle Persian and Parthian (Durkin-Meisterernst 2004, 3:150f.) lists ten occurrences for dēnān, the morphological plural of dēn, in Parthian and two occurrences in Middle Persian. Taking the texts published after this dictionary was published into account, these figures increase correspondingly to twelve and three, correspondingly, in Parthian and Middle Persian. The difference between these two linguistic fields becomes even greater if one considers the ratio of appearances of the plural form to all appearances of the noun dēn: considering the occurrences collected in the above dictionary, the plural form constitutes less than 2 percent of all attestations of the noun dēn in Middle Persian; in Parthian, the ratio is approximately 17 percent. 97 The case of the Manichaean Sogdian corpus is similar to Middle Persian: in this corpus, only three occurrences of the plural form of δēn are attested. It accounts for circa 3 percent of all occurrences of the noun δēn in Manichaean Sogdian texts (2012, 76). 98 These statistics reveal an imbalance in the occurrences of the term 'religion' in the plural in Parthian, on the one hand, and in Sogdian 96 Manichaeism borrowed the eschatological meeting of the righteous man with his deeds, appearing as a beautiful women, from Zoroastrianism (see Henning 1945, 476f.;Ibn al-Nadīm 1871-1872Colpe 1981;Shaked 1990, 28f.;Sundermann 1992bSundermann , 2008Sundermann , 160-62, 2009bReck 2003). 97 Durkin-Meisterernst (2004, 3:151f.) lists 47 appearances for Parthian <dyn> in addition to two uncertain readings, one for <dynw> and two for <dyyn>, which equals 50 attestations in the singular. In Middle Persian, <dyn> is attested 106 times, <dʿyn> two times, and <dynʾ>, <dyyn> and <tyn> once each, which yields 111 attestations in the singular. The relation of the plural form to all attestations is therefore 16.67 percent for Parthian and 1.77 percent for Middle Persian. 98 See also Gharib (1995, 149), who lists 19 occurrences for different writings of declination forms of <δyn> in the singular in the Manichaean script, 48 occurrences in Sogdian script (excluding one very doubtful reading), and one occurrence in the plural each in both scripts. The relation of the plural form to all attestations is therefore 2.98 percent. and Middle Persian, on the other. This imbalance seems greater if we consider the Middle Persian and Sogdian passages in which the plural form occurs more closely. I will begin with the first three passages of M 5794, 99 in which one of the three Middle Persian attestations of the noun dēn in the plural appears: (0) u-š […] passox dād kū ēn dēn īg man wizīd az abārīgān dēn ī pēšēnagān pad [90] dah xīr frāy ud wehdar ast.
(1) "Firstly, that the religion of the ancients was in one country and in one language; my religion will be, however, manifested in every country and in every language, and it will be taught in the far countries.
(2) "Secondly, that the former religion … as long as their pure leaders existed. As soon as their leaders raised [scil. passed away], their religion was confused.
[…] (4) "Fourthly, that my revelation of two principles, my living books, my wisdom and knowledge is further and better than the religion of the ancients." It is striking that in the first passage of this text, a plural adjective qualifies a singular noun: [92] abārīgān dēn. This grammatical irregularity already caught Boyce's attention (Boyce 1975b, 2:29n1). The same irregularity is the case for pēšēnagān dēn in the fourth passage, if pēšēnag is regarded as an adjective, 'earlier, former.' 101 The syntactical irregularity attested in this fragment seems to express a semantic regularity of the Iranian religious field in the third century: at that time, it was unusual to use the noun dēn in the plural. Mani-if the fragment stems from the Šābuhragān-aimed to formulate 'other religions'; however, he could not easily decline the noun in the plural. Although he used the plural form of dēn in the fifth passage of this fragment, he could not apply it constantly. As Mani compared his religion with other religions in this text, one can expect the plural declination of the noun dēn in three other phrases: 1. Instead of "the religion of the ancients was in one country and in one language," one [93] would expect 'the religions of the ancients were in one country and in one language;' 2. In the phrase "greater and better than the other religion(s) of the ancients," one would expect the formulation * pēšēnagān dēnān * instead of pēšēnagān dēn. 3. This expectation, moreover, applies to the phrase "the former religion" in the second passage. The substantive dēn has been used in the singular 102 here, although it semantically designates former religions.
If we contrast the passages in this fragment in which the noun dēn has been declined in the [94] 99 See BeDuhn (2015, 271f.), who discusses this passage and compares it with its Coptic version. 100 M 5794 I (Lieu 2006, 524-26). 101 Pēšēnagān, however, can be considered a substantive with the meaning 'the ancients.' 102 The singular enclitic pronoun š in andom dā-š refers to dēn, which rules out that dēn is a syntactical plural although the verb of the sentence is in lacuna. singular but semantically necessarily denotes a group of religions with the only passage in which the noun is attested in the plural, the fragment seems to indicate Mani's difficulty to formulate the word dēn in the plural in third-century Iran more than its plural use in one case. Nevertheless, the passage lets us conclude that, in his theory of religion, Mani acknowledged not only religious plurality but also a hierarchy of religions. One segment of the Sogdian text Āzandnāmē, in which the noun δēn appears in the plural, [95] resembles the discussed Middle Persian text M 5794 (Sundermann 1985, 23n39 (50) βwt skwnw The religion which I chose is in ten things greater and better than the other religions of the ancients. Firstly that the religion of the ancients was in one country and in one language; my religion will be, however, manifested in every country and in every language, and it will be taught in the far countries. The corresponding Sogdian passage clearly expresses, without any irregularity, exactly [96] what we expected and what was absent in the Middle Persian passage from the third century, that is, the declination of dēn in the plural. Acknowledging that the Middle Persian passage belonged to the Šābuhragān, we have to consider the large stretch of time that lay between the authorship of the two texts. It seems that these centuries gave Manichaeism enough possibilities to unfold a generic term of religion. The author of the Sogdian text, moreover, uses a metaphor to compare Manichaeism with other religions: He equated Manichaeism with a world ocean (samudr) and former religions with lakes or rivers (āp). Doing so, he indicates the existence of a generic concept of religion in his understanding of the religious field, comparable to the generic concept bod of water. Using the conceptual metaphor religion is a bod of water, the author uses his freedom to speak of religions in the plural.

Conclusion
As demonstrated in section 2, through the Zoroastrian term dēn, pre-Sasanian Zoroastrian- [97] ism provided Manichaeism with the semantics necessary for a concept akin to religion, as determined by our working definition. Otherwise, it would not have been possible for Mani to choose this term in his Iranian texts for the designation of his concept of religion. Nevertheless, it seems that Zoroastrianism presented a dualistic scheme of religions in the third century, whereas Mani developed this further into a hierarchy of religions. For Mani, there are no 'bad' or 'good' religions, but 'good' and 'better' ones. This is a significant change of perspective, about which BeDuhn (2015, 273) aptly says: "Mani's 'universalism', therefore, represents in part simply another polemical and competitive strategy different from that of Kartīr's particularism: Supersession in place of suppression, appropriation in place of rejection." However, third-century Zoroastrianism does not seem to have adapted his concept. From the presented Zoroastrian material, we can infer that Zoroastrianism constantly projected its dualistic worldview on the categorization of religions. Whereas this perspective on religion seems to have dominated in Zoroastrianism all throughout late Antiquity, later Zoroastrianism increasingly acknowledged the plurality of 'evil' religions. We can thus assert a historical development in Zoroastrianism: In the third century C.E., Kerdīr applied dēn only to Zoroastrianism, whereas in ninth-century Zoroastrianism, the term dēn was applied to heresies or, liberally, to all religions.
From the presented evidence, the infrequent occurrence of the noun dēn in the plural in [98] Middle Persian and the formulation of dēn in the singular where one can expect the plural form, I conclude that it was unusual, especially in the third century, to formulate the noun dēn in the plural in Middle Persian. The infrequent occurrence of δēn in the plural in Sogdian attests the same circumstances for this language. In contrast, the relatively high frequency of the plural form of dēn in Parthian shows the more frequent occurrence of the generic concept of religion in Parthian Manichaeism. At the moment, I cannot answer the question whether this linguistic contrast includes a geographical significance, indicating that Parthia in the north of the empire was the hot spot of this formative process in Sasanian Iran, or whether the development of the abstract notion of religion did not affect Persis strongly because Zoroastrianism ruled there much more than Manichaeism. I do not mean by this that the alteration of pre-Sasanian Zoroastrianism to Sasanian Zoroastrianism was independent of various Manichaean developments. Kerdīr's inscription demonstrates the acceptance of a sort of religious plurality in the Sasanian empire of the third century from the Zoroastrian point of view. Nevertheless, the reception of the Manichaean development of the term religion does not seem to have been enough strong to initiate a similar way of thinking about religion in a sustainable way in Zoroastrianism, as suggested by the infrequent occurrence of the noun dēn in the plural in the Pahlavi corpus. Despite this formative process, the semantics of the substantive dēn did not achieve a high degree of abstraction in the religious field of Iran, so that the noun would designate only the abstract concept of religion. The lexeme widely designates Manichaeism or Zoroastrianism even without any further attribute.
Manichaeism claimed universality, and as a universal religion, it was highly missionary.
[99] Moreover, it was a religion of the book, with a strong emphasis on written texts (Puech 1949, 61-68). These components led to Manichaean programmatic multilingualism, declared in Mani's agenda for his future religion, as we read, for example, in M 5794, quoted in paragraph 90 above. Manichaean multilingualism gave rise to Manichaean literature in a variety of languages. Of Middle Iranian languages alone, the Manichaeans utilized three different languages: Middle Persian, Parthian, and Sogdian, translating their texts between them. 104 As highlighted earlier (see n. 3 above), Knut Stünkel's study concludes that contact situations present a stimulating factor for the development of the generic concept of religion. It is not far-fetched if we consider the translation of a contact situation as intra-religious contact. Translation, the transfer of a text from one language to another, constitutes a contact of two linguistic fields. Religions other than the one that the translated text belongs to influence these linguistic fields differently. This could have happened, for example, because of different geographical distributions of adherents, a synchronic feature, or because of the different periods of presence of these religions in these linguistic fields, a diachronic feature. I can refer to the hypothetically supposed difference between the abstraction of dēn in Middle Persian and Parthian because of the stronger influence of Zoroastrianism in the south of the empire as an example. Therefore, if the source and target linguistic fields of the translation involve a difference in the influence of other religions, the act of translation is not a mere linguistic undertaking but presents an example of religious contact between two different religious fields. The strong attempts of Manichaeism in the translation of their sources from and into Iranian and non-Iranian languages might have provided more such contact situations, which in their turn strengthened the process of forming a generic concept of religion.
Regarding the formative exchanges between Zoroastrianism and Manichaeism, an unbal-[100] anced reciprocity seems to have existed: Zoroastrianism provided Manichaeism with the necessary semantics in a term for religion in the third century. Having received this abstraction, Manichaeism developed the Zoroastrian dualistic concept of religion further into a hierarchical one and the Zoroastrian noun dēn into a generic term for religion. Zoroastrianism seems to have absorbed the result of this Manichaean development: The meaning 'religion,' which was absent in the Avestan corpus for daēnā-, emerges in Middle Persian Zoroastrian texts. Zoroastrianism, however, stayed with its dualistic categorization of religions and was inclined to use the noun dēn in a generic way similar to Manichaeism. The dynamics of the processes in these two religious traditions apparently differ: The Manichaean tradition more easily integrated the Zoroastrian concept of religion and developed it further; the Zoroastrian one, by comparison, shows rather conservative behavior and did not abandon its dualistic scheme of religion. In conclusion, one might say that regarding the production of an abstract concept of religion, Zoroastrianism was more formative to Manichaeism than vice versa.