Tu quoque?! On the second person pronoun tusya (tus̱a) and the second person verbal ending -tu (-du) in Niya Prakrit

2.1 The pronominal system of Niya Prakrit

Table 1 summarizes Burrow’s (1937: 32–33) analysis of the pronominal system of Niya Prakrit. The third person pronouns are not included, because Burrow (1937: 33–35) treats them as demonstrative pronouns. An analysis of these third person pronouns does in any case not contribute anything essential to the argumentation of this paper. The reader should take the following points into account. As in the rest of this paper, I use the term “direct case” for the merger of the nominative and the accusative case. Etymologically, these pronouns are all derived from the OIA nominatives. Infrequent variant spellings of the default forms are put in between brackets. Similar to what has happened in other MIA languages, the genitive and dative case have merged in Niya Prakrit. As far as the pronouns are concerned, both etymological datives and etymological genitives are found. While synchronically speaking they belong together, these are still separated from one another in the table, the old datives being listed in the first row and the old genitives in the second row.

Various aspects of the Niya Prakrit pronouns are in need of further investigation. It remains, for instance, to be studied whether to(ṃ)mi is a mere variant of tava (Burrow 1937: 96), or whether this form should be analysed differently. Another topic for further research is the synchronic distinction between the old datives 1sg mahi; 2sg tahi (tehi/dahi); 1pl amahu/asmahu (amaho); 2pl tumahu/tusmahu (tumaho) and the old genitives 1sg mama; 2sg tava/to(ṃ)mi; 1pl asmag̱a/asmakaṃ; 2pl tusmag̱a/tusmakaṃ. As the etymological datives are more frequent than the etymological genitives,^[6] it appears that the former are the productive forms, while the latter appear to be Sanskritic archaisms, possibly sociolinguistically marked for register. This impression needs, however, further research.

The point I want to focus on here is that Burrow’s system surprisingly lacks a dir.pl for the second person pronoun. In other words, why does Niya Prakrit lack a second person functional counterpart of first person veya(ṃ) (vayaṃ) ‘we.dir.pl’? While it is in theory possible that such a form has simply not yet turned up in our corpus, it is equally possible that the dir.pl of the second person pronoun is hidden in plain sight. In fact, when looking at Table 1, it catches the eye that Burrow assumes tusya (tus̱a) to be an additional form of the gen/dat.sg of the second person. Yet, this tusya (tus̱a) lacks then a counterpart in the first person, where there is only mahi, corresponding to tahi (tehi/dahi), and mama, corresponding to tava/to(ṃ)mi. It, therefore, seems worthwhile to investigate whether this tusya (tus̱a) could not be the missing dir.pl of the second person. In Section 2.2, I argue this is indeed the case.

2.2 tusya (tus̱a) is dir.pl

In this subsection, I provide evidence that tusya (tus̱a) should be classified as a dir.pl rather than as a gen.sg. In Table 2, I list all the securely attested examples of tusya (tus̱a), thereby indicating how many people are addressed and how tusya (tus̱a) is used syntactically. As briefly discussed in Section 1, Niya Prakrit second person pronouns (and verbal forms) are mainly used in letters and decrees. On the basis of the formulaic introduction to these types of texts, one can infer to how many people the letter is addressed and, by implication, thus also whether a pronoun as tusya (tus̱a) is used to refer to a plurality of persons or to only one person. Besides, the syntactic usage of tusya (tus̱a) will help us to properly define the case form of this pronoun.

On the basis of Table 2, one can make the following observations. First of all, out of the 19 documents that unmistakably contain a form of tusya (tus̱a), 15 mention more than one person in the opening to the letter (∼79%). Counting per token, 20 out of the 23 instances of tusya (tus̱a) are found in a document that is addressed to more than one person (∼87%). These statistics very much suggest that tusya (tus̱a) should be regarded a plural pronoun. Moreover, in the few cases tusya (tus̱a) occurs in a letter which addresses only one person (i.e. in CKD 272; 562; 639; 794), the pronoun is consistently used as the subject of a verb ending in -tha or -tu, which, as argued in Section 3.3, both appear to be used for 2pl. As regards the syntax of tusya (tus̱a), 14 out of 23 tokens of tusya (tus̱a) are used in subject function. This observation is a further confirmation that tusya (tus̱a) is the functional counterpart of 1pl veya(ṃ), vayaṃ ‘we’ which was missing in Burrow’s interpretation of the pronominal system. The fact that tusya (tus̱a) is also found in other grammatical functions can be explained on the basis of the general case confusion in Niya Prakrit (see Burrow 1937: 22–30, but also Jamison 2000). In short, the Niya Prakrit personal pronouns of the first and second person should rather be understood as in Table 3, whereby I indicate the main difference with Burrow’s analysis in bold face.

2.3 A new etymology for tusya/tus̱a

Now that it is clear that tusya (tus̱a) should be analysed as a dir.pl, a new etymology for this pronoun presents itself. Based on his interpretation of tusya (tus̱a) as a gen.sg, Burrow (1937: 32–33) segmented tusya (tus̱a) as the pronominal stem tu- and the nominal gen.sg ending -sya. While this etymology would be plausible if tusya (tus̱a) were indeed gen.sg, it is much less so when tusya (tus̱a) is in fact dir.pl, because the development of a 2sg to a 2pl pronoun is typologically very rare (Helmbrecht 2015: 184–188). In addition, the other direct case forms, i.e. ahu (ahaṃ/ahum), tuo (tuvo/tu) and veya(ṃ) (vayaṃ), are all etymological nominatives. Therefore, it is more likely that tusya (tus̱a) should be derived from OIA yūyam ‘you.nom.pl’ with analogical replacement of yū- by the oblique stem tus- seen in Niya tusmahu ‘you.gen/dat.pl’ and tusmag̱a/tusmakaṃ ‘you.gen/dat.pl’.^[7] The loss of the final -m in *tusyam (← yūyam) is paralleled by veya < vayam ‘we.nom.pl’ and e.g. the dir.sg ending of the a-stems, i.e. -a < -am (see Burrow 1937: 22).

3.1 The verbal system of Niya Prakrit

As is the case with other MIA languages, Niya Prakrit has simplified its verbal system as compared to OIA. According to the communis opinio (see Burrow 1937: 43–56), Niya Prakrit has eight moods (indicative; imperative; causative; optative; infinitive; absolutive; gerundive; participle), three tenses (present; past; future), three persons (1st; 2nd; 3rd) and two numbers (sg; pl). Traces of the OIA middle voice are rare and likely artificial archaisms; the passive, with suffix -ya-, is also used only occasionally. There are only two productive present stem formations, one with a stem in -a and one with a stem in -e (< -aya-). Taking the verb denati ‘to give’ as an example, Table 4 gives an overview of how Burrow seems to understand the present conjugation in Niya Prakrit (not all the forms of denati given are actually attested in the texts). The reader should note that Burrow assumes that the ending -tu (-du) alternates with the 2sg ending -si (-s̱i).

The same endings as for the present tense are used (1) for the future tense, where the suffix -iśa- (-iṣya-) is added to the present stem, (2) for the causative mood, which is characterised by the suffix -ave- < -āpaya-, and (3) for the optative mood (only used for present tense), which uses the suffix -eya-. In the past tense, the stem of the verb derives from the OIA verbal adjective in -ta (see Barchi and Peschl forthcoming; Burrow 1937: 50–53), to which enclitic forms of ‘to be’ are added. Again using the verb denati ‘to give’ as an example, the (partially reconstructed) paradigm of the past tense as it is understood by Burrow is presented in Table 5. Here again, Burrow assumes that the ending -tu (-du) is another 2sg ending, next to -si (-s̱i).

In what follows, I will re-examine Burrow’s suggestion that the verbal ending -tu (-du) is an alternative to the ending -si (-s̱i), coming to the conclusion that the ending -tu (-du) is, pace Burrow, not a 2sg ending, but is in fact replacing the 2pl ending -tha. The quantitative comparison between -tu (-du), -si (-s̱i) and -tha undertaken in the next subsection (Section 3.2) will yield some observations that already point in this direction. Further arguments are later on given in Section 3.3.

3.2 TAM of the endings -si (-s̱i); -tu (-du); -tha

In the present subsection, the “Tense, Aspect & Mood (TAM)” of the Niya second person verbal endings, i.e. -si (-s̱i), -tu (-du) and -tha, are subjected to a quantitative analysis, in order to understand how these endings fit within the Niya Prakrit verbal morphology. Because of the focus on morphology, “tense” and “mood” will be our main concern, while the syntactic usage of the endings, here loosely captured under the term “aspect”, will only be briefly touched upon.^[8]

Given that the inherited 2pl ending -tha does not require much comment, I will discuss this ending first. -tha is found only very infrequently (only 6 types/8 tokens),^[9] but, what is more, -tha is also only used in past tense forms. While not previously observed, these statistics are suggestive of a low productivity of the ending -tha. One, moreover, wonders how 2pl was then expressed in present and future, as the ending -tha is not found in these tenses.

As the basis for a comparison between the endings -si (-s̱i) and -tu (-du), Tables 6 and 7 give an overview of how often these endings are used in the present, past and future tense and in the optative mood.

The most intriguing conclusion one can draw from Tables 6 and 7 is that the ending tu (-du) is much less frequently used in past tense forms than the ending -si (-s̱i). This difference can be most clearly noticed if one looks at the token frequency, as there are only 7 instances of a past tense ending in -tu (-du) (∼4%) as against 50 instances of a past tense in -si (-s̱i) (∼43%). This rarity of past tense forms in -tu (-du) seems to indicate that past forms in -tu (-du) are less firmly anchored within the overall verbal system of Niya Prakrit. One is moreover inclined to think that this infrequency of the ending -tu (-du) in past tense forms is connected to the fact that the ending -tha is exclusively found in the past tense. In view of this, it would seem quite possible that -tu (-du) has taken over the place of -tha in present and future tense and is, by the time of our evidence, also gradually spreading towards the past tense. I will build further on this hypothesis in Section 3.3.

Tables 6 and 7 allow for a few other observations. Though slightly less important, these seem to be further confirmation that the endings -si (-s̱i) and -tu (-du) are, pace Burrow, not mere variants of one another. Future tense forms are, for instance, more frequent with the ending -tu (-du) (type frequency 37%; token frequency 59%) than with -si (-s̱i) (type frequency 18%; token frequency 15%). One should note, however, that, as far as token frequency is concerned, the data are slightly skewed because of the high frequency (74 tokens) of the verb form paribujiśatu (bujiśatu/paribhujiśatu/paribujiśadu) ‘you will understand’, which occurs in a formulaic sentence one encounters time and again in the Niya documents. Additionally, it appears that the ending -tu (-du) is never used for optatives, while there are a few instances (4 types; 4 tokens) of optatives ending in -si (-s̱i).^[10]

On top of these statistic differences between -tu (-du) and -si (-s̱i), there is one more difference in their syntactic usage that should not go unremarked, i.e. the ending -tu (-du) can be used to express commands in the present tense, which is not possible with -si (-s̱i). In his Niya Prakrit grammar (1937: 45), Burrow only acknowledges hotu (hodu/hutu/hoṃtu/bhavetu) ‘it should be’ (cf. also Balbir 1990: 26) and davyatu ‘it should be given’ (CKD 399) as old imperative forms.^[11] Yet, as Burrow seems to have recognized by the time of his translation of the Niya documents (1940), there are a few more present tense forms ending in -tu (-du) that are synchronically still used to express a command in the Niya documents. As will be discussed in somewhat more detail in Section 3.4, this synchronic fact suggests a historical connection with the OIA third person imperative ending in -tu. What follows is a list of the present forms in -tu (-du) that seem to express commands in the presently available Niya Prakrit corpus: CKD 31 + 764 upajivatu ‘one should live in dependence of’; CKD 134 choretu ‘you should abandon’; CKD 177 [karitu] ‘you should make’; CKD 295 pruchitu ‘you should ask’; CKD 338 prasavetu ‘you should produce’; CKD 373 kareṃtu ‘you should make’; CKD 399 vacitu ‘you should read’; CKD 475 denatu ‘you should give’; CKD 794 janatu ‘you should know’. A brief look at the negations used with present forms in -tu (-du) confirms that this ending -tu (-du) can be used in two distinct functions in the present tense, i.e. for declarative statements and for commands.^[12] Verbs ending in -tu (-du) use the negation na ‘not’ in declarative statements,^[13] while ma iṃ ci ‘not (at all)’ < OIA mā kiṃ cit is the default negation when present forms in -tu (-du) express a command or, more precisely in this case, a prohibition.^[14] Needless to say, this distinction is in line with OIA (Delbrück 1888: 541–546; Speijer 1886: 315–320).

In short, the syntactic usage of verbs containing the endings -si (-s̱i) and -tu (-du) and especially the frequency of these endings in specific tenses makes it clear that one cannot simply consider them variants of one another. It is, for instance, essential to take into account that the ending -tu (-du) is only infrequently used in past tense forms and that this same ending can be used in present tense forms to express commands.

3.3 -tu (-du) is 2pl

As one of the conclusions of the preceding subsection is that the 2pl ending -tha is only used for past tense forms, should there not be another 2pl ending for present and future tense? As -tu (-du) is not at all frequent in past tense forms, but more so in present and future tense, it seems worth investigating whether -tu (-du) is not rather the 2pl ending that is lacking for present and future (and which may then also be gradually taking ground in the past tense). The following two pieces of evidence seem in favour of this hypothesis. First, it will be pointed out that both the ending -tu (-du) and -tha correlate with the personal pronoun tusya (tus̱a) ‘you.dir.pl’, while the ending -si (-s̱i) is combined with tuo (tuvo/tu) ‘you.dir.sg’. Second, there is a general tendency for verbs ending in -si (-s̱i) to be attested in letters and decrees that are addressed to only one person, whereas verbs ending in -tu (-du) and -tha are by comparison found more often in documents that are sent to more than one person.^[15] These two arguments will be taken up in turn.

Regarding the first argument, it may first of all be noted that the ending -tu (-du) is less often combined with an overt personal pronoun in subject function than the other second person endings. Compare the statistics in Table 8.

Related to this, verbs ending in -tu (-du) are never combined with both a personal pronoun in subject function and a nominal subject standing in apposition with the pronoun. For the verbal ending -si (-s̱i), on the other hand, there are five examples of this construction,^[16] one of which (example sentence (1)) is found in CKD 144, where the king speaks to the cozbo-official Soṃjaka.

Second and still more important, when verbs ending in -tu (-du) are combined with an overt personal pronoun in subject function, this is, with only one exception, the pronoun tusya (tus̱a) ‘you.dir.pl’ (cf. Section 2.2). In this respect, -tu (-du) aligns with 2pl -tha, which also twice has tusya (tus̱a) as its subject, but not with 2sg -si (-s̱i), which is, notwithstanding one exception, by default found together with tuo (tuvo/tu) ‘you.dir.sg’. See Table 9 for the details. These observations are a strong indication that -tu (-du) should be regarded as 2pl and not as 2sg, as has been assumed before.

Regarding the second argument that -tu (-du) is a 2pl ending, I have first counted in how many documents a particular second person ending is attested and I have then investigated whether these documents specify, either in the introduction or in the delivery instructions, to how many people this document was addressed. Having so defined my corpus, I then calculate the percentage of documents that are sent to only one person and those that are sent to more than one person.

If we apply the methodology described above to the verbal ending -tha, we obtain the following results. There are seven documents that contain verbs ending in -tha, one of which is too fragmentarily preserved to be of any use (CKD 705), which means there are six documents to be considered. Exactly half of those (3 docs. ∼ 50%) are addressed to only one person (CKD 213; 272; 275), while the other half (3 docs. ∼ 50%) is addressed to more than one person (CKD 162; 375; 470). While the latter examples do not necessitate further comment, the former ones can be accounted for if we assume that, precisely as in Sanskrit (cf. Speijer 1886: 195), the ending -tha could also be used as a polite form when referring to one person only. Needless to say, using a 2pl verb form as a polite form is typologically frequent (cf. e.g. French or Modern Greek).^[17]

When doing the same calculations for the 65 documents that contain at least one verb form in -si (-s̱i),^[18] one can observe the following. 18 documents lack or do not preserve delivery instructions and are thus not helpful for this research.^[19] Hence, we are left with a corpus of 47 documents, out of which 38 (∼81%)^[20] address only one person, as is expected for the 2sg ending -si (-s̱i). The remaining 9 documents (∼19%)^[21] address more than one individual.

In documents addressing more than one person, it is often the case that a 2sg verb form is only intended for one person specifically.^[22] I will briefly discuss two examples of this phenomenon. A first instance comes from CKD 140, a letter which is sent to the ṣoṭhaṃgha ‘accountant’ Lýipeya and three of his sisters. Being a ṣoṭhaṃgha, Lýipeya is concerned with the finances of the kingdom (cf. Burrow 1937: 127–128), while his sisters are not. As a result of this, when Kupṣiṃta, the author of CKD 140, makes a reference to the reckoning of taxes, it is most likely that he only addresses Lýipeya and not his sisters and, indeed, in those cases a 2sg verb form is used.^[23] See example sentence (2), which occurs thrice (with minor variants) in CKD 140.

Another telling example is found in CKD 552. This document is addressed to the same ṣoṭhaṃgha ‘accountant’ Lýipeya, this time together with his two scribes (divira) Sodaya and Lýimsu. In example sentence (3), again having to do with the administration of the Shanshan kingdom, the usage of a 2sg personal pronoun tuo ‘you’, combined with a nominal subject bhaṭarag̱a ‘master’ standing in apposition with the pronoun (cf. supra for this construction), makes it clear that again only Lýipeya is invoked here.

For the ending -tu (-du), there is a total of 133 documents which contain at least one verb form with this ending,^[25] out of which 6 are either not well enough preserved to be used for this investigation or simply do not indicate to whom they are addressed (CKD 125; 177; 240; 423; 741; 761). This leaves us with 127 documents. 75 of those (∼59%)^[26] have only one individual as their addressee, while 52 of them (∼41%)^[27] have more than one person addressed. In other words, it is still the case that verb forms in -tu (-du) are found more often in letters and commands sent to only one person, but the difference with verbs ending in -si (-s̱i) is still notable, as for this ending 81% were addressed to only one person. The percentages we have for -tu (-du) are also not too far from the fifty-fifty ratio which we observed for the ending -tha, commonly believed to be plural.^[28] As a result, I take these numbers as another indication that the ending -tu (-du) is marked for 2pl, but that it can at the same time also be used as a polite form when referring to one person. As noted above, this same usage as a polite form seems to be there for the ending -tha as well.

While the figures discussed above and also summarized in Table 10 already suggest that the ending -tu (-du) is marked for plural, there are a few additional arguments to back this up. First of all, it is instructive to have a look at sentences which contain both a verb form in -tu (-du) and a form in -si (-s̱i). Such an example is, for instance, found in CKD 165, which is a personal letter sent by the ogu-official Kirtiśarma to both the cozbo-official Kranaya and the same ṣoṭhaṃgha Lýipeya which we encountered above. Example sentence (4) comes from this document and contains both the verb forms vis̱ajiṣyatu ‘you will send’ and vyoṣiśas̱i ‘you will reimburse’. If the ending -tu (-du) is indeed marked for plural, the assumption would be that vis̱ajiṣyatu would refer to both Kranaya and Lýipeya and 2sg vyoṣiśas̱i to only one of them. This suspicion can be easily confirmed, because vyoṣiśas̱i is combined not only with the 2sg pronoun tuo, but also by a nominal apposition, ṣoṭhaṃgha Lýipeya, which is akin to a vocative. As such, while vis̱ajiṣyatu can be taken to refer to more than one person, vyoṣiśas̱i only refers to one person in particular.^[29]

A second point concerns the usage of second person verbs in the greeting formulae which open many letters in the Niya Prakrit corpus. One such formula translates into English as ‘I/we am/are happy that you are healthy’ and both arog̱es̱i and arog̱etu ‘you are healthy’ are found in this formula. What has not been observed before is that there is a distribution between these two forms. arog̱es̱i is found in CKD 302; 305; 385; 666; 721; 797; 840, all of which are addressed to only one person.^[30] arog̱etu, on the other hand, is attested in CKD 399; 714; 819. Two documents of this group (714; 819) are addressed to more than one person, while the remaining document (399) has a predilection for plural forms, as will be discussed in more detail infra. In addition, when arog̱es̱i and arog̱etu are combined with an overt personal pronoun in subject position, this pronoun is invariably tuo (tuvo/tu) ‘you.dir.sg’, while in CKD 714 arog̱etu is combined with tus̱a ‘you.dir.pl’ (for which see also below).^[31] Taken together, these observations are further confirmation that the ending -tu (-du) should be seen as 2pl. Besides, one can now confidently restore the only partially preserved greeting formula of CKD 247 as ṣadosmi tusya arog(*etu) ‘I am pleased that you (pl) are healthy’, because of the pronoun tusya ‘you.dir.pl’ and the fact that this letter is addressed to two individuals (Cug̱apa and Priya[śaya]).

A third point worthy of note is that, with one exception (vis̱arjitetu ‘you have sent’ in CKD 399), all of the past tense forms ending in -tu (-du) are found in documents that are addressed to more than one person (CKD 126; 157; 162; 320; 519; 714).

As has been noted above, it seems likely that the ending -tu (-du) could have been used as a polite form as well. In some other cases, 2pl and 2sg may merely have been confused, for which parallels exist in Niya Prakrit.^[32] One sees such mistakes with the pronouns and verb endings of the first person too. A notable instance of such confusion is found in the introduction to CKD 475, here printed as example sentence (5). While no less than three people, named Yapg̱u, Cimg̱ayae and Parsug̱eya, are authoring this document, they still use 1sg verb forms and, strikingly, they combine the 1pl pronoun veyaṃ ‘we’ with a 1sg verb form arog̱emi ‘I am healthy’.^[33]

Confusion between sg and pl forms can also be observed in nominal morphology. In the first part of CKD 399, for instance, one cozbo-official, named Ṣamas̱ena, is greeted by a string of honorific epithets, some of which are in the expected gen.sg (e.g. priyadarśanasya ‘having a dear look’), while others are gen.pl (e.g. priyadevamaṃnuṣyana ‘dear to gods and humans’). Addressing this cozbo Ṣamas̱ena with plural forms also continues in the rest of the document, where we find 2pl pronouns as tumahu and tusmag̱a and, interestingly, verbs ending in -tu (-du).

In view of the preceding arguments, it is now fair to conclude that the ending -tu (-du) marks 2pl in Niya Prakrit and not 2sg, as previously thought. As a result, Burrow’s paradigms of the present and the past tense, which were given above in Tables 4 and 5 respectively, should now be adapted. In Tables 11 and 12, the new paradigms are given with the differences indicated in bold face. Note that, in accordance with the absence of any attestation of the ending -tha for present tense, I have left it out of the paradigm for the present. I have further also included the possible translation of present forms ending in -tu (-du) as imperatives.

3.4 The etymology of the verbal ending -tu (-du)

As an offshoot of the synchronic description, this subsection presents a few stray remarks on the historical explanation of the ending -tu (-du). Various, often mutually exclusive, proposals have been made on the etymology of this verbal ending in Niya Prakrit. Thomas (1934: 49, fn. 5; 51, fn. 3; fn. 4; 57, fn. 1), for instance, presented no less than three different hypotheses on the origin of the ending -tu (-du). One suggestion of his is that this ending would be connected to the absolutive (“gerund”) in -tu, which has been observed in various Aśokan Prakrits (cf. Bloch 1950: 79). Yet Niya verbs ending in -tu (-du) do not have the meaning of an absolutive and absolutives in Niya Prakrit normally end in -ti.^[34] Another proposal by Thomas was to derive -tu (-du) from the OIA mediopassive imperative ending 3sg -tām, but this etymology is phonologically unlikely. It remains moreover unclear why Thomas (and, with him, Konow 1938: 155) considers these forms in -tu (-du) to be passive in origin, as davyatu ‘it should be given’ (CKD 399) is in fact the only example of a morphological passive with the ending -tu (-du).

The most straightforward of Thomas’ etymological suggestions is that -tu (-du) comes from the OIA active imperative ending 3sg -tu, derivatives of which are widely found in MIA languages (e.g. in Pāli; cf. Oberlies 2019 [2001]: 401–402). This etymology is not only phonologically attractive, but it also offers us an explanation as to why present tense forms in -tu (-du) can also express a command in Niya Prakrit. When taking the pragmatic context into account, one can moreover comprehend how such a third person ending could have been re-analysed as second person.^[35]

In this regard, it is useful to take a brief look at two passages from the Pāli canon where such an imperative in -tu is used parallel to a second person imperative. A first example comes from the final verse of the Punabbasusutta contained in the Pāli Saṃyuttanikāya (SN I 210) – here printed as example sentence (6) – where a yakṣa mother is addressing her two children, using a second person imperative for her son Punabbasu and a third person imperative for her daughter Uttarā.

Even though the yakṣa mother uses the third person imperative suṇātu, second person deixis is implied, because we are dealing here with a speech act whereby person A, in this case the mother, is directly addressing person B, in this case the daughter. This point is nicely captured by Bhikkhu Bodhi’s translation, as he renders both the 2sg hohi and the 3sg suṇātu as a second person imperative in English.^[37] Letters and royal commands represent this same type of speech act whereby person A is addressing person B (C, D etc.), which in turn makes it understandable that a third person verb form could have become second person in Niya Prakrit.^[38]

It should still be admitted that suṇātu is paired together with a second singular imperative hohi in example sentence (6), whereas I assume this ending -tu to have become a second plural ending in Niya Prakrit. That this is indeed possible can be seen from another passage of the Pāli canon, this time from the Mahāvagga (II, 3.3 = Vin. I 102–103). The excerpt from the Uposathakkhandhaka of the Mahāvagga in (7) is concerned with the recitation at the 15th day of the month (the Uposatha-day) of the Pātimokkha-rules Buddhist monks have to comply with.

The introductory sentence makes it clear that a senior, experienced monk should address the community of Buddhist monks.^[39] As the Buddhist sangha [monastic order] consists by necessity of several monks, this means that 3sg suṇātu in suṇātu me bhante saṃgho ‘the sangha, venerable ones, should listen to me’ is pragmatically comparable to a 2pl imperative of the type ‘listen to me, monks’. Therefore, this usage of suṇātu by the senior monk can be compared to the Buddha’s use of a 2pl imperative suṇātha ‘listen’ in a stock phrase of the Pāli canon used to address the monks, here example (8).

Further confirmation for the view that suṇātu more or less equals suṇātha can be drawn from what follows in the rest of the Mahāvagga-portion. Without there being a change in the type of speech act, the senior monk suddenly switches to the 2pl imperative ārocetha ‘announce’ in his address to the community of monks.

Using these Pāli examples as a comparandum, we can thus hypothesize that this type of third person imperatives came first to be re-analysed in the prehistory of Niya Prakrit as second person (plural) imperatives because of the pragmatic context is which they were used. The assumption would then be that, once the originally imperative ending -tu (-du) was re-interpreted as a second person (plural) ending, its usage got extended, so as to be also useable in declarative statements.^[40]

On a more general level, there is a further reason why the ending in -tu could have been re-interpreted as a plural ending specifically. In the Pāli examples cited above, the imperative is constructed with a nominal subject, e.g. Uttarā … suṇātu, literally ‘Uttarā should listen’. Yet, this type of imperatives can also be used without an overt subject, e.g. suṇātu ‘one should listen’. In such cases, the agent is not always specified, because of which suṇātu can at times even mean ‘anyone should listen’. Typologists sometimes refer to this type of construction as a “non-referential indefinite” (e.g. Ramat and Sansò 2007) and what is relevant for us is that such a construction may come to be marked for plural number. This is so because different options are left open as the exact referent is not specified. In this respect, it is worth pointing to the so-called MAN-impersonals that are found in various European languages (see e.g. Ramat and Sansò 2007; Siewierska 2011). These MAN-impersonals are a type of impersonal construction whereby the indefinite pronoun in subject position is etymologically derived from a lexeme meaning ‘man’. An example is the French impersonal construction on + 3sg verb, e.g. on verra, which literally means ‘one shall see’. Incidentally, this construction with on is further grammaticalized in French, especially so in informal registers, to that of a 1pl, thereby gradually replacing the older construction with the 1pl personal pronoun nous ‘we’ + 1pl verb (cf. e.g. Ramat and Sansò 2007: 104–106; Heine and Song 2011: 616). In Danish, these MAN-impersonals may also refer to the second and third person (cf. Siewierska 2011: 65), in which regard they are closer than the French to what I assume for the ending -tu (-du) in Niya Prakrit.