1. Introduction

Grammaticalization is concerned with the development of grammatical forms over time. As Kuteva et al. (2019: 3) define it, this process involves the emergence and evolution of grammatical structures and constructions across both temporal and spatial dimensions, while also accounting for their structural organization. In this paper, we argue that the verb āmadan ‘to come’, which serves as a lexical source, has undergone grammaticalization to become a multifunctional grammatical marker, in addition to its role as a light verb in certain compound constructions[1]. In Middle Persian (MP), āmadan predominantly retains its original meaning of ‘to come’. On occasion, it is preceded by preverbs[2] that modify its sense, as in andar āmadan ‘to come in’ and bē̆ āmadan ‘to come out’. During this period, passive constructions have been formed via suffixation (cf. Abolghāsemi, 2011: 49), and āmadan has played no part in the passive formation. Nevertheless, it has acquired the meaning of ‘become’, a semantic extension that has persisted into Early New Persian (ENP). According to Dehkhodā (1998: 205), āmadan in ENP is used—alongside its lexical meaning—as an auxiliary verb in passive constructions (see also Naghzguy-Kohan, 2011). Khānlari (1986: 375–380) notes that other verbs, such as šudan ‘to go’, gaštan/gardidan ‘to turn’, and raftan ‘to go’ also appear in passive constructions in ENP, with šudan being the most frequently employed and raftan the least[3]. In Contemporary New Persian (CNP), however, āmadan no longer conveys the sense of ‘become’ and does not serve as a passive auxiliary (Sadri Afshār, 2002: 44). Consequently, šudan has become the primary verb used in passive[4] constructions in this later period. In this study, we contend that the grammaticalization of āmadan toward a passive marker originates in its semantic extension to denote ‘change-of-state’.

Sadri Afshār (2002: 44) identifies several meanings for āmadan in CNP (CNP), two of which are particularly relevant to the present study. The first glossed as ‘intending to do something’ is argued here to encode inchoative aspect[5]. The second, meaning ‘supposition’—and inflected only for the 3^rd person singular and 1^st person plural—is proposed to express epistemic modality. In addition, we contend that in this period, the frequently used motion verb āmadan has also functioned as an avertive aspect marker and as a hortative modal marker[6]. Furthermore, it assumes a role as a discourse marker in CNP. Unlike the aforementioned grammatical functions, these discourse-related uses do not derive from the notion of ‘change-of-state’; rather, they represent metaphorically extended applications of the lexical meaning of the verb, i.e., ‘to come’, toward more abstract domains. The diachronic development of these diverse grammatical functions from a single motion verb may be accounted for by the universal pathway proposed by Bybee (1988: 374) for motion verbs, such as ‘go’ and ‘come’ outlined as follows:

Fig 1- Diachronic path for the development of future morphemes from 'come'/'go'

As the data suggest, the function of āmadan as an inchoative marker denoting intention precedes its use in expressing prediction and supposition, which in turn precede its imperative-like function as a hortative marker and its subsequent role as a discourse marker.

Section 2 outlines the grammaticalization framework adopted in this study and defines the relevant grammatical categories. It also reviews the most pertinent previous research on the topic. Section 3 describes the methodology, materials, and procedures employed in the study. The discussion (Section 5) follows the data analysis presented in Section 4. The final section, Section 6, is devoted to the conclusion.

2. Theoretical Framework

Given that the grammaticalized verb āmadan functions as a multifunctional grammatical marker in NP, this section briefly addresses several theoretical issues relevant to its analysis. The grammaticalization pathway of the verb under investigation is traced within the framework proposed by Narrog and Heine (2021).

According to Narrog and Heine (2021: 56), "Grammaticalization is the development from lexical to grammatical forms and, once the grammatical form has evolved, the development of further grammatical forms."

Narrog and Heine (2021: 56) identify four parameters for identifying and reconstructing grammaticalization, which apply in the following order: context extension, desemanticization, decategorialization, and erosion. For the first parameter, they introduce the context extension model (2021: 58), which comprises four stages and illustrates the semantic change of lexical items. In the initial stage, a given linguistic form serves as the input for grammaticalization and extends to a new context termed the bridging context stage (second stage). At this point, the target (new) meaning—which coexists with the old (source) meaning—becomes more readily inferable. In the third stage known as the switch context stage, the use of the linguistic form in a novel context foregrounds the target meaning to the extent that it becomes incompatible with the source meaning. Finally, at the conventionalized stage, this new preferred meaning no longer requires contextual support and speakers may no longer perceive its connection to the source meaning (Narrog and Heine, 2021: 58–61). This model is summarized in Table 1.

Table 1- Distribution of Source and Target Meanings in the Context Extension Model

Desemanticization—an immediate consequence of context extension from the second stage onward—refers to the loss of semantic properties[7] that a linguistic expression undergoes during grammaticalization. This process may result from various conceptual mechanisms, including subjectification, metonymy, and metaphorical transfer (Narrog & Heine 2021: 67–70). The second parameter, decategorialization, involves the loss of morphological and syntactic features in an item undergoing grammaticalization (Narrog & Heine 2021: 72). The final parameter, erosion, refers to the attenuation of phonetic substance of a linguistic expression; however, unlike the other two, erosion is not considered a requirement in the grammaticalization process (Narrog & Heine 2021: 78).

2.1. On Modality

Bybee et al. (1994: 176) define modality as the grammaticalization of the speaker’s subjective attitudes and opinions. They further argue that a comprehensive understanding of the range of modal meanings in any language requires an examination of the diachronic development of modal elements. In their framework, they distinguish four types of modality: epistemic, agent-oriented, speaker-oriented, and subordinating[8]. Agent-oriented modality refers to conditions imposed on the agent with respect to the completion of an action. These conditions may be internal (e.g., ability[9] and desire), external (e.g., obligation), or physical (e.g., necessity). Furthermore, agent-oriented modalities may appear in directives intended to elicit action. Since the speaker is involved in permitting or instigating the obligation, such uses are considered subjective (Bybee et al., 1994: 177). Speaker-oriented modalities[10], by contrast, allow the speaker to impose conditions directly on the addressee, rather than merely reporting their existence. This category encompasses directives, such as commands, demands, requests, and entreaties, as well as warnings, exhortations, recommendations, and permissions. Among the grammatical categories they identify for speaker-oriented modality (1994: 179), the hortative modality is particularly relevant to the present investigation. In this context, the hortative is defined as a construction, in which “the speaker is encouraging or inciting someone to action”. In the present study, the hortative construction is marked by the present subjunctive prefix *be-*, which, according to Mahootian (1997: 247), falls under the mood category in Persian. This construction serves various functions in subordinating clauses, including optative, intentional, conditional, potential, and hortative meanings. Mahootian (1997: 251) further notes that “the imperative form is used to express a hortatory meaning”, providing the example berim ‘Let’s go’.

Finally, epistemic modalities—which convey the speaker's degree of commitment to the truth of a proposition—commonly include possibility, probability, and inferred certainty (Bybee et al., 1994: 179). Of particular relevance to the present study is the expression of possibility, which, according to Bybee et al. (1994: 179), "indicates that the proposition may possibly be true".

2.2. On Aspect

Comrie (1976: 3) defines aspect as "different ways of viewing the internal temporal consistency of a situation". The perfective aspect presents a situation as a single indivisible whole, whereas the imperfective aspect distinguishes among the various constituent phases that make up a situation (Comrie, 1976: 16). For the purposes of this paper, two aspectual categories were of particular significance: the avertive aspect and the inchoative aspect. Given that the proximative aspect is closely related to the avertive, it is useful to define it here in order to clarify the similarities and differences between the two. According to Narrog and Heine (2021: 340), the proximative aspect denotes "a category of verbal aspect denoting 'be about to', 'be on the verge of doing'". In contrast, the avertive aspect—in their terminology (2021: 341)—refers to "a category of verbal aspect denoting 'almost, nearly', that is, an action or event that was on the verge of taking place but did not take place". Finally, the inchoative aspect[11], which serves a crucial function in the use of āmadan, marks "the beginning of a state or event" (Narrog & Heine, 2021: 336). As noted by Naghzguy-Kohan (2011: 649), the motion verb āmadan, when followed by a verb marked with a subjunctive marker, yields an inchoative reading, thereby indicating the inchoative aspect in CNP.

2.3. On Discourse Markers (DMs)

According to Narrog and Heine (2021: 303), Discourse Markers (DMs) are defined as "(a) invariable expressions, which are (b) syntactically independent from their environment, (c) typically set off prosodically from the rest of the utterance, and (d) whose function is meta-textual, relating to the situation of discourse—that is, to the organization of texts, speaker–hearer interaction, and/or the attitudes of the speaker". Although DMs are considered part of grammar, Narrog and Heine (2021: 306–311) observe that they exhibit certain features that challenge a straightforward reduction of their emergence to grammaticalization (see also Narrog & Heine, 2021). After examining three hypotheses on the development of DMs—namely grammaticalization, lexicalization, and cooptation—they propose the following cline (Narrog & Heine, 2021: 315–318):

Fig 2- Hypothesized Development of Discourse Markers (Heine, 2013: 1238–1239)

Narrog and Heine (2021: 316–317) define cooptation[12] as a discourse strategy, through which parentheticals are developed, and they argue that DMs exhibit much the same features as these parenthetical units. When a coopted unit is used recurrently, it can mark the onset of a grammaticalization process and subsequently give rise to DMs, a trajectory that aligns with the established parameters of grammaticalization. Although cooptation involves only a brief phase in the overall development of DMs, the authors contend that it can effectively address the challenges that grammaticalization faces in accounting for the emergence of DMs—challenges that accompany the entire history of DMs, from their initial rise to their current usage (Narrog & Heine, 2021: 315–318).

3. Methodology

The data selected for this study span three historical periods and are drawn from a range of texts covering diverse subjects. Regional variation has also been taken into consideration in the selection process. For NP, the chosen texts originate from various centuries. The CNP data, likewise selected from a variety of texts by different authors and on different subjects, are intended to represent everyday usage of the Persian language.

The MP data are sourced from the MPCD project[13] and include the following texts, along with their respective manuscripts:

• Great Bundahišn(GBd) according to manuscript TD1
• Mādayān ī yošt ī friyān(MYF) according to manuscript K20
• The second and third letters of Nāmagīhā ī manuščihr(NM2, NM3) according to manuscript K35
• Zand ī wahman yasn(ZWY) according to manuscript K43a

The ENP data are drawn from the following texts, spanning the period from 913 to 1299 CE. Approximately 15,000 words have been examined per text as detailed below:

• 4^th century AH (913–1009 CE):

Rag-Shenāsi [Angiology] (Ebn-e Sinā, 2004), Tarjome-ye tafsir-e Tabari [Translation of Tabari's Commentary] (Olamaˁ-e Māvarāˁ al-nahr, 1977), Tārix-e Balˁami [Balˁami's History] (Balˁami, 1974), and Al-abniye an haqāyeq al-adviye [Foundation of the Realities of Drugs] (Heravi, 1967)

• 5^th century AH (1010–1105 CE):

Qābus-nāme [The Book of Qābus] (Kaykawus ebn-e Eskandar, 1956), Siyāsat-nāme [The Book of Politics] (Khāje Nezām al-molk, 1965), Safar-nāme [Travelogue] (Qobādiyāni Maruzi, 1956), and Dibā-ye Khosravāni [The Royal Brocade] (Beyhaqi, 1996)

• 6^th century AH (1106–1202 CE):

Kimiyā-ye saˁādat [The Alchemy of Happiness] (Tusi, 2001), Kalila va Demna [Kalila and Dimna] (Monshi, 2009), Koliyāt-e čāhār maqāle [The Compendium of the Four Discourses] (Nezami Aruzi Samarqandi, 1948), and Asrār al-tohid [The Secrets of Divine Unity] (Mihani, 2023)

• 7^th century AH (1203–1299 CE):

Bargozide-ye Axlāq-e Nāseri [Selections of Nāseri's Ethics] (Tusi, 1994), Tazkerat al-owliyaˁ [Biographies of the Saints] (Attār-e Neyshāburi, 1967), Fih-e mā fih [There is in It What is in It] (Molavi, 2023), and Golestān-e Saˁdi [The Rose Garden of Saˁdi] (Saˁdi Shirāzi, 1997)

The CNP data are drawn from the following plays and novels, which are representative of colloquial and spoken language:

• Nun va qalam [By the Pen] (Āl-e Ahmad, 2005)
• Zan-e ziyādi [A Superfluous Woman] (Āl-e Ahmad, 1952)
• Zende be gur [Buried Alive] (Hedāyat, 2019)
• Haft Film-nāme [Seven Screenplays] (Farhādi, 2013)

3.1. Some Considerations on the Procedure of the Study

Given the considerable size of the investigated data, not all attested cases are included in the analysis. The most salient instances are discussed in Section 4. Examination of the data reveals that, in some cases, the verb is preceded by pre-verbs—namely adverbs, nouns, and adjectives—while in others, it appears in idiomatic constructions that yield new, modified lexical meanings. Since the notion of 'change of state' constitutes the primary semantic shift relevant to the grammaticalization of āmadan, the present study focuses exclusively on its meaning of 'become'. Other meanings of the verb, as well as its functions in compound constructions, are not examined here.

4. Data Analysis

In this section, the development of āmadan as a grammaticalized form is traced across three historical periods: Middle Persian (MP) in Section 4.1, Early New Persian (ENP) in Section 4.2, and Contemporary New Persian (CNP) in Section 4.3. Although grammaticalized uses of the verb are attested in New Persian, the onset of its semantic change can be traced back to MP; accordingly, illustrative examples from this period are also provided. The examples have been organized according to their respective stages within the context extension model for MP, ENP, and CNP, respectively.

4.1. Middle Persian

The Middle Persian (MP) data are examined to identify instances of grammaticalization during this period. Both verb forms āmadan and madan[14] are included in the analysis. Since the examples from this period are drawn from the same texts and the same linguistic stage, the term 'stage' here refers to coexisting abstract phases in the developmental process rather than to distinct chronological periods. Example 1 illustrates the lexical meaning of āmadan 'to come' and corresponds to the first stage of the context extension model, namely the initial stage. Example 2 represents the bridging stage—the second stage—in which both the source and target meanings are inferable although the source meaning remains prominent. Finally, Example 3 exemplifies the meaning 'become', which conveys the notion of 'change of state'. At this stage, this new meaning has become the highlighted and conventional interpretation of the verb; nevertheless, it still requires contextual support to be fully realized.

• MYF66 (K20.036r.8-168)

• GBd19 (TD1.092r.3-5)

• GBd98 (TD1.074r.13-1664)

There is a discernible correlation between the degree of development within the context extension model and both the concreteness of the agent and the physicality of the source/goal phrases. When āmadan/madan is used in its lexical meaning, the agent typically belongs to the higher levels of the animacy hierarchy[15] and the motion verb is most frequently accompanied by concrete source and goal phrases. In contrast, when the verb expresses 'change of state', the agent tends to be abstract and source/goal phrases are generally absent.

In total, the verb āmadan/madan is attested in 233 sentences across the data examined in this study. Of these, 219 instances reflect its lexical meaning, 4 cases correspond to the second stage, and 10 instances are classified as belonging to the third stage of the context extension model. As the data indicate, the final stage of the model is not attested in MP; rather, the process is completed in subsequent historical periods, leading to desemanticization and decategorialization.

4.2. Early New Persian

The verb form of madan is not attested in Early New Persian (ENP); āmadan is the only form that appears in the data. All attested cases have been classified according to the stages of the context extension model. In this period, āmadan undergoes both desemanticization and decategorialization and the grammaticalized form functions as a passive marker. Example 4 illustrates the first stage of the context extension model, in which only the lexical meaning is available. Examples 5 and 6 correspond to the second and third stages, respectively, where the source meaning and the target meaning are highlighted in turn. Finally, Example 7 represents the last stage of the model—the conventionalization stage—at which the relationship between the source meaning and the target meaning is no longer readily perceivable.

• Tārix-e Balˁami (1974: 5)

• Tarjome-ye tafsir-e Tabari (1977: 46)

• Tarjome-ye tafsir-e Tabari (1977: 24)

• Kalila va demna (2009: 43)

As observed in the MP period, abstractness increases progressively toward the final stage in ENP, while the use of physical source and goal phrases correspondingly decreases. At the second stage, the agent is characterized as 'affected' and is marked by the object marker rā, thereby functioning syntactically like an object phrase. As illustrated in Example 5, the agent—anger—is abstrac and its metaphorical motion exerts an effect on Eblis, who undergoes the state of 'being angry'—that is, a 'change of state' interpretable as 'become angry'. Thus, in addition to the primary lexical reading, the new abstract reading is closely tied to the affectedness of the agent in the data. At the third stage, the agent undergoes a 'change of state', a meaning that becomes conventionalized by the final stage. At this point, the target meaning constitutes the only inferable sense of the verb, which subsequently contributes to the formation of passive constructions. Consequently, āmadan undergoes desemanticization within the verb category in ENP and, as it comes to function as an auxiliary in the passive construction[16], it also undergoes decategorialization.

In total, the verb āmadan is attested in 1,356 sentences in the ENP data. Of these, 1,063 instances reflect its lexical meaning. In 5 cases, the verb belongs to the second stage of the context extension model, where it indicates the 'affectedness' of the agent. In 156 instances, āmadan conveys the meaning 'become', expressing the notion of 'change of state' and thus representing the third stage of the model. In 117 cases, āmadan is assigned to the final stage, functioning as a passive auxiliary. Additionally, in 15 instances, the verb participates in a specific inchoative construction that denotes 'the beginning of an action'. This construction consists of two fixed slots and one variable slot and may be literally paraphrased as 'enter an action'. The preposition dar/be 'in', which occupies one of the fixed slots, together with the action (the variable slot), forms the goal phrase, toward which the act of āmadan is directed. Consequently, the inchoative reading does not arise solely from āmadan itself, but rather from the construction as a whole. Given that this inchoative meaning follows a systematic pattern, it can be represented within the framework of Construction Morphology (Booij, 2010) as shown in Schema 1. Examples 8 and 9 illustrate this inchoative meaning through the following sub-schemas:

1. [[be/dar]Prep [action]N āmadan]VP ↔ ['to begin doing sth']
   [dar [šarāb xordan] āmadan] 'to begin drinking wine'
   [be [soxan] āmadan] 'to begin speaking'

• Dibā-ye Khosravāni (1996:76)

• Tarjome-ye tafsir-e Tabari (1977: 44)

The following diagram provides a comprehensive picture of the development of āmadan in each period of ENP.

Fig 3- Distribution of different functions of the motion verb āmadan in ENP

4.3. Contemporary New Persian

Examination of the CNP data reveals that, in this period, āmadan has been predominantly used in its lexical meaning and no longer functions as a passive marker[18]. Instead, the grammaticalized verb āmadan in CNP assumes aspectual and modal functions. Examples 9 and 10 illustrate the first stage of the context extension model, namely the lexical usage of āmadan. In total, the verb āmadan appears in 1,536 sentences in the CNP data, of which 1,477 instances reflect its lexical meaning. In contrast to MP and ENP, āmadan no longer conveys the notion of 'change of state' in this period; this function has been largely taken over by other verbs, most notably šudan (see discussion below). Accordingly, no instances corresponding to the second or third stages of the context extension model are attested in the CNP data. Instead, this period attests to the verb function in inchoative constructions (Examples 11 and 12), occurring in 22 cases in specific contexts, as well as its role as an avertive marker in 9 instances (Examples 13 and 14). Both functions fall within the category of aspect. The aspectual function as an inchoative marker represents a continuation of Schema 1 with the sole difference that the goal phrase is marked exclusively by be 'to[19]' in CNP. The avertive function of āmadan is attested in certain contexts, in which an action that has been 'about to be performed' is subsequently canceled, with this cancellation signaled by tā 'as soon as' and ke 'when'. Thus, the avertive reading of the verb is only inferable in specific constructions, which may account for the relatively low frequency of this function. As illustrated in Examples 13 and 14, āmadan in the subordinate clause conveys the meaning of 'being about to', an action that is subsequently canceled by an event occurring in the main clause. In addition to its aspectual functions, the verb also operates within the domain of modality. It serves as a hortative modal marker[20] (Examples 15 and 16) in the second person singular and plural attested in 8 cases and expresses epistemic modality in 11 instances. As shown in Examples 17 and 18, āmadan is inflected only for the first person plural followed by the clitic -and to indicate epistemic modality in the subordinate clause; in 8 cases, the verb is fronted in the sentence to convey possibility. The main verb in the main clause is not restricted in terms of inflection and appears in past tense forms (simple past, past perfect, or past perfect subjunctive) or the present subjunctive. In 4 cases, āmadan appears in the present tense, expressing a negative prediction (Example 19). Finally, Examples 20 and 21 demonstrate that āmadan can occur both as an interpolated and as a peripheral lexical item within the utterance, functioning as a discourse marker that indicates objection or emphasis in 8 cases. In Example 20, the speech participants are engaged in an argument; the speaker, who has not anticipated such a response from the interlocutor, uses biyā to register objection. In Example 21, the speaker employs the verb to place emphasis on the poor decision made, while simultaneously expressing dissatisfaction with it.

• Nun va qalam (2005: 169)

• Haft Film-nāme (2013: 173)

• Haft Film-nāme (2013:239)

(13) Nun va qalam (2005:115)

(14) Zan-e ziyādi (1952:11)

(15) Nun va qalam (2005:12)

(16) Haft Film-nāme (2013:6)

(17) Nun va qalam (2005:44)

(18) Nun va qalam (2005:92)

(19) Haft Film-nāme (2013:63)

‘Let’s assume (that) nobody calls (lit. called) to ask for two old man and woman workers.’

(20) Haft Film-nāme (2013: 192)

(21) Haft Film-nāme (2013: 43)

(22) Zende be gur (2019: 30)

5. Discussion

Āmadan is one of the most frequent Persian motion verbs and has undergone grammaticalization, emerging as a multifunctional verb in Contemporary New Persian (CNP) as demonstrated in Section 4. This verb has followed two distinct developmental paths. Its metaphorical extension aligned with the trajectory illustrated in Figure 1 originates from its lexical meaning 'to come'. Its function as a passive marker, however, derives from its 'become' meaning—a later lexical development. It is worth noting that āmadan is not the only verb in New Persian that conveys the notion of 'change of state' and subsequently develops into a grammatical marker. We examined other verbs expressing this notion from MP to NP, drawing on data from the MPCD project. The results indicated that, out of 582 attested instances of budan 'to be'—which in MP means both 'to be' and 'become'—375 cases reflected the former meaning, while 207 instances conveyed the notion of 'change of state'. The 'become' meaning of budan, although also attested in ENP, did not persist into CNP. Another relevant verb is šudan 'to go'. In the MP texts examined in this study, out of 118 attested cases, 3 instances of šudan indicated 'change of state', a usage that continued into both ENP and CNP. In ENP, the frequency of the 'become' meaning of āmadan decreased over time and was gradually replaced primarily by šudan, which itself became grammaticalized and functioned as a passive marker in CNP. Furthermore, gaštan and gardidan, both of which expressed 'change of state' in ENP, continued to be used in this sense in CNP; according to Naghzguy-Kohan (2011: 649), their use in passive constructions persists in very formal contexts. Thus, it could be concluded that 'change of state' was a notion largely shared by the verbs employed in Persian passive constructions. The question that remained to be addressed was why, in Persian, older verbs were successively replaced by newer ones in the passive construction budan → āmadan → šudan / gaštan / gardidan—while some continued to coexist, seemingly in contradiction with the principle of economy. This issue warrants further investigation. A comprehensive diachronic study is therefore needed to examine the coexistence of different verbs in representing the passive construction, taking into account various factors, such as regional distribution, frequency of use, and their diachronic development in expressing different grammatical functions—factors that may ultimately lead one verb to relinquish its passive-marking role while another expands its functional scope in this construction.

6. Conclusion

This paper argued that the highly frequent motion verb āmadan, in addition to its lexical meaning, serves as a lexical source for various grammatical functions and functions as a multifunctional grammatical marker in New Persian. In Middle Persian, it acquired a new, albeit infrequent, 'become' meaning alongside its original sense 'to come'—a development that subsequently gave rise to the passive construction in Early New Persian, a later historical period. It should be noted that during this period, the verb šudan 'to go' was also used in the passive construction. The passive function of āmadan did not persist into Contemporary New Persian; instead, šudan 'to go,' gaštan/gardidan 'to turn', and in some cases raftan 'to go' were employed in the passive construction. As a lexical motion verb, āmadan was used in physical contexts with an animate agent and was typically accompanied by concrete source and goal phrases. When grammaticalized as a passive marker, however, the animacy of the agent decreased along the animacy hierarchy, physical source and goal phrases were absent and the verb no longer denoted motion. In CNP, the verb was predominantly used in its lexical meaning 'to come' and no longer expressed 'change of state'. The grammaticalized āmadan functioned instead as an aspect marker (inchoative and avertive), a modal marker (hortative and epistemic), and a discourse marker in this period. As illustrated in Figure 1, the developmental trajectory begins with the lexical meaning of the verb as a movement verb. Its function as an inchoative marker—which implicitly denotes intention—precedes its uses in prediction and supposition (epistemic modality), while the imperative form functioning as a hortative marker and its role as a discourse marker represent later developments. Throughout the grammaticalization process, abstraction increased through metaphorical extension. Thus, even when the agent remained animate, the verb conveyed abstract notions, such as prediction and possibility, rather than its lexical motion meaning, while physical source and goal phrases were no longer present. The only construction that included a goal phrase was the inchoative construction in both ENP and CNP, which featured a non-physical goal. The following table gives a quick overview of different functions of āmadan in MP, ENP, and CNP.

Table 2- An overview of different functions of āmadan in MP, ENP, and CNP

Abbreviations

[1] The function as the light verb was not the focus of this study (for more information on this function see Naghzguy-Kohan (2011: 649) , Karimi-Dustān and Eshāghi (2019:171), and Davari and Naghzguy-Kohan (2025)).

[2] For more information on MP pre-verbs, see Skjærvø (2009: 214).

[3]This verb is still used in the passive construction in some parts of Khorāsān and is regarded as one of the dialectal characteristics of certain regions in ENP (Dehkhodā, 1998: 379).

[4] For more information on the passive marking function of šudan, see Karimi Doostan and Shahbazi (2025: 15-19).

[5] Davari and Naghzguy-Kohan (2017: 335) also refer to the inchoative aspect marking function of this verb.

[6] According to De Hann (2005: 32-45), modality is formally expressed across languages by means of modal auxiliary verbs, mood, modal affixes, lexical means, including modal adverbs and adjectives, modal tags, modal particles, and modal case.

[7] The entire meaning of the lexical unit does not necessarily disappear and the degree of the desemanticization correlates with the degree of context extension advancement. The more semantic properties a form loses, the more contexts it becomes compatible with (Narrog and Heine, 2021: 69).

[8] For more information on subordinating moods, see Bybee et al. (1994:180).

[9] Root possibility is a very important type of agent-oriented modality, which is related to ability and takes external factors into account as well.

[10] For a discussion on these grammatical terms, see Bybee et al. (1994: 179)

[11] Also referred to as ‘ingressive’ and ‘inceptive’ with some slight differences in definition. For a comparison, see Bybee et al. (1994: 318), Narrog and Heine (2021: 331), and Kuteva et al. (2019: 29).

[12] In this process, a chunk of sentence grammar or any other text piece is developed to be used on the meta-contextual level of discourse processing and is either interpolated in an utterance or placed at its periphery and is used as a syntactic standalone (Narrog and Heine, 2021: 316).

[13] https://www.mpcorpus.org/

[14] According to Bartholomae (1904: 300), the preverb ā in Middle Persian indicates direction toward the speaker. The difference between the two forms is in the existence of this preverb (mostly the difference is invisible as the Aramaeogram YATWN could be read either way), which, according to Abolghāsemi (2011: 183), is no more productive in New Persian.

[15] Extended animacy hierarchy: first/second person pronouns < third person pronoun < proper names < human common noun < nonhuman animate common noun < inanimate common noun (Dixon, 1979: 85)

[16] From a syntactic point of view, passivization results in the change of valency and the number of arguments decreases. That is, the agent is demoted and the patient is promoted in the passive clause. For more information on change of argument structure with change of valency, see Song (2018:372).

[17] Note that just as with English, in Persian, “turn” has turned to an auxiliary verb beside its lexical usage.

[18] For more information on the passive construction in CNP see section 5.

[19] The interesting point about the goal phrase is that except for one case, that is, be hayajān āmadan ‘become excited’ (Āl-e Ahmad, 2005:115), all of the goal phrases relate to the speech event.

[20] According to Kuteva et al. (2019: 92, 93, 257), verbs like ‘go’, ‘come’, and ‘leave’ give rise to hortative constructions in some languages. In this process, ‘certain verbs assume an interpersonal function in specific contexts involving imperatives and related interpersonal functions’ (Kuteva et al., 2019:257).

[21] This is a title used to refer to people, who visit the holy shrine of Emām Rezā in Mashhad.