1. Premise

This paper analyses how the Tamil verb system was understood and described in six grammars of the Tamil language produced by missionaries between the 16th and 18th centuries with the purpose of creating useful tools to help missionaries how to communicate with natives and, thereby, spread the Gospel. This immediately raises the issue of the diglossic situation that characterises Tamil for, as Chevillard (2018) suggests, it would more correctly be considered triglossia to include the poetic language inasmuch as it is “difficult to separate [the three varieties] in practice, because each component of the Tamil triglossia had its own role to play in the global picture of the everyday life in Tamil Nadu” (Chevillard 2021: 2). As such, the data collected in these texts are representative of three varieties albeit with a different degree of representativeness: variety A which is the vernacular Tamil, variety B which is the Modern formal Tamil, and variety C which is the Classical Tamil – as described in Chevillard’s schema (2021: 3). Among the oldest texts, Henriques’ Arte (cf. below in this §) is more representative of A. Later works are more illustrative of variety B.¹ Unsurprisingly, data from the poetical language are marginal or absent from earlier texts – despite missionaries having always been aware of the existence of different varieties (cf. Chevillard 2021; Muru 2010). Nevertheless, data from A, B and C co-exist within the grammars, at least up until when Italian Jesuit missionary Constantius Joseph Beschi [or Costantino Giuseppe Beschi] (1680–1747) compiled two different grammars: one describing the common language (koṭuntamil) while the other focused on the high language (centamil).

The first text considered in this study is among the most studied Tamil grammars, written in the 16th century, namely the Arte da lingua Malabar (ca. 1548) by Henrique Henriques (1520–1600) (henceforth HH, no. 1 in Table 1) of which the only extant manuscript is held in the National Library of Lisbon. Published by Vermeer (1982) it also provides an almost complete classification of all the Tamil lexicon found in the grammar, while an English translation by Hein & Rajam appeared later in 2013. The other manuscripts studied in this paper include the Tamil Arte wrote by Gaspar de Aguilar (1548–?) and copied by Philippus Baldaeus (1623–1671) (henceforth GA/PB, no. 4 in Table 1), already discussed in Francis (2011), Muru (2014, 2020), and Pytlowany (2018). The third text is the Tamil grammar produced by Balthasar Da Costa (c. 1610–1673) (henceforth BC, n. 3 in Table 1) which the author of this paper has transcribed and translated into English (as yet unpublished). The fourth text is an anonymous Tamil grammar manuscript now preserved in the French National Library (henceforth Ind. 188, no. 2 in Table 1) allegedly written earlier than BC’s Arte (see discussion in § 4). The fifth is the Grammatica Damulica by Bartholomäus Ziegenbalg (1682–1719) printed in 1716 (henceforth BZ, no. 5 in Table 1) while the sixth is the famous grammar of Common Tamil (koṭuntamiḻ) produced by Constantius Joseph Beschi (1680–1747) (henceforth CJB, no. 6 in Table 1). Table 1 below provides detailed information about these texts, their authors, and their location, as well as accessible editions and translations.

This article is organised into two main sections, each introduced by a paragraph briefly explaining the main features of Tamil verbs (§ 2) as they are described in modern grammars of standard Tamil. The first section discusses (1) how the verb system was described in the Indian grammars with which missionaries were familiar (cf. § 3); and (2) which Western grammatical framework missionaries would have considered while describing the Tamil verb system (cf. § 4). The second section analyses in detail (3) how Tamil verbs were described by early missionaries between the second half of the 16th century and the first half of the 18th century, prior to Graul’s classification (of Tamil verbs) (cf. § 5). The conclusion section (§ 6) reveals the way in which the classification of Tamil verbs is used today and the extent to which the system highlighted in the Grammar Outline (1855: 38–39) by Karl Friedrich Leberecht Graul (1814–1864) has its roots in these early descriptions, pointing out the achievements of missionaries in the early 18th century. This in turn opens up perspectives for further research and underlines once again the significance of these early documents for the History of the Language Science.

2. A short introduction to the Tamil verb system

Although an outline of the main features of the verb system (i.e. verb conjugations, verb valency, and codification of agentivity) might appear redundant for those who are familiar with Tamil, it seems useful to include it here since it is functional to the analysis provided in subsequent sections. Indeed, even though data gathered in this section derive from descriptions of standard Modern Tamil, these demonstrate the typological distance from the missionaries’ mother tongue and the metalanguage of grammaticisation they encountered while producing their grammars of Tamil.

Verbs in Tamil are divided into finite and non-finite, identified as muṟṟu and eccam respectively by the traditional Tamil grammars. The distinction between finite and non-finite verbs is syntactical rather than morphological: almost all finite verb forms occur as an end predicate of a sentence and may be inflected for tense, mood, person, number, and gender. Non-finite verb forms, by contrast, occur as a predicate of an embedded or subordinate clause, or in compound verb constructions (Lehmann 1989: 48). The main distinction within non-finite verbs is between viṉai-y-eccam, verbs that combine with a following predicate, and peyar eccam, verbs that combine with a following nominal expression (Steever 1988: 11). Both the forms, in traditional Tamil grammar, are syntactically incomplete achieving completion only when followed by another verbal or nominal form (see Chevillard 2021: 5). In addition, there are nominalised forms which may or may not display tense inflection (e.g. verbal nouns, participial, and adjectival nouns). As illustrated in Figure 1, the elements representative of the morphology of a typical Tamil finite verb are the verb base and two grammatical formative affixes, of which the first denotes the tense and the second the person-number-gender (PNG) agreement:

Figure 1

Tamil verb form.

Additionally, the verb stem to which the inflectional suffixes are added may be simple, consisting of the verb root (a), or derived, consisting of a verb root plus a voice morpheme (b), and both may or may not be followed by a causative suffix (c):

The verb forms in (b), tiruppu and tirumpu, are paired verb stems derived from the same verb root tiru ‘turn’ to which a voice morpheme is added.⁵ Their structure is illustrated by Figure 2.

Figure 2

Structure of derived verb stem.

In Lehmann’s (1989) analysis, voice morphemes are considered as part of the derived verb base, in contrast to a simple verb base like tiru or eḻutu. Steever (1983) provides an alternative analysis, considering the voice markers as combining with an adjacent morpheme to create a portmanteau morph.⁶ Indeed, Steever (1983: 55) states that Tamil verb morphology, despite being “predominantly agglutinative; … exhibits certain fusional tendencies so that 12 (VII) conjugational patterns are distinguished according to the morphophonemic alternations that take place between the verb base and the grammatical formatives”. Hence, tirumpu and tiruppu are taken as one morph which combines two morphemes: the root morpheme and the voice morpheme. They are realised by one pormanteau morph at the phonological level where the voice combines with the preceding verb root (cf. examples 5–6), or with the following tense marker (cf. examples 7–8):

Klaiman (1991: 73) states that the basic voice system contrasts weak and strong verbs, while Paramsivam (1979) classifies it as an opposition between affective (cf. example 9) and effective (cf. example 10) voice. As Zúñiga & Kittilä (2019: 7) state, “studies of grammatical voice can be classified into two groups according to whether the notion of voice is defined structurally or functionally”. While a discussion on voice in Tamil language is not within the scope of this article, I here follow Paramasivam (1979) and Klaiman (1991). The latter stands within the functional approach and suggests different kinds of voices and different values for the same kind of voice.⁷ The alternation of voices in Tamil does not necessarily imply suppression or reassignment of arguments among verbally assigned nominal positions. It signifies that the opposition cannot be easily considered as an intransitive/transitive one, but rather that the selection of voice depends on whether the verb does or does not assign a subject argument upon whose referent the denoted action’s principal effects devolve. As evidence of this, one may cite Klaiman’s (1991: 60): “small and unsystematic sample consisting of 60 invariant Weak and 46 invariant Strong verbs … [that] have been isolated from a listing of 223 verb roots in Arden (1969: 150–169) and checked against dictionary entries in Burrow & Emeneau (1961) and the Tamil Lexicon (1924–36)”. This survey specifically highlights that detransitivisation or valence reduction is not a weak or middle voice function. Both invariant strong and invariant weak verbs may be transitive or intransitive:⁸

As Steever (1983: 59) observes, “affective and effective voice are mutually implicating terms: if a root can occur with one, it can occur with the other […] morphologically, the two are mutually exclusive; semantically, mutually implicating. Effectivity thus marks an inflectional, not a derivational, relationship between the two members of the opposition”. Examining a small sample of 140 verbs, Paramasivam (1979: 5, fn. 8) also estimated – without counting Sanskrit loans and verbs derived from onomatopoeia – that about 60% of Tamil verbs contain such a voice morpheme in their stem and can be considered as paired verbs of the same verb root. However, the observation of the grammars analysed in this article, as well as the investigation carried out by Chevillard (2021) on AP’s dictionary both provide a lower estimation of paired verbs compared to Paramasivam’s estimation. Indeed, the value is 15% for the grammars analysed in this article and 14% for AP’s Dictionary, while the percentage may increase up to ca. 31% if one considers the total amount of paired verbs which can be identified in the verb list provided in Appendix 3. Despite this, the role played by voice morpheme in Tamil verbs is unquestionable. In addition to the opposition of the morphemes -mpu to -ppu already discussed above, voice is also marked by further pairs of allomorphs, as summarised by Lehmann (1989: 51):

The examples listed above show how “an affective verb is one the subject of which undergoes, volitionally or not-volitionally, the action (or state or change of state) described by that verb stem” (Paramsivam 1979: 20) while effective verbs “represent actions of the subject on external object” (Paramasivam 1979: 23). Hence, affective and effective voices characterise the subject’s role in the narrated event where it appears as an actor in affective voice and as an agent in effective voice (Duranti 2004, 2007). The subject of the affective verb does not necessarily direct the process expressed by the verb; by contrast, the effective verb is characterised as one whose subject directs the action or process named by the verb root which may reflect on another object (animate or inanimate; human or non-human).

From a formal standpoint, the affective verb stem tirumpu seen in example (9) is characterised by a voice morpheme with a single obstruent, while the effective verb stem tiruppu seen in example (10) has a voice morpheme with a geminated obstruent. The single/geminated obstruent distinction appears in the inflectional process, but it can also occur in the derivational process since the affective/effective distinction can also appear in the first affix-initial position. As such, Paramasivam (1979: 5) argues that “paired verbs fall under three groups on the basis of how the single/geminated obstruent distinction manifests itself”:

Additionally, as already stated, a third class of verbs can also be identified: these are morphological and periphrastic causatives. Paramasivam (1979) discusses the contrast between the two. The morphological causatives are distinguished formally by an affix represented by the weak allomorph -vi- (cf. example 11) and the strong allomorph –(p)pi- (cf. example 12):

The second type, the periphrastic (syntactic) causatives, involve the auxiliary verbs cey ‘make’ (13), vai ‘place’ (14) or paṇṇu ‘do’ (15):

Today, the conjugational verb patterns are separated into different classes on the basis of the morphophonemic alternations hereto discussed. The number of classes may vary between seven (VII) and thirteen (13). This subdivision into seven (VII) or thirteen (13) classes is based primarily on morpho-phonology and, as Steever (1983: 55) observes, when the contribution of affective versus effective voice is taken into consideration in the form of the verb class, the number of classes is likely to decrease. It is noteworthy that the subdivision into seven classes (VII), maintained by some authors (i.e. Schiffman 1999), was devised by Karl F. L. Graul (1814–1864),¹¹ a member of the German Oriental Society, who was also the director of the Leipzig Lutheran mission. Graul arrived in South India in 1849 with the Lutheran mission, followed on the Danish–Hale mission (cf. Neill 1985; Pytlowany 2018), and mastered the Tamil language. This same subdivision was later extended by the authors of the Madras Tamil Lexicon (henceforth MTL), to a total of 13 classes on the basis of a more specific morphophonemic alternation. Table 2 summarises both classifications: Graul (1855: 38–39, =KG) is denoted in bold and in Roman numerals, while MTL (1982 [1924–36, 1939] lxvii) is indicated in Arabic numerals.

There are two particular issues worthy of note that emerge from Graul’s classification: firstly, as the starting point from which the MTL developed its classification, hence it represents the contribution made by missionaries to Tamil studies. Secondly, appearing as it did in the 19th century, Graul’s classification can be considered the end product of the later missionaries, achieved as a result of the work of their predecessors, as will be discussed in later sections of this paper (cf. § 5). What path did the earlier missionaries follow in their identification and understanding of Tamil verb paradigms? How did they interpret paired verbs? How did missionaries describe verbs applying the Latin grammatical framework? Did they have any awareness of traditional Tamil grammar?

The following sections, which demonstrate how earlier missionaries contributed to Graul’s classification, provide answers to these questions.

3. Descriptions of the Tamil verb system in the early Tamil grammars

When I refer to Indigenous Tamil Grammatical Tradition (henceforth ITGT) in this paper, this is represented by two texts, namely the Tolkāppiyam¹² (henceforth Tol.) which describes Old Tamil and the Naṉṉūl ‘the good book’ (ca. 13th century, henceforth Naṉ.) which describes Middle Tamil. Both texts explain the written variety of Tamil. The former, the oldest grammatical treatise, was allegedly compiled by Tolkāppiyaṉār between the first and third century C.E. and continued to have pre-eminence above all other grammatical texts (cf. Annamalai 2016). The latter, written by the Jain, Pavaṇanti, “is a judicious accommodation of some features from the Sanskrit tradition responding to change in language and the linguistic milieu of the medieval period” (Annamalai 2016: 725). It became the new standard for the description of Tamil despite never eclipsing the Tol. Furthermore, the Naṉ. was the grammar to which missionaries had access, as it is evident from Beschi’s references to its grammatical framework in his description of the Tamil language, such as when he refers to an opposition identified in the Naṉ. to explain the difference between Latin nomina propria and nomina appellativa (cf. Chevillard 1992: 83).

Other than a mention of the ‘fluctuations’ that occur in verb forms expressing tense, represented by the diversity of letters placed before word endings, neither the Tol. nor the Naṉ.¹³ explain Tamil tense morphology and verbs are only described in terms of their endings.¹⁴ With regard to paired verbs, the only reference in the Tol. is found in its distinction between itaṉ pālāṉ and piṟaṉ pālāṉ (sūtra 243), which in the Naṉ. are rendered as ēval viṉai and piṟa vinai (Subramanya Sastri 1934: 146). The terms taṉ viṉai (self-act verbs) and piṟa viṉai (other-act verbs) appeared for the first time only in later commentaries on these grammars in order to differentiate verbs with a single obstruent from verbs with a geminated obstruent (Paramasivam 1979: 34–35; also Chevillard 2008: 414).

By contrast, the Latin and Greek grammatical traditions, as does Sanskrit Aṣṭādhyāyī (cf. Klaiman 1991: 82), not only specify information about verb conjugations, but also provide rules for the selection of active, as opposed to middle, voice inflections. In fact, in Indo-European languages like Sanskrit, and indeed Greek, the middle voice has semantic functions consistent with affectedness. In addition, it is also associated with neuter verbs, which are considered as special subcategories of the intransitive verbs, semantically inchoative and alternating formally with the related transitive, or are associated with non-eventuality, non-punctual, or atelic temporal-modal senses (Klaiman 1991: 103). Similarly, in some Latin grammars, special emphasis is placed upon media tantum (middle-only) class as opposed to activa tantum (active-only) class: the former consists of verbs that are characterised as non-active in form, but which have an active meaning. They are referred to as deponents and cannot be marked by the passive inflection (Klaiman 1991: 78, 97). This matter could be expanded further but a deeper examination falls beyond the scope of this paper. Suffice to say that, with regard to the treatment of verbs in Tamil and the identification of paired verbs, we can confidently state that missionaries were better informed by their Western grammatical framework than by the Tamil grammatical tradition since the latter neither differentiates between paired verbs and morphological causatives, nor does it provide a paradigm of verbs declension on the basis of the tense morpheme.

4. The missionaries’ grammatical model of reference for their description of Tamil

In terms of a model of reference for these early missionaries, I considered João de Barros’ grammar (1540) for HH’s Arte (MS no. 1, Table 1) as suggested by author at the beginning of his text where he points out to the reader the need to be familiar with Barros or any Latin grammar in order to navigate his Tamil grammar (cf. Muru 2018: 15). Indeed, examining his technical terminology and the conceptual framework, it seems that HH made use of Barros’ theoretical framework and process (cf. Appendix 1) in devising his Tamil grammar, although he did not strictly follow the structure of Barros’ grammar. Additionally, HH’s Arte shares some features with manuscript no. 2 (Ind. 188) in Table 1. Both authors itemise declensions for nouns, both identify different conjugations in Tamil (nine conjugations in HH, five conjugations in Ind 188), both include categories such as personal and impersonal verbs absent in later works and employ technical terms such as especies (species)¹⁵ and creçencias (augments),¹⁶ largely found in Barros’ text. Based on the metalanguage used in the description of Tamil I am inclined to consider manuscript no. 2, or at least its contents, to be older than manuscripts no. 3–6 in Table 1. However, this hypothesis of dating for manuscript no. 2 requires further investigation.¹⁷

Indisputably, manuscripts no. 1 and no. 2, and manuscript no. 3 in part, differ from the other missionary grammars. This is likely due to the innovation introduced by the Ratio studiorum in 1599 which promoted the Latin grammar composed by Manuel Álvares (1526–1582) De Institutione Libri Tres (1572), and in particular its abridged version (Ars Minor, 1573), as a textbook for teaching Latin language in Jesuit colleges. Subsequently, Álvares’ grammar became one of the texts most widely translated into different European languages (Fernandes 2007; Gómez Gómez 2002: V; Springhetti 1961–62: 287), and was adapted to the study of non-European languages in various parts of the world, becoming the principal reference model for most missionaries (Assunção & Toyoshima 2012; Fernandes 2015; Muru 2020).

As stated above, all these grammars (cf. Table 1) are representative of the earliest versions of Tamil descriptions which today exist in the form of autograph, or copied, manuscripts and printed books among which CJB’s grammar represents a turning point, since he was the first missionary to attempt to combine and find the links between the ITGT and Western Grammatical Tradition (henceforth WGT).

A detailed reconstruction of the circulation and diffusion of these manuscripts within the missionary community is beyond the scope of this paper; nevertheless, a few facts about the interweaving of these grammatical works in the missionary network (some of which were already addressed by Muru 2018; 2020) might provide a better appreciation of how the knowledge and description of Tamil (verbs) progressed.

There are two kinds of references which offer evidence of the connections among these works. The first is a direct reference to an author/work by another author. For example, in two manuscripts of BC’s Arte and in the preface in PB’s Arte there are (not entirely favourable) references to GA’s work (see Muru 2018: 15; Muru 2020: 60–63). Connections are also evidenced by indirect references, such as those inferred by comparisons of the contents of the manuscript, as when Jeyaraj (2010) demonstrates that BZ’s grammar relied a lot on the copy of BC’s work – even though the Tamil verb forms found in BZ’s text are unique when compared to the other manuscripts (cf. Appendix 3). Indeed, in reproducing BZ’s grammar, Jeyaraj (2010) identifies all instances of BZ copying from BC’s manuscript without even mentioning it. These connections highlight how the first contribution of Westerners to Tamil studies, made by Jesuits, also found a following among the networks of Protestant missionaries, who relied on the Jesuits’ work for their translations of religious books (cf. Israel 2011; Muru 2019). As such, for manuscripts no. 3 and no. 4 (cf. Table 1), and indirectly also for manuscript no. 5 (cf. Table 1),¹⁸ Álvares’ text offers the best reference model, extended by missionaries to describe the Tamil language (cf. Muru 2020). Appendix 1 compares Álvares’s (1573) table of contents with the manuscripts no. 3–5 (cf. Table 1) highlighting the similarities between Álvares’ content and organisation and those of the missionary grammars.

As already mentioned, manuscript no. 6 stands apart, CJB being the first to have worked to establish a theoretical connection between ITGT and WTG. Manuscript no. 6 also marks the differences between these manuscripts and those for which Barros (1540) can be considered as a model of reference (cf. manuscripts no. 1 and 2). Since, however, the detailed analysis of the application of Western or Indian grammatical frameworks is not the focus of this paper, this aspect will not be discussed further. Germane to this paper is specifically the treatment of verbs in both traditions, through which the principle behind missionaries’ classifications of Tamil verbs can be better understood, and to which I will return in the conclusions, having first discussed in detail how missionaries classified Tamil verbs.

5. The description of verbs in the early missionary grammars of Tamil

Each of the following sub-section discusses in detail how the missionaries, analysed in this study, classified, and understood Tamil verbs. A comparison of their descriptions highlights the progression the missionaries made in their understanding of Tamil verb paradigms and paired verbs (see also Chevillard 2021: 25–29)– which will be further discussed in the conclusion –, and it clearly reveals how much the classification of Tamil verbs provided by later descriptors like Graul owed to these early missionary descriptions.

5.2 Verb conjugations and paired verbs in MS Ind 188

As already stated, the author of MS Ind 188 (manuscript no. 2) – as did HH – differentiated between personal and impersonal verbs. He, however, only identified five conjugations. He found that there is no passive in Tamil, although there are some verbs through which it is possible to derive it (i.e. paṭu ‘to suffer’ when added to the infinitive).

However, with regard to the verb paradigm (cf. Appendix 2, section B), he went further than HH. He not only reduced the number of conjugations but also made almost every conjugation correspond to a specific verb class. In fact, apart from the 1^st conjugation which includes strong and weak verbs, all the others include verbs belonging to the same class. Thus, the 2^nd conjugation includes strong verbs, while the remaining conjugations include weak class verbs. Furthermore, he also clearly states that some verbs cannot be included in any conjugation. These are the verbs belonging to the 7^th, 8^th, and 13^th classes.

As HH did, the author of MS Ind 188 identified the conjugations on the basis of the final ending of the verb stem when it precedes the present tense morpheme. Thus, verbs of the first conjugation have -i, -ai, and -a as the final sound of the verb stem. In the identification of this first conjugation, the past tense morpheme is not included as a selective criterion since it varies among the verbs included in this conjugation.²⁰ By contrast, past tense allomorphs are used in the identification of the 3^rd and 5^th conjugations for which all the verbs form the past in -nt- (i.e. uri-tal, ‘to peel’) and -iṉ- respectively (i.e. ettu-tal ‘to inveigle, lure, cheat’). With regard to the 4^th conjugation, once again the only valid criteria are the sounds which precede the present tense morpheme (i.e. lu, ḷu, ru) rather than the past tense allomorphs which vary greatly. Indeed, the verbs included in this conjugation belong to different classes, specifically the 2^nd, 3^rd, 4^th, and 5^th, and form the past with different allomorphs, namely -nt- for verbs in the 4^th class (i.e. akaḻ-tal ‘to leave, vanish’), -ṭ- for verbs of the 2^nd class (i.e. koḷḷu-tal ‘to buy, size’), and -iṉ- for verbs in the 5^th class (i.e. pal(u)ku-tal ‘to increase in measure, to multiply’). Finally, as was the case with HH and for later missionaries discussed in the following sections, the author of MS Ind 188 based the verb paradigm exclusively on formal aspects. The semantics of the verb, as well as the arguments that a verb may take, is not remotely considered.

In terms of the kind of verb stem, the author of MS Ind 188, as well as HH did (cf. Hein & Rajam 2013: 99; Vermeer 1982: 47) identified the morphological causative -(p)pi- and -vi-, and interpreted it as deriving from the future tense and as being used for generating compound verbs:

a.

“há verbos compositos que fazem executar a sua significação e componse da 1à pessoa do futuro do Indicativo, mutando o, en, em, icren, ut eu crerey, fazo crer; eu fazerey, , fazo fazer, ou mando fazer e vaõ pella prima coiugaçaõ” (MS Ind 188, f. 41r).

[There are compound verbs which make [someone else] execute their meaning, [the action expressed by the verb]. They are composed of the first person of the future of the Indicative. changing en, in, icren, as in vicuv[ā]cippēṉ ‘I will believe’, vicuvācippikkiṟēṉ, ‘I make someone to believe’; ceyvēṉ, ‘I will do’, ceyvikkiṟēṉ, ‘I make someone to do, I order someone to do’. They are in the first conjugation.]

The morphological causatives, formally distinguished by an extra affix and considered compound verbs, are compared to other compound verbs with a similar causative meaning (verbs meaning “fazer que outrem faça” [‘make others do’] (MS Ind 188, f. 41r), formed by adding a particle -tugren-, that results in the geminated obstruent (-tt-) occurring in the base-final position of affective verbs followed by the present tense morpheme (-kiṟ-) and the PNG marker (-ēṉ). Indeed, the listed verb forms are paired verbs of affective versus effective voice, like naṭa (12^th) vs. naṭattu (5^th) corresponding to the second class of basic Tamil verbs identified by Paramasivam (1979: 11) and included in the third group of paired verbs, characterised by a single obstruent in the affix-initial position but with a geminated obstruent in the base-final position (Paramasivam 1979: 9–10).

5.4 Verb conjugations and paired verbs in BC and BZ

As highlighted in the previous section, the texts of BC and BZ share similarities above all in their discussion of the verbs. Given this, I have considered them as a single work, since the sole differences between the two authors are the different verb forms used for explaining the rules, the different number of rules for the formation of the preterite in Tamil found in BZ, compared with those outlined in BC, and in some misinterpretations BZ made while he was relying on BC’s Arte.²³ As already mentioned, this is due to the fact that, as Jeyaraj (2010: 20) states, “BZ made use of Da Costa’s Tamil grammar entitled Arte Tamulica”²⁴ while composing his Grammatica Damulica printed in 1716. For this reason, I will refer exclusively to BC’s work and will mention BZ only when differences occur, or it is relevant to the discussion.

BC deals with Tamil verb conjugations and paired verbs in the Third Title, after having explained, through the verb vicuvāci-ttal (‘to believe’), the only declension existing in the Tamil language. In this section, BC explains how to form past and future tenses, the imperative mood, and noun genders. He starts by declaring how forming the past tense is awkward and confusing for beginners because verb conjugations cannot be reduced to one general rule. Thus, in order to overcome this difficulty and to limit problems for beginners, he provides just twelve simple rules (rather than conjugations) for forming the preterite, leaving aside the usage of the language and all other information necessary for the Tamil teacher to know about Tamil verbs:

f.

“Daõse alguãs regras per preteritos Futuros Imperativos e plurais dos pretéritos. He materia esta assâs embaraçada e cõfuza p(ar)a os Principiantes porq(ue) pella variedade de preteritos se naõ pode todos reduzir a regras geraes, se bem despois de poucos annos de exerciçios se vem a cair sem se sentir tanto, nesta materia q(ue)em nenhu’ ou poucos preteritos se erra. Nos p(er)a q(ue) nella digamos alguã cousa e por evitar por todos os preteritos, no vocabulario, formaremos alguas regras gerais, ou quase gerais Remettendo os preteritos que nellas senaõ cõprenderem, ao Vocabulario, e deixando muitas particularidades ao uzo por naõ cauzar confuzaõ aos principiantes”. (MS 50, f. M-34–42, Left column, lines 9–28)

[Wherein some rules are given for preterite Future Imperatives and plurals of preterites. This is a rather awkward and confusing matter for Beginners, for due to the variety in preterites we cannot reduce them all to general rules, although after a few years of exercises one comes to use them properly, so that one will eventually barely misuse them, if at all. We will state a few things so that we do not need to have all preterites in the vocabulary, we will form some general rules or quasi-general. We will place the preterites not included in (these rules) in the vocabulary, leaving their many particularities to usage, so as not to create confusion among beginners.]

Essentially, BC started from the key distinction identified by GA/PB (cf. Appendix 2, section D), namely that the 1^st rule involves verbs ending in cradu (-kkiṟatu) which correspond to the class of the strong verbs, while the 2^nd rule regards weak – those verbs ending in gradu (-kiṟatu). It is hardly surprising that BC abandoned the intention identify different conjugations in Tamil, suggesting instead only rules for the formation of different tenses and moods. In reality, BC was reproducing and, at the same time, extending what GA/PB included in his Arte, a grammar with which BC (or the copyist of his Arte) seems to have been well acquainted. As already discussed in Muru (2018, 2020) and stated above (cf. § 4), BC mentions GA’s Arte on several occasions in his work, at times critically. However, it is evident that he agreed with GA that there is no variation in the final declension of Tamil verbs, that thus there are no conjugations, and that some verbs are derived from others. To what extent then, did BC extend GA’s grammar? He explored the variation, identified but not explained by GA, which occurs in the formation of the past tense – to which the 3^rd –12^th rules are devoted. These rules are identified on the basis of the final sound of the verb stem (V or C) that precedes the present tense morpheme and on the type of present tense morph, with or without gemination of the obstruent. These two parameters are also combined, as was the case in HH, with the type of past tense allomorphs taken by the verb. Nevertheless, the number of verb classes identified and the divisions into classes vary; and BC’s verb classification is better defined than GA/PB’s. The rules outlined by BC are as follows:

•3rd rule = verbs ending in a+ kkiṟ		➔	pst: nt	includes verbs of the 12th class
•4th rule = verbs ending in ē/ō+ kiṟ		➔	pst: nt	includes irregular verbs (13th), 4th and 5th class
	but ō + kkiṟ	➔	pst: iṉ
•5th rule = verbs ending in u+ kiṟ		➔	pst: iṉ	includes verbs of the 5th class
•6th rule = verbs ending in ḷu+ kiṟ		➔	pst: ṇṭ	includes verbs of the 2nd class
•7th rule = verbs ending in lu + kiṟ		➔	pst: ṇṭ	includes verbs of the 3rd and 5th class
•8th rule = verbs ending in ru +kiṟ		➔	pst: ṉṟ, ṉṉ	includes irregular verbs (13th), 4th and 5th class
•9th rule = verbs ending in ṭu+kiṟ		➔	pst: ṭṭ	includes verbs of the 6th class
•10th rule = verbs ending in ṅ+kiṟ		➔	pst: iṉ	includes verbs of the 5th class (neuter verbs = effective)
•11th rule = verbs ending in ṟ + kkiṟ		➔	pst: tt, ṟṟ	includes verbs of the 6th class

The final rule (12^th) is connected to the 10^th, since it includes all the active (of the) neuter forms given under the 10^th rule.

•12th rule =	in kkiṟ = active verbs (=affective)	includes the corresponding pairs of
		verbs of the 5th class (effective)

What emerges thus far is that in BC verbs are classified, as in GA, largely according to morpho-phonemic criteria, taking into consideration:

• 1) verbs which differentiate between geminate (cradu, -kkiṟatu) versus non-geminate (gradu, -kiṟatu) present tense morpheme;

• 2) the final sound of the verb stem;

• 3) the past tense allomorphs taken by verbs;

but he also took into account verb diathesis inasmuch as:

• 4) verbs in the 12^th rule are a counterpart to those in the 10^th rule, the former being active and the latter neuter.

If one excludes HH, all the manuscripts discussed so far differentiate between active and neuter based on a morpho-phonological criterion (gemination vs. non-gemination). However, when discussing the causative in addition to the passive, their differentiation is also based on semantic criteria.

With regard to the opposition within paired verbs, as in HH, it is perceived as a contrasting of verbs in the active and ‘like passive’ voice, presumably taking into consideration both the formal structure and the meaning:

g.	“Da Voz Passiva
	Naõ fallando nos verbos passivos que há nesta lingoa acabados em gren, q(ue) como naõ tem regra certa só cõ o uzo se podem aprender, como saõ de eu acabo, eu sou acabado; de eu quebro, eu sou quebrado, e outros muitos. Há outros compostos e propriamente passivos.” (MS 50, M-34–36, Left column, lines 4–13)
	[Of the Passive Voice
	Leaving aside the passive verbs in this language ending in gren (-kiṟēṉ), which, as they have no certain rule, can only be learned by use, as is the case of (mukikkiṟēṉ) I finish, (mukikiṟēṉ) I am finished; from (muṟikkiṟēṉ) I break, (muṟikiṟēṉ) I am broken, and many other modes; there are other [verbs] that are composite and passive proper.]

In this respect, it is worth mentioning that BZ stated that the formation of the passive voice is twofold: passives which occur without adjustments within the same verbs and passives that are formed through the compounding of the infinitive with the verb paṭu-tal (to suffer). The paired verbs are listed as the first kind of passive that “must be learnt, not by many rules, but from practice” (Ziegenbalg 1716: 80–81; Jeyaraj 2010: 111):

(h)	“	Finio	[mukikkiṟēṉ	I complete [something]
		Finior	mukiyiṟēṉ	I get completed
		Frango	muṟikkiṟēṉ	I break
		Frangor	muṟikiṟēṉ	I get broken
		Corrumpo	aḻikkiṟēṉ	I destroy
		Corrumpor	aḻikiṟēṉ	I get destroyed [i.e., I perish]
		Contero	nacukkukiṟēṉ	I crush [something]
		Conteros	nacuṅkiṟēṉ	I am crushed
		Flecto	vaḷaikkiṟēṉ	I bend [something]
		Flector	vaḷaikiṟēṉ	I am bent [i.e., I bend]
		Concoquo	cerikkiṟēṉ*	I digest
		Concoquor	cerikiṟēṉ*	I digest [myself]
		Excito	eḻuppukiṟēṉ*	I arise
		Excitor”	eḻumpukiṟēṉ	I am caused to arise [i.e., I stand up]
	(Ziegenbalg, 1716: 80–81)		Jeyaraj 2010: 111, *apart from the vowel lenght which has been added, these transcriptions respect what is written in BZ’s original book and differ from Jeyaraj]

Morphological causatives are explained in BC in the section devoted to verb composition:

(i)

“3º se fâz cõ o verbo do mesmo modo, ut Eu creyo. Esta composição se acha as vezes cõ significaçaõ passiva, ut sou espancado, mas õ ordinario he significaçaõ activa, ut creyo, venci naõ confia, etc.25 Tembem da p(rimei)ra pessoa do futuro affirmativo de qualquer verbo se formaõ outros effectivos [emphasis mine] mudando o , em ou em , et , e signifiçaõ [o fazer] exercitar por outro tem a significaçaõ do verbo simples ut Eu faço crer Eu faço chamar. eu faço fazer etc. E vem a formar outros verbos q(ue) tẽ todos os [modos], tempos e pessoas.” (MS 50, M-34–37, Left column, lines 7–31)

[3rd [composition of verbs] is made with the verb of the same mode, as in (vicuvācaṅkoḷḷukiṟēṉ) ‘I believe’. This composition can sometimes be found with the passive signification [meaning], as in (aṭi koḷḷukiṟēṉ) ‘I am [beaten]’, but (the most common) is the active signification [meaning], as in (meykoḷḷukiṟēṉ) ‘(I) believe’, (ceyṅkoṇṭēṉ) ‘(I won)’; (nampikkaikoḷḷāṉ) ‘(he) does not trust’, etc. Also, from the first person of the future affirmative of any verb, other effective [emphasis mine] verbs are formed by changing the (peṉ), or (veṉ) into (pikkiṟēṉ), or (vakkiṟēṉ), meaning to make someone do something, and another has the signification (meaning) of the simple verb as in (vicuvācippikkiṟēṉ) ‘I make someone believe’. (aḻaippikkiṟēṉ) ‘I make someone call’ (ceyyikkiṟēṉ) ‘I make someone do’ and forms other verbs that have all moods, tenses, and persons.]

It is worthy of note that, for the first time, the term effectivos appears to define the semantics of the Tamil morphological causative verbs. The same term is also found in the version of BC’s Arte, copied in 1685, and attached to AP’s dictionary, printed in 1679 (manuscript Borg. Ind. 12 from the Vatican Library). Surprisingly, the three copies of the original exhibit differences: those found at the State Central Library of Goa (MS 15 and MS 49) display the term affectivos rather than effectivos, while the manuscript traced by Jeyaraj (2010) at the British Library (OC Sloane 3003) does not include either term.²⁶ Despite the fact that both effectivos and affectivos seem to anticipate Paramasivam’s terminology, in this context they should be interpreted through the lens of the Latin grammatical framework where the term affectus,²⁷ an imperfect calque from Greek, is used along with signficatio, genus, and vox in reference to the verb diathesis and, only in certain Latin authors, is used to address the patientive role in the pair agere and pati (cf. Benedetti 2001: 225–226; Coccia 2012: 58). Thus, in this context, they should be understood as expressing the semantic and formal change which Tamil verbs undergo with certain suffixes, and through which the verb expresses a cause, or a result of an action produced on someone by the influence of another.

BZ added to BC’s list of verbs (included in Appendix 3), but reduced the explanation regarding verb composition and, in terms of the morphological causative, he merely stated:

(j)	“Non minus fit quædam verbi compositio, quando prima persona Futuri affirmativi vel mutatur in , qua mutatione significatio majorem habet valorem, quam in verbo simplici…” (BZ, 1716: 84–85)
	[“A certain adjustment of the verb occurs when the first person of the future affirmative – (vēṉ) or (pēṉ) is changed into - (pikkiṟēṉ). As a result of this mutation the meaning has greater force than in the simple form of the verb …” (Jeyaraj 2010: 115)]

5.5 Verb conjugation and paired verbs in CJB

I conclude this investigation into how missionaries organised the Tamil verb conjugation and treated Tamil paired verbs with Constantius Joseph Beschi (henceforth CJB), one of the most prominent Western grammarians of Tamil of the early 18th century, whose work, translated into several languages (see Chevillard 2021, cf. § 2), was to become the model of reference for later descriptors of the Tamil language. For example, Rhenius (1836: 79) acknowledged CJB’s deep appreciation of the Tamil verb paradigm, by contrast with the ITGT:

(k)

“the past tense is characterised by the middle particles t, ṭ, ṟ, and iṉ. The native grammarians do not point out which roots require the one or the other of these middle particles. I shall therefore deviate from their course, and rather follow Beschius in describing the formation of this tense; the only one that has any difficulty.”

CJB may be considered as the turning point from the previous missionary tradition for two more key reasons: as already stated, he described Tamil with reference to the WGT and to the ITGT with which he was familiar; secondly, in particular with the Naṉ., he shed light on those questions that had arisen among his predecessors, such as questions relating to the passive meaning of the neuter (affective) verbs.

However, CJB would not have accomplished so much if his predecessors had not worked so extensively on the Tamil language and had not collected linguistic data: primary evidence of this is his explanation of the past tense formation where the synthesis is clearly derived from the grammatical descriptions that preceded CJB. He summarised in eight rules the formation of the past tense, taking into account the same formal criteria that were based on the observation of the final sound of the verb stem preceding the present tense morphemes cradu (-kkiṟau) or gradu (-kiṟatu), thus on internal sandhi. As is the case for the other manuscripts, the full list of verb forms found in CJB’s grammar (Beschi 173: 50–58, parag. 54–65; Mahon 1848: 44–63) is given in Appendix 2, section E. CJB provided three rules for present tense in -kkiṟ- (strong verbs):

•1st rule:		➔	pst = tt-ēṉ	verbs of the 11th class
•2nd rule:	verbs in ă	➔	pst= nd-ēṉ	verbs of the 12th class28
Except. if monosyllabic verbs in ă		➔	pst= (kk)-iṉ-ēṉ	verbs of the 5th class
•3rd rule:	verbs in long syllable	➔	pst= (kk)-iṉ-ēṉ	verbs of the 5th class

The first rule (a, b) includes the strong verbs of the 11^th class; the second rule includes verbs which mainly belong to the 12^th class,²⁹ while the third rule includes weak verbs of the 5^th class³⁰ that have a long vowel in the final syllable of the verb stem preceding the present tense morpheme. CJB considers these forms the correspondent active form of the neuters that he lists in the rule n. 2 among those verbs with a single velar stop in the present tense marker (Beschi 1738: 53, parag. 60; Mahon: 1848: 46), i.e. “aṭakkiṟatu (‘to shut’) is the active form of the neuter aṭaṅkiṟatu (‘to be shut’) that has the past in iṉēṉ”.

With regard to the verbs ending in gradu (-kiṟatu), five further rules are given along with some exceptions:

•1st rule:	a.	verbs with ṅga (ṅ-kiṟ)	➔	pst= ṅk-iṉ-ēṉ		verbs of the 5th class
	b.	verbs with ṇ (ṇ-kiṟ)	➔	pst= ṇṭ-ēṉ		verbs of the 7th, 13th class
	c.	verbs with ṉ (ṉ-kiṟ)	➔	pst= ṉṟ-ēṉ		verbs of the 8th class
•2nd rule:	verbs ending in over short u		➔	pst= iṉ-ēṉ		verbs of the 5th class31
Except. Verbs ending in ku, ṟu, ṭu			➔	pst= dropping ‘u’ and doubling the final C
					(verbs of the 6th class)
•3rd rule:	a.	verbs ending in single ḷ+u	➔	pst=ṇṭ-ēṉ		verbs of the 2nd class
	b.	verbs ending in geminated ḷ+u	➔	pst=iṉ-ēṉ		verbs of the 5th class
	c.	verbs ending in ru	➔	pst=nt-ēṉ		verbs of the 4th class
	d.	verbs ending in lu	➔	pst=ṉṟ-ēṉ		verbs of the 3rd class
	e.	verbs ending in ḻu	➔	pst=varies		verbs of the 1st and 4th class
•4th rule	verbs ending in ya, i, ai		➔	pst=nt-ēṉ		verbs of the 4th class
•5th rule	verbs ending in long vowel		➔	pst=varies		verbs of the 4th, 5th, 13th class

It is worth noting, from the above rules, how CJB identified almost all the classes included in the contemporary MTL with a re-organisation of BC’s classification. Indisputably, he achieved this thanks to earlier missionaries, in particular BC via BZ, as is suggested by the similarities among verbs used as examples (cf. Appendix 3) as well as his references to a previous author – presumably BZ.³²

CJB mentions paired verbs in a section on the passive voice but, unlike his predecessors, he is adamant that they are not passive or ‘like passive’ verbs, but rather that they should be considered as neuter verbs carrying an intransitive meaning:

(l)	“Siquidem quæ ab Alio passiva dicuntur, v.g. , , , &c. non passiva, sed neutra dicenda sunt; non enim passionem a causa exprimunt, sed sensum intransitivum important: v.g. ex augere, activè; numquid erit augeri, passivè? Non certè, sed est crescere,
	neutrum. Sic dicitur, , crevit filius tuus: et ad explicandum augeri, passivè, dicendum . Sic , non est compleri, passive; sed desinere, neutrum. […] Quod ut calriùs adhuc pateat, Latini habent luceo, neut. illumino, act. illuminor, pass. his correspondent , neut. , act. , pass”. (Beschi 1738: 69, parag. 77)
	[Those which, by Another person, are called passives, e.g. (mukkiṟatu), (vaḷarukiṟatu), (vaḷaṅkiṟatu), &c. are not to be called passive, but neuter; for they do not express passion or suffering from a cause, but import an intransitive sense: e.g. means, to increase, actively; is there any (vaḷarukiṟatu) that will mean to be increased, passively? Certainly not, but it means to grow larger, a neuter verb. Thus it is said, (uṉ makaṉ vaḷarntāṉ), your son has increased in size: and to express, to be increased, we must say (vaḷarkkappaṭukiṟatu). So too (mukkiṟatu) does not mean to be completed, passively; but to cease, to leave off, neuter. … To make this appear yet more clearly; the Latins have luceo, I shine, neuter; illumino, I illumine, active; illuminor, I am enlightened, passive. To these correspond (vaḷaṅkiṟatu), neuter; (vaḷakkiṟatu), active; (vaḷakkappaṭukiṟatu), passive (Mahon 1848: 59)].

With regard to morphological causatives, following the Naṉ. (cf. § 3), CJB calls these the verbs of command (ēval viṉai), which means verbs that imply an action done at the command, or at least at the persuasion of that person, who is in the nominative. As in earlier missionary grammars, the causative morphemes are considered as derived from the word used in the formation of the future tense; the word for the future -pēṉ gives -pikkiṟatu; and the word for the future -vēṉ gives -vikkiṟatu (Beschi 1738: 115–116; Mahon 1848: 95).

In conclusion, following the main subdivision already identified by earlier missionaries that split verbs into two main groups – with and without gemination of the velar stop in the present tense morpheme, CJB managed to provide a classification of verbs identifying the different classes. In addition, CJB categorises the paired verbs on the basis of verb diathesis. As such, neuter verbs carry an intransitive significance because the action expressed by the verb refers to the subject itself, while the action expressed by the active verbs produces effects on a referent other than the subject and for this reason, these are different from the neuter.

6. Conclusions

My analysis of the classification of verbs by missionaries between the second half of the 16th century and the first half of the 18th century demonstrates the progress these early descriptors made in understanding Tamil verbs through their application of a Latin grammatical framework. It was a gradual process of comprehension which led them to certain assumptions about the nature of Tamil verbs that had not previously been observed by the ITGT.

There are three principal issues related to the Tamil verbs under observation: the verb conjugations, the paired verbs, and the causative verbs. All six manuscripts, through the adoption of formal criteria, identified a variable number of verb conjugations (in HH and MS Ind 188) or rules (in BC, GA/PB, BZ, CJB) for the formation of present, past, and future tenses in Tamil. In defining these, missionaries also became aware of certain types of verbs which appeared to be the counterpart of other verbs (Tamil paired verbs) which they struggled to describe using categories of Latin grammar. As such, in HH, paired verbs were impersonal forms with a significance ‘like a passive’ which could be derived from personal verbs; in the other manuscripts, they were the neuter of the active verbs. In addition, missionaries also encountered verb forms that signify ‘to make someone do something’ which they deemed to be derived from the paradigm of the future tense because of the formal similarity between the causative morphemes -vi- and -ppi- and the morpheme of the future tense, -v-, -p-, and -pp-.

Although none of the missionaries discussed here made any attempt to identify or isolate a proper ‘voice morpheme’, by applying the category of neuter and active, they all clearly recognised a connection between some paired verbs, including a deviant form (neuter) from another one (active). As observed in earlier sections, only HH and MS 188 offer a differentiation between personal and impersonal verbs, including within the latter the active tantum verbs (HH) and kuṟippu viṉai (MS 188). Only HH articulates that “in some of the conjugation there is a way of speaking that is almost like the passive” (Hein & Rajam 2013: 211), although he acknowledges that the case to which he is in fact referring, namely the affective voice, cannot be considered as completely overlapping with the Latin passive. Despite this, later grammarians like GA/PB continued to question the possibility of considering these verbs as passive, and also introduced a new category, the neuter, while BC, anticipating Paramasivam’s terminology– albeit with a slightly different meaning– made use of Latin categories such as effectivos/affectivos, one of the terms used along with signficatio, genus, and vox to refer to the verb diathesis, even though he still considered the verb ending in gradu (-kiṟatu) as a passive. However, both GA/PB and BC, as well as, BZ, abandoned the differentiation between personal and impersonal verbs and the idea of conjugations for the Tamil verbs. Lastly, CJB clarified these uncertainties, arguing that certain verbs should be considered as neuter rather than passive forms of the corresponding active ones, because they “do not express passion or suffering from a cause but import an intransitive sense” (Beschi 1738: 69, parag.; Mahon 1848: 59).

In summary, missionaries:

• 1) identified the Tamil verb classes on the basis of the final sound of the verb stem and of the tense morphemes – in particular of the present and past tense;

• 2) identified the morphologically causative verbs from each of the paired verbs;

• 3) identified the paired verbs on the basis of the subject’s role in the narrated event as active verbs (i.e. the subject works as an agent producing effects on another entity who is not the subject), or as neuter verbs (i.e. the action carried out by the verb does not necessarily involve another entity and may affect the subject itself), which were associated with intransitive verbs in the first half of the 18th century.

In establishing the main criteria guiding the missionaries’ identification of classes and classified verbs, one can consider Duranti’s (2004: 453) definition of agency: “agency is understood as the property of those entities (i) that have some degree of control over their own behaviours, (ii) whose actions in the world affect other entities (and sometimes their own), and (iii) whose actions are the object of evaluation (e.g. in terms of their responsibility for a given outcome)”. This appears to suggest that for the missionaries the prevailing criteria in the definition of the neuter in Tamil were semantic despite the fact that the main criterion for the differentiation into conjugations or rules was formal. In fact, in their identification of affective verbs with ‘like passive’ or ‘neuter’, there is an implicit understanding of a different degree of agentivity for the subject of the verb, codified through the presence or absence of a gemination of the obstruent in the tense morpheme, or in the verb base. Hence, it is clear that for the division of verbs into neuter and active, missionaries were thinking of the Latin Genera Verbi, which found its roots in the Greek grammar (diathesis) and became, in Latin linguistics, an autonomous inflection category from the 4th century onwards (see Hovdhaugen 1986: 315–329; Hovdhaugen 1987: 133–147). The selection of the category of neuter may have been a device to explain a feature of a linguistic item that differed from a prototype. The Latin Genera Verbi identifies verb species or significations. There were nine³³ species in the 4th century which were later reduced to five distinct types:³⁴ activum, passivum, neutrum, commune, deponens. As such, active verbs are straight, and the neuters come from a deviation of this straightness; they are defective in forming the passive voice.³⁵ Furthermore, the active verbs, like the passive, involve two people (two arguments), the agent and the patient. In contrast, those verbs that do have none of these forms are considered as neuters or absolutive (Stoppie, Swiggers & Wouters 2007: 213–215).

As Benedetti (2001: 228) states, “the Genera Verbi may be considered as the extreme attempt grammarians made to integrate the semantic with the formal description in order to be able to cover all the possible combinations among the series of endings and the conceptual categories of facere and pati [translation mine].” In the same way, the application of two categories of the Genera Verbi to the description of Tamil verbs by missionaries may be considered as an extreme attempt to give an account of the semantic variation that the affective verbs display when compared to the effective ones.

As the definition of affective verbs progressed, it seems that, through the use of impersonal and neuter categories, missionaries were applying the canonical definitions as given in the WGT such as Barros (1540) and Álvares (1573)³⁶ but were also extending them, taking into account not solely the formal aspect but also the significatio related to neuter. As such, one should trace the principle of categorisation within the semantic, rather than the formal sphere, despite the fact that affective differs from effective in the stops that precede the tense morpheme. Indeed, missionaries were trying to explain a slightly different meaning which is codified through the tense morpheme combined with the voice morpheme, according to which certain verbs not only change their tense, but also the degree to which the subject is involved in the action described by the verb. If two entities are involved in active forms, in the neuter form it seems that the only entity predominantly interested by the action is the semantic subject (cf. CJB’s definition of neuter as intransitive).

In conclusion, this paper demonstrates how throughout the application of the Latin framework and its extension, missionaries were able to account for some of the peculiarities of the Tamil verb system, which are not described by the Tamil grammatical tradition. It also highlights how the data offered by these first external descriptions of Tamil, in part representative of data collected through missionaries’ fieldwork, may enable, even today, a better comprehension of the Tamil language – i.e. according to these sources, the incidence of paired verbs in Tamil is lesser if compared with Paramasivam’s estimation thereof (1979). Furthermore, bearing in mind the fact that the missionaries’ identification of Tamil verb classes, as well as of affective versus effective voice continues to be used in modern descriptions of Tamil, it seems almost redundant to highlight the missionaries’ contribution to Tamil studies and to the History of the Language Science in general.

Abbreviations

Competing Interests

The author has no competing interests to declare.