Menu
Free Access
Issue
Histoire Epistémologie Langage
Volume 42, Number 1, 2020
La grammaire arabe étendue
Page(s) 73 - 92
DOI https://doi.org/10.1051/hel/2020004
Published online 28 September 2020

© SHESL, 2020

1 Introduction: the sources

Arabic treatises of other languages present an intriguing type of source material for a number of reasons. These treatises, written by specialists educated in the Arabic grammatical tradition, deal with Turkic vocabulary, grammar, morphology and phonology. In their concise descriptions and analyses, they apply extant concepts, yet they occasionally find that they have to adapt them to this new language. In this contribution, I intend to analyse what their authors write about the passive and about passive constructions.1

The sources used for this paper are the following:

Dīwān Luġāt at-Turk by Maḥmūd al-Kāšġarī (eleventh century; ed. Kültür Bakanlığı 1990);

The Margin Grammar — a large compilation of sections on Turkic grammar (eleventh-fourteenth century) in the margins of a copy of the Kitāb al-ˀIdrāk Veli ed-Din manuscript;

Kitāb al-ˀIdrāk li-Lisān al-ˀAtrāk by ˀAbū Ḥayyān al-ˀAndalusī (fourteenth century; ed. Caferoǧlu 1931);

Al-Tuḥfa al-Zakiyya fī al-Luġa al-Turkiyya, anonymous (fourteenth century; ed. Halasi-Kun 1942);

Kitāb al-Qawānīn al-Kulliyya li-Ḍabṭ al-Luġa al-Turkiyya, anonymous (fourteenth century; ed. Kilisli 1928);

Kitāb Ḥilyat al-Insān wa-Ḥalbat al-Lisān by Ǧamāl al-Dīn ibn al-Muhannā (fourteenth century; ed. Kilisli 1921).

Most of them date from the fourteenth century, the one exception being Dīwān whose original dates from the eleventh century (al-Kāšġarī, Dīwān; Auezova 2005; Dankoff and Kelly 1982‑1985).2 All of these sources have been edited and published, except the one I have called the Margin Grammar.

The Margin Grammar is an as yet unpublished grammar of Turkic, scribbled in the margins and between the lines of the Veli ed-Din ˀIdrāk MS. The author of the Margin Grammar is unknown; in fact, it is probably not an entirely original work, considering the fact that a considerable number of fragments are almost literal copies from other known sources, mainly Qawānīn, ˀIdrāk and Dīwān, while the style and set-up of some others shows strong resemblance to Tarǧumān (ed. Houtsma 1889) (Ermers 1999: 41f).

Of these sources, ˀAbū Ḥayyān’s work is of particular interest, because, in addition to ˀIdrāk, other works of his on Arabic grammar have been preserved, such as his Irtišāf (e.d al-Namās 1984, 1987, 1989) and his commentary on Ibn Mālik’s Alfiyya. This makes it possible to compare his statements on grammar in ˀIdrāk with passages in his other studies.

The sources describe different varieties of Turkic. Dīwān describes Ḫāqānī, a Turkic language spoken in Central Asia at the time, whose modern descendant is Sariġ Yugur or Yellow Uygur (Clauson 1965). The fourteenth-century sources describe the language of Mamlūk military slaves in Egypt and Syria, who spoke a Kipchak language —probably (Crimean) Tatar— with Oġuz (Turcoman) material (Berta 1998). A precise determination of the Turkic language described in each work is complicated, one reason being that there are large overlaps between languages and subgroups regarding morphology and vocabulary, and another that the Arabic script as it is used in the sources does not easily allow for the expression of nuances in vowels, such as the distinction between e/a, o/u and ö/ü respectively (Ermers 1999).

From both morphological and syntactical perspectives, there are considerable differences between Arabic and Turkic. Most Arabic nouns, verbs, and adjectives can be derived from a root, whereas in Turkic word formation is based upon stems. Not surprising, therefore, is the notion that in Arabic grammar a central concept of a verbal root (rather than a stem) was developed, consisting of consonants to which other consonants (ḥurūf, sg. ḥarf) are added as prefixes and infixes, thus adding derived meanings (maˁnā, pl. maˁānī) and forming new patterns (wazn, pl. ˀawzān) that are more or less predictable in form and function. Probably for this reason, the Arabic grammatical tradition could develop into a functional grammar, in which given forms (and positions) are correlated with certain meanings (see e.g. Bohas and Guillaume 1984; Owens 19881990; Versteegh 19951997; Sheyhatovitch 2018: 169ff; Ayoub and Versteegh 2018). This perspective was enhanced by the divine status of Arabic as the language of the Qurˀān, which, it was believed, thus could only have been chosen because of its perfectness and logic in form and structure. As this perspective on grammar was the only one available, all scholars were educated in this model regardless of their origin.

2 Passive in Arabic and in Turkic

2.1 Arabic morphology: consonants of augmentation (ḥurūf al-ziyāda)

Arabic uses morphological infixes, prefixes and suffixes which are placed onto a three-consonantal root. In the Arabic grammatical tradition, the notion of prefixes and infixes to this root is illustrated in an abstract way with the help of the paradigmatical root f-ˁ-l, which in turn is derived from the verb faˁala ‘to do’.3 When filled with the appropriate vowels, the pattern faˁala signifies ‘he did’.

With the help of the appropriate paradigmatic form, the base consonants from any root can easily be distinguished from any additions. In Arabic grammar, the augmented meaningful elements are called ḥurūf al-ziyāda; all augments are consonants. For example, the augmented consonant in the paradigmatical form ˀa-fˁala (paradigm IV in Western studies of Arabic) is ˀa, and the same holds for the prefix in-, in-faˁala where the augment is n only (paradigm VII). After the insertion of the pre- and infixes, often a (secondary) shift of the vowels (naql al-ḥaraka) occurs, for example, often the first root consonant, f-, loses its vowel.

In some paradigms, meanings are added by the simultaneous augmentation of two consonants. For example, the paradigm (VI) ta-f-ā-ˁala signifies reciprocity, i.e. the action of the verb is carried out together or reciprocally by several agents.

Note that in the Arabic morphological tradition the concept of long vowels as morphemes did not exist. The so-called long vowels were understood as a sequence of a consonant, a glide, preceded by a vowel sign: ā = /a”/,4 ī = /iy/, ū = /uw/. In this way, for example, the verbal pattern of the conative form, fāˁala, is understood as /fa”ˁala/, and the additional meaning is attributed to ˀalif, represented as /”/.

Earlier (Ermers 1999: 270‑282) I analysed how Arab grammarians dealt with transitivity and causativity in Turkic. My conclusions were that they tried to uphold their theories regarding the correlations between form and function for Turkic as well (see also Ermers 2007). Yet they had to recognize that in other languages functions could be expressed by different syntactical and morphological elements that did not resemble their Arabic counterparts.5 They were forced, to some extent, to engage in comparative and universal linguistics. In this contribution, I intend to examine how they dealt with the passive form.

2.2 Two types of passivity in Arabic

In Arabic grammar, there are two distinct notions of the passive. The first is the so-called internal or apophonic passive of the verb, on the pattern fuˁila. This passive stands out in Arabic grammar because it is not expressed by means of an infix or, in Arabic terms, a meaningful particle, but by a vowel change: faˁala → fuˁila. The imperfect tense is subjected to changes in the pattern too: yafˁilu → yufˁalu. The subject of an internal passive verb is referred to in Arabic grammatical theory as an-nāˀib ˁan al-fāˁil, which can be translated as “subject by proxy” or “substitute subject” (Soltan 2009: 535). The verb is “built for the patient” or “the logical object” (mabnī li-l-mafˁūl).

In the traditional Arabic grammatical theory, passivization is regarded as a process in which the agent of a transitive verb (fiˁl mutaˁaddin) is deleted or kept hidden or “unknown” (maǧhūl), although not absent, leaving the former object and the verb (Bazzi-Hamzé 2007b: 94). “The subject of the apophonic passive in Arabic is obviously not the agent of the process but rather one who is affected by the process” (Maalej 2008: 224).

This new situation is reflected in the verb, which assumes a new vowel pattern. Then there occurs a (superficial) syntactical shift, in which the direct object of the original verb assumes the subject position, i.e. the syntactic role of agent, and receives a corresponding nominative case ending (Carter 1981: 168ff).

The second way for indicating a passive is the use of the pattern infiˁāl. This pattern indicates an action without an evident cause:

The infiˁāl-pattern by itself expresses an action which the subject carries out by itself, without an agent being implied. According to Abboud-Haggar (2006: 616), it “was used as an alternative for the internal passive” in early texts, such as Qurˀanic Arabic.

However, the infiˁāl pattern is also used in relation to the semantical notion of muṭāwaˁa ‘compliance’, where there is a cause implied; hence Larcher’s (2003: 69) interpretation as a ‘résultatif’. Versteegh (2014: 119) explains the morphological reasoning behind compliance as follows:

Muṭāwaˁa was regarded as the opposite of taˁdiya, that is, decreasing the valency of the verb with one, for example, kasara ‘to break [transitive]’ versus inkasara ‘to break [intransitive]’; ˁallama ‘to instruct someone about something’ versus taˁallama ‘to be instructed, to become learned in something’ (Larcher 2012: 75‑77). What mattered to the grammarians was the fact that the augment (ziyāda) correlated with an additional meaning.

“[t]here is implicit causation underlying all forms of compliance” summarizes Maalej (2008: 225), e.g. fataḥtu l-bāba fa-nfataḥa ‘I opened the door and it opened’ (An-Nādirī 1995: 353 apud Maalej 2008: 225). An-Nādirī, still according to Maalej (2008: 226), writes that the complier is not necessarily intransitive (lāzim), it can also be transitive (mutaˁaddī), which makes sense when the muṭāwaˁa is basically a resultative. The muṭāwaˁa, from a morphological perspective, therefore does not depend on one single verbal pattern, but can be expressed with several, intransitive and transitive, i.e., VII infiˁāl, infaˁala; VIII iftiˁāl, iftaˁala ġamamtu-hu fa-ġtamma ‘I saddened him, so he was filled with grief’; V tafaˁˁul, tafaˁˁala: kassartu l-ˀaqlāma fa-takassarat ‘I broke the pencils, so they broke’ (Maalej 2008: 226), and others.6 Larcher (2009: 642) writes that Form VII itself is already “the resultative of Form I”, although “in many dialects it is used as a passive of the base form.” The internal passive has become rare in modern Arabic dialects, and this process must have begun a long time ago (Carter 1981: 171). It subsists in some modern dialects on the Arabian Peninsula, e.g. in Qatar it occurs with the u-i form, e.g. ḥad qutil hina? ‘was somebody killed here?’ (Belova 2009: 306), while for Oman the imperfect 3msg form prefix has been registered, e.g. yibāˁ ‘it is sold’ (Al-Balushi 2016: 107; for Yemen see Simeone-Senelle 1997: 407). Instead of the internal passive, therefore, in most variants of Arabic the passive is expressed by means of consonantal prefixes and infixes to the verbal root. As a consequence, the distinctions between the agent-less fuˁila passive on the one hand and patterns like infiˁāl on the other have disappeared.

(1) ḍuriba zayd-un
  hit.PASS.3sg.PAST Zayd-NOM
  ‘Zayd was hit [by someone]’  
(2) in-kasara al-kaˀs-u
  COMPL-break.3sg.PAST DEF-glass-NOM
  ‘the glass broke (by itself)’  

2.3 Passive forms in Turkic languages

In all Turkic languages, semantic and voice changes to verbs are added to a stem. The morphemes added can involve one or more consonants (and occasionally vowels). The most common passive suffix is -(V)l-, ört- ‘cover’ → ört-ül- ‘be covered’, kör- ‘see’ → kör-ül- ‘be seen’ (Róna-Tas 1998: 75; Johanson 1998: 42). In Kazakh (and Turkish), the verbal form with ‑l‑ is bi-functional. It serves to form both “non-passives without implied agents”, e.g. yesik aš-ıl-dı ‘the door opened’ (Kazakh), and “true passives with implied agents” ‘the door was opened (by someone)’ or, in other words, “intransitivized transitive verbs” (Şahan Güney 2006: 128).

When a verbal stem ends in a vowel or ‑l‑, ‑(V)n‑ is in Turkic used to indicate passivization, e.g. sı:- ‘break (trans.)’ → sı-n- ‘break’ (Clauson 1972).7 This then coincides with the verbal suffix ‑(V)n‑ which indicates reflexivity, e.g. yu:- ‘wash (trans.)’ → yu:-n- ‘wash oneself’ (ibid.: 870 & 942). Therefore, verbal forms in -n- under some conditions can express either passivity or reflexivity, e.g. kör-ün- (< kör- ‘see’): ‘to be seen’ (passive), ‘become visible’ (reflexive). A third signification of ‑(V)n‑ is the middle voice, e.g. Karakhanid al-ïn- ‘take for oneself’ (< al- ‘take’) (Johanson 1998: 42). The form of the vowel V in the passive and reflexive suffixes is subjected to the principles of fourfold vowel harmony in nearly all Turkic languages: u/ü (rounded: back/front) and ı/i (unrounded: back/front). These cannot be rendered in Arabic script (on Kipchak passive suffixes, see Berta 1998: 160).

3 Passive and compliance in the sources

The question is how the sources deal with passivization and compliance as semantic concepts, how they link the voice to morphemes such as ‑(V)l‑ and ‑(V)n‑, the distribution of these morphemes, and the concepts these morphemes are associated with —i.e. the internal passive or a derived form— and which terminology is used for the notion of stem for suffixes.8

The Turkic passive is dealt with in the sources in different contexts. Unlike Arabic, the past tense verb is not close to the unmarked form, but expressed by an ending, -DI, attached to the stem. The Turkic stem alone expresses the imperative. Some sources, as we shall see, discuss the passive in the context of the passive participle, which in Turkic merely requires the appropriate passive suffix to the stem, plus one other suffix which expresses the participle. In Arabic, the passive participle is expressed by means of a combination of a pre- and infix.

In ˀIdrāk, ˀAbū Ḥayyān uses three different expressions: the first is related to the voice of the verb (mā lam yusamma fāˁiluhu ‘[the verb] whose agent is not expressed’), the second to the consequences for the syntax (al-nāˀib ˁan al-fāˁil ‘substitute agent’), and the third to the semantic notion of the resultative (muṭāwaˁa, lit. ‘compliance’).

3.1 The distribution of l and n

In some sources, the logic behind the distribution of l and n is not explained at all. In the Margin Grammar for example, the author writes quite vaguely that the rule regarding the use of l and n is what one actually hears (al-samāˁ: MG 36a top). Kāšġarī writes at the entry ˀaġirla-n- (< ˀaġirla- ‘praise’, ˀukrima ‘he was praised’) that n in this verb can be replaced (mubdala) by l, yielding the alternative form ˀaġirla-l-, but he does not explain why (Dīwān 148,6; Clauson 1972: 94).

The Margin Grammar writes that an unvocalized l is added before the marker of the personal pronoun (muḍmar), i.e. probably in case of the past tense verb, or before the marker of the future tense (istiqbāl) (MG, 37B right) —an important note, since in Arabic the imperfect tense, used for the future tense, contains elements that express gender and number.

Kāšġarī states:

For every biradical (ṯunāˀī) transitive (mutaˁaddī) verb, if you add an l to it, it becomes an intransitive (lāziman) and passive verb (maǧhūlan) as explained before.

Wa-kull fiˁl ˀiḏā kāna ṯunāˀiyyan mutaˁaddiyan fa-ˀiḏā ˀadḫalta fīhi al-lāma yakūnu fiˁlan lāziman wa-fiˁlan maǧhūlan kamā maḍā. (Dīwān 490)

The anonymous author of Qawānīn provides a short analysis:

The substitute agent (al-nāˀib ˁan al-fāˁil).9 The rule (qāˀida) in this is that you insert (tuqḥim) an unvocalized l [or an unvocalized n] between the imperative form and whatever marker follows.

Al-nāˀib ˁan al-fāˁil − al-qāˀida fīhi ˀan tuqḥim lāman sākina [ˀaw nūnan sākina] bayna fiˁl al-ˀamr wa bayna mā yalī min ˁalāma. (Qawānīn 26)

The author does not elaborate further. Yet ˀAbū Ḥayyān in a very concise manner provides more detailed rules regarding the distribution of -n-:

If [the verb] is uniradical or biradical, and its second consonant is either vocalised or a silent l, or if [the verb ends in] , which serves the action (ˁamal), the addition (mazīd) is an unvocalized n (ˀIdrāk 133).

ˀAbū Ḥayyān bases his distribution of -n-, on the following criteria:

  • 1. the verb is uniradical, i.e. CV- or

  • 2. the verb is biradical:

  • a. the final consonant of the [biradical] verb is vocalized (i.e. the stem ends in a vowel), i.e. CVCV- or

  • b. it ends in an (unvocalized) l, i.e. CVl- or

  • 3. the verb, of any length, ends in lā- (examples below).

Qawānīn describes four contexts which determine the form of the suffix, i.e. when an n is used:

The rule (ḍābiṭ) regarding the position (mawḍiˁ) of the n is [1] that the verb10 consists of one single consonant, like y [i.e. ya-] meaning ‘eat!’ (kul), [2] or of two consonants, the second of which is vocalised (mutaḥarrik), like tuša, meaning ‘spread!’ (ufruš), [3] or it consists of two consonants the second of which is not vocalised, but it is an l [that is used], e.g. ˀal meaning ‘take!’ (ḫuḏ), ṣal meaning ‘throw!’ (irmi), [4] or the verb has more consonants, the final one being which serves the action (ˁamal) [i.e. used to construct denominal verbs], e.g. yumruq-lā [< yumruq ‘fist’] ‘punch!’ (ulkum) ˀaylā! ‘work!’ [< ˀay ∼ e:ḏ ‘material’11] (iˁmal) and sūy-lah! [< söz ‘word’] ‘talk!’ (taḥaddaṯ) and the like. These are the positions of the n, and all the rest [is] the position of the l. [5] So you say in the constructions of these examples according to their order: yi-n-il-dī [‘he was eaten’] and tuša-n-dī [‘it was spread’], ˀal-in-ḍī [‘it was taken’], ṣal-in-ḍī [‘it was thrown’], yumruq-la-n-dī [< yumruq ‘fist’ > yumruq-la- ‘to punch’] [‘he was punched’] and sū-la-n-dī [<  ‘water’ > sū-la- ‘to water’] [‘it was watered’] and use this as a general rule (qis).

Wa-l-ḍābiṭ li-mawḍiˁ al-nūn ˀan yakuna al-ismu [sic, al-fiˁl, R.E.] [1] ˁalā ḥarf [wāḥid] naḥwa ya bi-maˁnā ‘kul’. [2] ˀaw ˁalā ḥarfayni wa-l-ṯānī minhumā mutaḥarrik naḥwa tuša bi-maˁnā ufruš’. [3] wa-ˁala ḥarfayni wa-l-ṯānī minhumā sākin wa-lakinnahu lām, naḥwa ˀal bi-maˁnā ‘ḫuḏ’, wa-ṣal bi-maˁnā ‘irmi’. [4] aw yakūna al-fiˁl ˁalā ˀakṯar min ḏālika wa-ˀāḫiruhu , allati li-l-ˁamal naḥwa yumruq-lā12 bi-maˀna ‘ulkum’, ˀaylā bi-maˁnā ‘iˁmal’ wa-sūylah bi-maˁnā ‘taḥaddaṯ’ wa-naḥwa ḏalika. Fa-hāḏihi mawāḍiˁ al-nūn wa-mā ˁadāhā mawḍiˁ al-lām. [5] Fa-taqūlu fī bināˀ hāḏihi l-ˀamṯila ˁalā t-tartīb yi-n-il-dī wa-tuša-n-dī, ˀal-in-ḍī, ṣal-in-ḍī, yumruq-la-n-dī wa-sū-la-n-dī wa-qis ˁalā ḏālika. (Qawānīn 26)

If we rephrase Qawānīn’s statements in a more formal notation, the following picture of the distribution of n emerges:

  • 1. CV- —one consonant, e.g. ya-;

  • 2. CVCV- —two consonants, a vowel follows the second consonant, e.g. tuša-;

  • 3. CVl- —two consonants, the last one being an l, e.g. ṣal-;

  • 4. -lā —verbal stem ends in -lā— this category includes denominal verbs —e.g. yumruq-lā-, a denominal verb from yumruq ‘fist’.

Yet the examples Qawānīn gives still deviate from these rules. For example, instead of yi-n-il-dī, [‘it was eaten’], which contains a combination of n and l, i.e. a two passive suffixes on the stem yi- ‘to eat’, one would expect a form like yi-n-.13 Another point is that in his account the anonymous author does not account for the distribution of l.

If we combine this statement with ˀAbū Ḥayyān’s concise summary, it evolves that they are essentially identical:

  • 1. (CV)CV- —one or two consonants;

  • 2. (CV)CVl- —two consonants, the final one being an l;

  • 3. verbal stem ending in -lā.

The limitation in both sources on the number of consonants is difficult to understand as there are many verbs consisting of more than two consonants, not ending in -lā to which -n- can be added.

3.2 L and n as markers of the internal passive

All sources deal with the internal passive, yet not in the same way. Qawānīn, for example, relates Turkic l and n to the Arabic u-i pattern in the unmarked verb: “The rule (qāˁida) in this is that n is that you insert (tuqḥim) an unvocalized l [or an unvocalized n] between the imperative form and whatever marker follows” (cf. also 4.1). Examples (without Arabic equivalents) are ya-n-il-dī ‘it was eaten’ (which contains a double passive, one -n, directly after the stem, ya- ‘eat’, and a second passive suffix in -il-), tuša-n-dī ‘it was spread out’ [< tuša-], ˀal-in-ḍī ‘it was taken’ [< ˀal- ‘take’], ṣal-in-ḍī ‘it was set free’ (< ṣal-), yumruq-la-n-ḍī ‘he was punched’ [< yumruq-la- ‘to punch’ [denominal v. < yumruq ‘fist’], sūla-n-dī ‘it was watered’ [< sū-la-, denominal v. < sū- ‘water’] (Qawānīn 26).14

ˀAbū Ḥayyān briefly discusses the passive form with -Vl- under the heading “Chapter on the addition” (al-Qawl fī al-ziyāda) where a great number of nominal and verbal are listed (ˀIdrāk 111‑116). He writes: “[the l] is added (tuzādu) as an indication (dalālatan) that [the verb] is ‘formed for the passive form’” (bināˀ al-fiˁl li-l-mafˁūl) (ˀIdrāk 115), i.e. the internal passive:

Not only l, according to ˀAbū Ḥayyān, but also n serves the function of marker of the internal passive as well. Elsewhere, in the “Chapter on the substitute of the agent” (ˀIdrāk 133), ˀAbū Ḥayyān gives the following examples of n and the internal passive: ya-dīˀakalaya-n-dīˀukila; ṣi-dīkasarasi-n-dīkusira. Some of these examples recur in a summary under the header “the logical object whose agent is not mentioned” (al-mafˁūl mā lam yusamma al-fāˁilu-hu) (ˀIdrāk 112) albeit without translations into Arabic.

(3) ˀur-il-dī
  beat-PASS-3sg.PAST
  ‘he was beaten’

3.2.1 The passive participle

The author of the Margin Grammar chooses an approach based upon the passive participle, which in Turkic is marked by a passive stem plus the ending -KAn:

[1] The “passive participle” in Arabic —as is well-known— can only be [derived] from the transitive verb (al-fiˁl al-mutaˁaddī); the same is true in Turkic. [2] Its marker (ˁalāma) [i.e. of the passive participle] is that you insert (tudḫil) an unvocalized (sākin) l or an unvocalized (sākin) n between the basic imperative verb (fiˁl al-ˀamr al-muǧarrad) and the marker of the connected agent (al-fāˁil al-mawṣūl).

[1] Ism al-mafˁūl − wa-qad ˁulima fī al-ˁarabiyya ˀannahu lā yakūnu ˀillā min al-fiˁl al-mutaˁaddī fa-kaḏālika fī al-turkiyya. [2] Wa-ˁalāmatuhu ˀan tudḫila bayn fiˁl al-ˀamr al-muǧarrad wa-ˁalāma al-fāˁil al-mawṣūl lāman sākinan (sic) ˀaw nūnan sākinan (sic). (MG 36a top)

The author here says, first, that a passive form can only be construed from a transitive verb. This is not entirely true, because in Arabic grammar intransitive verbs, e.g. ḏahaba ‘go’, the formation of impersonal passives is allowed (Girod 2007: 315): e.g. ḏuhiba ˀilā al-qudsi, lit. ‘it was went to Jerusalem’ (Saad 2019 [1982]: 2).

He then explains that the marker l is put after the stem, but before -KAn. He exemplifies this with wur-ġān (al-ḍārib) ‘the hitter’ and wur-ul-ġān (al-maḍrūb) ‘the one that is hit’. He then analyses the participle ending -ġān as the marker of the connected agent (al-fāˁil al-mawṣūl), and thus appears to assign other significations to it, perhaps because this same ending is also used for the active participle. The Turkic ending does not contain any information regarding gender, number or passivity:

A similar statement can be found in Tuḥfa. The passive participle (ism al-mafˁūl) and the passive according to the pattern fuˁila (al-mabnī li-mā lam yusamma fāˁiluhu):

“Its marker [i.e. of the passive participle] is an unvocalized l which follows (talī) the verb (fiˁl) [i.e. the verbal stem] in the three tenses (ḥālāt) for all pronouns”.

ˁAlāmatuhu lām sākina talī al- fiˁl ṯalāṯ ḥālāt fī jamīˁ aḍ-ḍamāˀir. (Tuḥfa 47v5)

The Arabic examples given here, surprisingly, contain conjugated verbs with the internal passive rather than their passive participles, e.g. musiktu (‘I was taken’) and their Turkic equivalents with ‑(V)l‑, e.g.:

(4) wur-ġān
  hit-PART.PAST
  ‘the hitter’, ‘who was hit’
(5) wur-ul-ġān
  hit-PASS-PART.PAST
  ‘the one that is hit’
(6) ṭūṭ-ūl-dū-m
  take-PASS-PAST-1sg
  ‘I was taken’

3.3.2 K as a marker of passivity

In Turkic, there is a considerable number of deverbal adjectives ending in −(V)K15 (i.e. -ik, -uq, etc.).16 It makes sense to consider it a marker of passivity, albeit in a limited context, e.g. ač-uk ‘open’ (< ač- ‘open’), yar-uq ‘split’ (< yar- ‘to cleave’), oy-uq ‘hole’ (< oy- ‘to hollow out’), and yül-ük ‘shaven’ (< yül- ‘to shave’).17 The formation of these adjectives is, as far as I know, not productive.

These adjectives in -(V)K can be translated with passive participles in Arabic. This is the approach chosen in Tuḥfa. Tuḥfa (p. 48r) lists thirteen Arabic passive participles (of the mafˁūl pattern) along with their Turkic equivalents, all of them ending in -Vq/-Vk, e.g. maftūḥˀaǧ-īq ‘open’, mašqūqyār-īq ‘split’, maḥzūzkārt-īk ‘notched’ [< kert- ‘to notch’],18 suggesting that -(V)K is a marker of passivity. As we shall see, Ibn Muhannā too considers q to have this function (cf. 4.3.1), but he mistakes sin-uq- for a verbal stem.

Kāšġarī takes this same reasoning regarding the use of the variants of -(V)K one step further as he considers them verbal augments in relation to the passive voice of a verb (p. 328). He gives two exemplifying phrases with a verb in -(V)K which he translates with an Arabic internal passive. Note that in Turcological studies -(V)K is not considered a productive suffix with verbal stems.19

The Turkic verb [bassiq-], Kāšġarī writes, derives from an original form (ˀaṣl) bas-dī plus the ḥarf q. Bas- is, in fact, a common verb which is used in the sense of ‘to attack’.

In Kāšġarī’s view, the verb in 7b too consists of a stem (bal-) plus the suffix k/q. In Turcology though, baliq is not derived from the stem bal-, but from an obsolete stem ba:.20 Kāšġarī seems to acknowledge this, because he lists bāliġ as a noun elsewhere in his work, with the meaning ‘the wounded one’ [al-ǧarīḥ, p. 205; also p. 107 & 131]. Perhaps a more accurate translation of 7b therefore may be ‘the man was a wounded one’.

(7a) bass-iq-tī ˀar
  buyyita al-raǧulu
  ‘the man was suddenly attacked’
(7b) ˀar baliq-ti
  ǧuriḥa al-raǧul
  ‘the man was wounded’

3.3 Turkic l and n as markers of infiˁāl and muṭāwaˁa

The grammarians could not always decide whether the Turkic augments l/n stood for the internal passive or for other forms. It appears that they prefer the infiˁāl rather than the internal passive, possibly because this also involves the addition of a consonant.

ˀAbū Ḥayyān in his ˀIdrāk deals with the l/n both in the context of the internal passive, as I discussed above, but also in the compliance. He writes:

If the verb consists of one consonant (ˁalā ḥarf wāḥid), like their utterance ‘he broke (tr.)’ (kasara) si-dī, in the compliance (muṭāwaˁa) a silent n is used instead of an l. Thus for ‘he broke (intr.)’ (inkasara) you say si-n-dī.

Fa-in kāna al-fiˁl ˁalā ḥarf wāhid naḥwa qawlihim kasara si-dī fa-l-ḥarf allaḏī gīˀa bihi li-l- muṭāwaˁa nūn sākina badala al-lām fa-taqūlu fi inkasara si-n-dī. (ˀIdrāk 110)

He does not give any other conditions for the change.

On the same page, ˀAbū Ḥayyān, again, discusses this l under the heading of “consonant of the compliance” (ḥarf al-muṭāwaˁa). There he translates the Turkic passive verbal form 3.sg kas-il-dī with Arabic in-qataˁa, which we can analyse as follows:

It is difficult to tell at this point whether ˀAbū Ḥayyān here refers to the muṭāwaˁa in the semantical-interpretative sense, or whether he interprets all instances of infiˁāl as muṭāwaˁa throughout. The lack of any context in this phrase suggests that the latter may have been the case.

The author of the Margin Grammar is more explicit. According to him [in Turkic], no distinction is made between the “passive form” (mā lam yusamma fāˁiluhu) and the infiˁāl form: they overlap. In Turkic, he writes:

[1] There is no difference between this form [i.e. the form in fuˁila] and the infiˁāl-form; both are rendered by means of addition (ziyāda) of the l, regardless of whether it is in the past tense (māḍī) or the future tense (mustaqbal).

[2] You say for example ˀur-ul-dī i.e. ‘he was beaten’ (ḍuriba) and likewise aṣ-il-dī ‘he was hung’ (ṣuliba), and ˀič-il-dī ‘it was drunk’ (šuriba) and the like.

[3] The difference between al-infiˁāl and [the form fuˁila] is that al-infiˁāl is intransitive (lāzim) [while] this [the base form without ‑l‑] is transitive (mutaˁaddin). [...]

[4] the infiˁāl occurs with an n, you say ˀari-n-dī [‘he is cleansed’] or the infiˁāl-form of ‘the cleansing’ (al-naẓāfa), and likewise kur-u-n-dī [‘he is seen’] or the infiˁāl-form of ‘the staring’ (al-ˀibṣār).

[1] Bāb mā lam yusamma fāˁilu-hu − lā farqa bayna-hu wa-bayn al-infiˁāl fī ziyāda al-lām fī l-māḍī wa-l-mustaqbal; [2] taqūl min ḏālika ˀur-ul-dī ˀay ḍuriba wa-kaḏālika ˀaṣ-il-dī ˀay ṣuliba wa-kaḏālika ˀič-il-dī ˀay šuriba wa-naḥwa-hunna. [3] Wa-l-farq bayna al-infiˁāl wa-bayna-hā ˀanna al-infiˀāl lāzim wa-haḏā mutaˁaddin [...] [4] wa-qad yaˀtī al-infiˁāl bi-l-nūn fa-taqūlu ˀār-in-dī ˀay infaˁala min al-naẓāfa w-kaḏālika kur-un-dī ˀay infiˁāl min al-ˀibṣār. (MG 52A right)

The reference to the tenses is not without importance either. In Arabic, the past tense verb of the 3.m.sg is the standard form without any additional consonants. The present/future tense (muḍāriˁ or mustaqbal) is formed by means of an extra prefix and a change in the pattern of the stem, e.g. in ḍaraba/yaḍribu (‘he hit’/‘he hits or will hit’). In the Turkic verbal paradigm, these are only added to the verbal stem.

According to Kāšġarī, the effect of the insertion of n is that “the verb shifts from transitivity to intransitivity” (fa-lamma ˀadḫalta al-nūna yuqlabu al-fiˁl min al-taˁdiya ˀilā l-lāzim: (Dīwān 490).21 Kāšġarī suggests that both in Turkic and Arabic there is a similar morphological process in which the n causes intransivity:22

ˀar tukūn yaz-dī (ḥalla al-raǧul al-ˁuqda) ‘the man loosened the knot’ but then the n is attached and they say tukūn yaz-in-dī23 i.e. ‘the knot is loosened’ (inḥallat al-ˁuqda) and the verb has become intransitive because of the attachment of the n to it.

ˀAr tukūn yaz-dī ˀay ḥalla al-raǧulu al-ˁuqdata ṯumma yulḥaqu bi-hi al-nūn fa-yuqālu tukūn yaz-in-dī ˀay inḥallat al-ˁuqdatu fa-ṣāra al-fiˁl lāziman bi-ˀilḥāq al-nūn bihi. (Dīwān 490‑491)

Perhaps they maintain that the functions of n in Turkic and Arabic here coincide, i.e. namely passivization of transitive verbs.24

(8) in-qataˁa
  PASS.RESULT-cut-3m.sg.PAST
  ‘it was cut’ (ˀIdrāk 110, 12‑15)

3.3.1 Ibn Muhannā on the internal passive and infiˁāl

In his work Ḥilyat al-Insān, Ibn Muhannā writes a brief yet elaborate explanation of the passive forms. He announces a discussion of three items: “the fourth chapter on [1] the verb whose agent is not mentioned (fiˁl mā lam yusamma fāˁiluhu), i.e. the fuˁila form [2] the form infiˁāl and [3] the form tafaˁˁul25 (Ḥilyat 129). The patterns infiˁāl and tafaˁˁul are in Arab grammar often used in the context of compliance (muṭāwaˁa).

Then Ibn Muhannā proceeds with a description of four instances in which in Turkic an unvocalized l is added. His point of departure is formed by the contexts in which one of the passive forms is used in Arabic. The first of these is the internal passive (maǧhūl) in which the l is, in Ibn Muhannā’s terms, inserted (ˀadḫalta) to the “roots” (ˀuṣūl) of the verb, e.g.:

  • ˀaḫaḏaˀal-dī, ˀuḫiḏaˀal-il-dī* ‘he took’, ‘he was taken’;

  • ḍarabaˀur-dī, ḍuribaˀur-ul-dī, ‘he hit’, ‘he was hit’;

  • kasarasin-dur-dī, kusirasin-dur-.l-dī, ‘he broke’, ‘he was broken’.

In all of these examples, the Arabic equivalents have the internal passive. The Turkic forms differ from another. The choice for ˀal-il- is peculiar, for the regular passive form of ˀal- is, according to the rules, ˀal-in-, ˀal-il- being quite rare (Clauson 1972: 145). Further, sin-dur- is surprising too, because it is a causative form (-dur-) and the passivization process shown here (sin-dur-ul-) thus contains a cluster of one causative and one passive suffixes.

Secondly, Ibn Muhannā explains, the unvocalized l in Turkic occurs as a marker where in Arabic the verbal pattern infaˁala is used. He illustrates this with the following examples:

  • ṭahura ‘he was clean’ — ˀarī-dī, taṭahhara26ˀar-īl-dī ‘he was cleansed’;

  • ˁallaqa ‘he hung’ — ˀas-dī, taˁallaqaˀas-īl-dī ‘he was hung’;

  • farraqa ‘he separated’ — taġ-dī, tafarraqataġ-īl-dī ‘it was dispersed’.

Interestingly, while all Turkic forms indeed contain a passive in -Vl-, none of the Arabic examples are an actual illustration of the VII pattern (infaˁala). Instead, they are all V forms (tafaˁˁala). Perhaps Ibn Muhannā did not intend to refer literally to the VII infaˁala form but rather the notion of muṭāwaˁa often associated with this pattern.

In the third place, according to Ibn Muhannā, a q is used as a marker of passivity, e.g.:

  • kasara, s.n-dī ‘he broke’, takassara, s.n-uq-dī ‘it was broken’.

Yet a verbal stem s.n-uq-, as far as I know, does not exist; the form sın-uq is an adjective to which, in a regular procedure, a past tense ending can be added (Clauson 1972: 837). Moreover, while no doubt derived from the verb sı- ‘break’, the adjective already contains -n-, which denotes passivity. In proposing here q as a suffix, Ibn Muhannā either follows Kāšġarī (cf. 3.2.2), who also proposes q as a marker of passivity, albeit with a less adequately chosen example, or, alternatively, he has had access to sources used by Kāšġarī.

In a fourth statement, Ibn Muhannā remarks that instead (ˁiwaḍ) of the l and the q an unvocalized n can be used. The condition for using n is, he writes, that the preceding consonant is vocalized with an a (maftūḥ) or a u (maḍmūm). This same n also serves as the marker of the equivalent to the V tafaˁˁul pattern, the reflexive:

  • ġasalayū-dī ‘he washed’, taġassalayū-n-dī ‘he washed himself’

  • ḥarraka ‘he moved (tr.)’ — t.brā-dī ‘he moved (intr.)’, taḥarrakat.brā-n-dī ‘he (was) moved’.

Ibn Muhannā here appears to be saying that in regard to n, in Turkic no difference is made between the internal passive and the other passive forms. However, again, neither of the two examples he gives is an internal passive. In addition, his choice of the second Turkic example, t.brā-dī is not very adequate, since it is intransitive and clearly not the equivalent of the transitive verb ḥarraka.27

3.3.2 Kāšġarī on the combination of l and n

The suffixes l and n occasionally occur in combination with each other in Turkic verbs, e.g. yaz-l-in- ‘become loose’ and yuv-lu-n-28 ‘roll’. In the reasoning of the Arab grammatical tradition, doubling poses a problem, since these are both meaningful suffixes which essentially serve the same function. Although Kāšġarī (490‑411) does not mention this theoretical problem, he analyses the facts in some detail, in relation to the passive-reflexive verbs yaz-li-n- ‘become loose’ and yuv-lu-n- ‘roll (pass.)’ [< yuv- ‘to roll (trans.)’], which convey the same meaning as the simpler alternative passive forms yaz-il- and yaz-in- and yuv-ul-, respectively. In Kāšġarī’s analysis, many of the issues discussed above come together:

[1] Then the n is combined (turakkabu) with the l and they say yaz-li-n-dī, i.e. ‘the knot loosens by itself’ (inḥallat al-ˁuqdatu bi-ṭabˁihi) (sic).

[2] They also say ˀar tubuq yuv-dī29 ‘the man rolled the ball’ (daḥraǧa al-raǧul al-kurrata). Then they say tubuq yuv-ul-dī ‘the ball was rolled by the action of something else’ (duḥriǧat al-kurra bi-fiˁili ġayrihi). The same applies in case of [the Arabic verbal form] tadaḥraǧa ‘it rolled (intr.)’ [i.e. there is no implied agent]. Then the n is attached to it (yulḥaqu), and they say yuv-lu-n-dī, i.e. ‘it rolled by itself’ (tadaḥraǧa bi-ṭabˁihi).

[3] Before the attachment of the n to the l [i.e. yaz-il-], the verb was transitive (lāziman) in two aspects (waǧhayni). One of them30 was that the action affected it [i.e. the semantic object] (wāqiˁan ˁalayhi) through an unknown agent (fāˁil maǧhūl) and the verb follows the same course (maǧrā) as the l in it.31

[1] ṯumma turakkabu al-nūn maˁa al-lām fa-yuqāl yaz-li-n-dī ˀay inḥallat al-ˁuqdatu bi-ṭabˁihi [2] wa-kaḏālika yuqālu ˀar tubuq yuv-dī ˀay daḥraǧa al-raǧul al-kurrata. Ṯumma yuqālu tubuq yuv-ul-dī ˀay duḥriǧat al-kurra bi-fiˁli ġayrihi wa-kaḏālika ˀiḏā tadaḥraǧa ṯumma yulḥaqu bihi al-nūn fa-yuqālu [491] yuv-l-un-dī ˀay tadaḥraǧa bi-ṭabˁihi. [3] fa-qabla ˀilḥāq al-nūn bi-l-lām kāna al-fiˁl lāziman ˁalā waǧhayni, ˀaḥaduhumā kāna yaǧūzu ˀan yakūn al-fiˁlu wāqiˁan ˁalayhi min fāˁil maǧhūl fa-yaǧrī al-fiˁlu maǧrā al-lām fīhi.

What Kāšġarī appears to be saying in this section is that yuv-ul- is a passive form with an implied, hidden (or unknown) agent. Yet after the addition of the n, resulting in yuv-lu-n-, which contains a (vowel shift and a) combination of suffixes —impossible in Arabic—, the verb looses the notion of implied or hidden agent and the action is carried out by itself (bi-ṭabˁihi), expressed in Arabic by a passive-reflexive form such as tadaḥraǧa.32 Thus he not only distinguishes semantically and functionally the Arabic internal passive from the infiˁāl-form, he also assigns distinct functions to Turkic verbal morphemes.

3.4 The distinction between stem and root

While in Arabic the pre- and infixes are inserted before the verbal (or nominal) root called ˀaṣl33 and between its consonantal elements, in Turkic all suffixes are attached to an (almost) invariable stem and clustered to one another. Because of the distinct qualities, the Turkic stem cannot be equivalent to the Arabic root

In the Margin Grammar and in Qawānīn, as we have seen above, therefore the terms fiˁl al-ˀamr ‘imperative verb’ and fiˁl al-ˀamr al-muǧarrad ‘the bare imperative verb’ are used. This makes sense, because the Turkic bare stem, devoid of any suffixes, conveys the imperative. Other sources (Tuḥfa, Dīwān) use fiˁl for the verbal stem. Ibn Muhannā uses once ‘roots’ in relation to verbs (ˀuṣūl, see 3.3.1). In yet another context, the Margin Grammar uses al-ˀaṣl al-mufrad al-muǧarrad ‘the basic bare root’ (MG 36B top).

4 Conclusions

It is obvious that the Arab grammarians recognized the two Turkic suffixes, l and n, that are attached to the verbal stem in order to indicate the passive form. Yet there are different points of confusion as to the distribution of these suffixes. For example, for some reason they link the distribution of the n to mono- or biradical verbs.

Another problem is the signification. In Arabic, the passive can be expressed by means of an internal passive along the patterns /fuˁila/ —a change within the root (wazn)— or /mafˁuwl/, or via changes to the root, VII infaˁala or V tafaˁˁala each signifying different things. While the internal passive refers to an unknown, hidden agent, the VII and V forms refer to an absent agent, the subject of the verb carries the action out by itself. These forms are used in the concept of compliance, the resultative (muṭāwaˁa) in which an agent carries out the action, while there is a causing element. In Turkic no such differences exists and n can be assigned the same signification as the equivalent suffix in Arabic: intransitivization. Most sources take the functional overlap of l and n, when transferred to Arabic, for granted; only the Margin Grammar explicitly says that they coincide. It seems the authors are confused by this overlap; they would have preferred to assign the Turkic suffixes l/n distinct functions.

Primary sources

  • Al-Andalusī, ˀAbū Ḥayyān. Kitāb al-ˀIdrāk fī lisān al-ˀAtrāk. Ed. by A. Caferoǧlu. Istanbul: Evkaf. 1931. [Google Scholar]
  • Kitāb Manhaj al-sālik fī al-kalām ˀalā Alfīyyat Ibn Mālik. Ed. by S. Glazer. New Haven CT: American Oriental Society. 1947. [Google Scholar]
  • Irtišāf al-Ḍarab fī Lisān al-ˁArab, vol. I, II, III. Ed. by. M.A. al-Namās. Cairo: Al-Madanī. 1984, 1987, 1989. [Google Scholar]
  • Halasi-Kun, T. ed. Al-Tuḥfa al-Zakiyya fī al-Luġa al-Turkiyya / La langue des Kiptchaks d’après un manuscrit arabe d’Istanbul, II, Édition phototypique. Budapest: Bibliotheca Orientalis hungarica. 1942. [Google Scholar]
  • Houtsma, M. Th. ed. Tarǧumān Turkī wa-Aǧami wa-Muġālī / Ein türkisch-arabisches Glossar nach der Leidener Handschrift. Leiden: Brill. 1889. [Google Scholar]
  • Ibn Al-Muhannā, Ǧ.‑D. Kitāb Hilyat al-Insān wa-Halbat al-Lisān. Ed. by M.R. Kilisli. Istanbul: Matbaa-ı Âmire. 1921. [Google Scholar]
  • Al-Kāšġarī, M. Dīwān Luġāt at-Turk. Facsimile edition of Ali Emiri 4189. Ankara: Kültür Bakanlıǧı [Ministry of Culture]. 1990. [Google Scholar]
  • Kilisli, M.R. ed. Kitāb al-Qawānīn al-Kulliyya li-Ḍabt al-Luġa al-Turkiyya. Istanbul: Evkaf. 1928. [Google Scholar]
  • Abboud-Haggar, S. 2006. Dialects: Genesis. Encyclopedia of Arabic language and linguistics 1, 613-622. Leiden: Brill. [Google Scholar]
  • Al-Balushi, R. 2016. Omani Arabic: More than a dialect . Macrolinguistics, no 4/4: 80-125. [CrossRef] [Google Scholar]
  • Auezova, Z.‑A. 2005. Mahmud Al-Kashgari: Diwan Lughat at Turk. Kazakhstanskie Vostokovednye Issledovanii︠a︡. Almaty: Daik Press. [Google Scholar]
  • Ayoub, G. and Versteegh, K., eds. 2018. The foundations of Arabic linguistics, III, The development of a tradition: Continuity and change. Studies in Semitic languages and linguistics 94. Leiden: Brill. [Google Scholar]
  • Baalbaki, R. 2009. Aṣl. Encyclopedia of Arabic language and linguistics 4, 192-195. Leiden/Boston: Brill. [Google Scholar]
  • Bazzi-Hamzé, S. 2007a. Fāˁil. Encyclopedia of Arabic language and linguistics 2, 83-84. Leiden/Boston: Brill. [Google Scholar]
  • — 2007b. Fiˁl. Encyclopedia of Arabic language and linguistics 2, 90-96. Leiden/Boston: Brill. [Google Scholar]
  • Belova, A. Gr. 2009. South Semitic languages. Encyclopedia of Arabic language and linguistics 4, 300-315. Leiden/Boston: Brill. [Google Scholar]
  • Berta, Á. 1998. Middle Kipchak. In The Turkic languages, ed. by L. Johanson and É.Á. Csató, 158-166. London: Routledge. [Google Scholar]
  • Bohas, G. and Guillaume, J.‑P. 1984. Étude des théories des grammairiens arabes, 1, Morphologie et phonologie. PIFD 112. Damascus: Institut français de Damas. [Google Scholar]
  • Carter, M.G., ed. 1981. Arab linguistics: An introductory classical text with translation and notes. Amsterdam studies in the theory and history of linguistic science. Series 3, Studies in the history of linguistics 24. Amsterdam: J. Benjamins. [CrossRef] [Google Scholar]
  • Clauson, G. 1965. An Eastern Turki-English dictionary . By Gunnar Jarring. Lunds Universitets Årsskrift, N. F. Avd. 1, Bd. 56, Nr. 4. Pp. 338. C. W. K. Gleerup, Lund. 1964. Kr. 50. Journal of the Royal Asiatic Society of Great Britain and Ireland, no 97/1: 57. [CrossRef] [Google Scholar]
  • — 1972. An etymological dictionary of pre-thirteenth-century Turkish. Oxford: Clarendon Press. [Google Scholar]
  • Dankoff, R. and Kelly, J., eds. 1982-1985. Maḥmūd al-Kāšγarī, Compendium of the Turkic dialects (Diwān Luγāt at-Turk). 3 vols. Sources of Oriental languages and literatures; Turkish sources 7. Cambridge MA: Harvard University. [Google Scholar]
  • Ermers, R. 1999. Arabic grammars of Turkic: The Arabic linguistic model applied to foreign languages & translation of ˀAbū Ḥayyān al-ˀAndalusī’s Kitāb al-ˀIdrāk li-Lisān al-ˀAtrāk. Studies in Semitic languages and Linguistics 28. Leiden/Boston/Cologne: Brill. [Google Scholar]
  • — 2007. The use of morphological patterns in Arabic grammars of Turkic. In Approaches to Arabic linguistics presented to Kees Versteegh on the occasion of his sixtieth birthday, ed. by E. Ditters & H. Motzki, 435-453. Studies in Semitic Languages and Linguistics 49. Leiden/Boston: Brill. [Google Scholar]
  • Girod, A. 2007. Impersonal Verb. Encyclopedia of Arabic language and linguistics 2, 315-318. Leiden: Brill. [Google Scholar]
  • Johanson, L. 1998. The structure of Turkic. In The Turkic languages, ed. by L. Johanson and É.Á. Csató, 30-66. London: Routledge. [Google Scholar]
  • Khawla, M.F. 2012. Al-muṭāwaˁa wa-taˀṣīṣu-hā fī al-ˁArabiyya . Majallat Jāmiˁat Tikrīt Li-l-ˁUlūm al-Insāniyya, no 19/10: 122-135. [Google Scholar]
  • Larcher, P. 2003. Le système verbal de l’arabe classique. 1st ed. Didactilangue. Aix-en-Provence: Publications de l’université de Provence. [Google Scholar]
  • — 2009. Verbs. Encyclopedia of Arabic language and linguistics 4, 638-645. Leiden/Boston: Brill. [Google Scholar]
  • — 2012. Le système verbal de l’arabe classique. 2nd ed. Manuels. Aix-en-Provence: Presses universitaires de Provence. [Google Scholar]
  • Maalej, Z. 2008. Middle verbs. Encyclopedia of Arabic language and linguistics 3, 224-227. Leiden/Boston: Brill. [Google Scholar]
  • Maróth, M. 2009. Qiyās. Encyclopedia of Arabic language and linguistics 4, 11-14. Leiden/Boston: Brill. [Google Scholar]
  • An-Nādirī, M.A. 1995. Naḥw al-luġa al-ˁarabiyya: Kitāb fī qawāˁid an-naḥw wa-ṣ-ṣarf. Beirut: Al-Maktaba al-ˁAṣriyya. [Google Scholar]
  • Owens, J. 1988. The foundations of grammar: An introduction to Medieval Arabic grammatical theory. Amsterdam studies in the theory and history of linguistic science. Series 3, Studies in the history of the language sciences 45. Amsterdam/Philadelphia PA: J. Benjamins. [CrossRef] [Google Scholar]
  • — 1990.Early Arabic grammatical theory: Heterogeneity and standardization. Amsterdam studies in the theory and history of linguistic science. Series 3, Studies in the history of the language sciences 53. Amsterdam/Philadelphia PA: J. Benjamins. [Google Scholar]
  • Redhouse, J.W. 1890. A Turkish and English lexicon. Shewing in English the significations of the Turkish terms (Re-pr. 1978). Istanbul: Çaǧrı Yayınları. [Google Scholar]
  • Róna-Tas, A. 1998. Proto-Turkic and the Genetic question. In The Turkic languages, ed. by L. Johanson and É.Á. Csató, 67-80. London: Routledge. [Google Scholar]
  • Saad, G.N. 2019 [1982]. Transitivity, causation and passivization: A semantic-syntactic study of the verb in Classical Arabic. Library of Arabic Linguistics 4. New York: Routledge. [CrossRef] [Google Scholar]
  • Şahan Güney, F. 2006. Functions of the so-called passive morpheme −(I)l- in Kazakh . Bilig, no 36: 125-137. [Google Scholar]
  • Sheyhatovitch, B. 2018. The distinctive terminology in Šarḥ al-Kāfiya by Raḍī l-Dīn al-ˀAstarābāḏī. Studies in Semitic languages and linguistics 96. Leiden: Brill. [Google Scholar]
  • Simeone-Senelle, M.‑Cl. 1997. The Modern South Arabian languages. In The Semitic languages, ed. by R. Hetzron, 378-423. London: Routledge. [Google Scholar]
  • Soltan, U. 2009. Transitivity. Encyclopedia of Arabic language and linguistics 4, 535-542. Leiden/Boston: Brill. [Google Scholar]
  • Versteegh, K. 1995. The explanation of linguistic causes: Al-Zaǧǧāǧī’s theory of grammar: introduction, translation, commentary. Amsterdam studies in theory and history of linguistic science 75. Amsterdam/Philadelphia PA: John Benjamins. [Google Scholar]
  • — 1997. Landmarks in linguistic thought III: The Arabic linguistic tradition. Routledge History of Linguistic Thought Series. London/New York: Routledge. [Google Scholar]
  • — 2014. The Arabic Language. 2nd ed. Edinburgh: Edinburgh University Press. [Google Scholar]
  • Yavrumyan, M. 2006. Das System der Verbalstämme in der Arabischen linguistischen Tradition. Elemente der morphologischen und semantischen Analyse. Bayreuth: Universität Bayreuth. [Google Scholar]

Secondary sources

  • Al-Andalusī, ˀAbū Ḥayyān. Kitāb al-ˀIdrāk fī lisān al-ˀAtrāk. Ed. by A. Caferoǧlu. Istanbul: Evkaf. 1931. [Google Scholar]
  • Kitāb Manhaj al-sālik fī al-kalām ˀalā Alfīyyat Ibn Mālik. Ed. by S. Glazer. New Haven CT: American Oriental Society. 1947. [Google Scholar]
  • Irtišāf al-Ḍarab fī Lisān al-ˁArab, vol. I, II, III. Ed. by. M.A. al-Namās. Cairo: Al-Madanī. 1984, 1987, 1989. [Google Scholar]
  • Halasi-Kun, T. ed. Al-Tuḥfa al-Zakiyya fī al-Luġa al-Turkiyya / La langue des Kiptchaks d’après un manuscrit arabe d’Istanbul, II, Édition phototypique. Budapest: Bibliotheca Orientalis hungarica. 1942. [Google Scholar]
  • Houtsma, M. Th. ed. Tarǧumān Turkī wa-Aǧami wa-Muġālī / Ein türkisch-arabisches Glossar nach der Leidener Handschrift. Leiden: Brill. 1889. [Google Scholar]
  • Ibn Al-Muhannā, Ǧ.‑D. Kitāb Hilyat al-Insān wa-Halbat al-Lisān. Ed. by M.R. Kilisli. Istanbul: Matbaa-ı Âmire. 1921. [Google Scholar]
  • Al-Kāšġarī, M. Dīwān Luġāt at-Turk. Facsimile edition of Ali Emiri 4189. Ankara: Kültür Bakanlıǧı [Ministry of Culture]. 1990. [Google Scholar]
  • Kilisli, M.R. ed. Kitāb al-Qawānīn al-Kulliyya li-Ḍabt al-Luġa al-Turkiyya. Istanbul: Evkaf. 1928. [Google Scholar]
  • Abboud-Haggar, S. 2006. Dialects: Genesis. Encyclopedia of Arabic language and linguistics 1, 613-622. Leiden: Brill. [Google Scholar]
  • Al-Balushi, R. 2016. Omani Arabic: More than a dialect . Macrolinguistics, no 4/4: 80-125. [CrossRef] [Google Scholar]
  • Auezova, Z.‑A. 2005. Mahmud Al-Kashgari: Diwan Lughat at Turk. Kazakhstanskie Vostokovednye Issledovanii︠a︡. Almaty: Daik Press. [Google Scholar]
  • Ayoub, G. and Versteegh, K., eds. 2018. The foundations of Arabic linguistics, III, The development of a tradition: Continuity and change. Studies in Semitic languages and linguistics 94. Leiden: Brill. [Google Scholar]
  • Baalbaki, R. 2009. Aṣl. Encyclopedia of Arabic language and linguistics 4, 192-195. Leiden/Boston: Brill. [Google Scholar]
  • Bazzi-Hamzé, S. 2007a. Fāˁil. Encyclopedia of Arabic language and linguistics 2, 83-84. Leiden/Boston: Brill. [Google Scholar]
  • — 2007b. Fiˁl. Encyclopedia of Arabic language and linguistics 2, 90-96. Leiden/Boston: Brill. [Google Scholar]
  • Belova, A. Gr. 2009. South Semitic languages. Encyclopedia of Arabic language and linguistics 4, 300-315. Leiden/Boston: Brill. [Google Scholar]
  • Berta, Á. 1998. Middle Kipchak. In The Turkic languages, ed. by L. Johanson and É.Á. Csató, 158-166. London: Routledge. [Google Scholar]
  • Bohas, G. and Guillaume, J.‑P. 1984. Étude des théories des grammairiens arabes, 1, Morphologie et phonologie. PIFD 112. Damascus: Institut français de Damas. [Google Scholar]
  • Carter, M.G., ed. 1981. Arab linguistics: An introductory classical text with translation and notes. Amsterdam studies in the theory and history of linguistic science. Series 3, Studies in the history of linguistics 24. Amsterdam: J. Benjamins. [CrossRef] [Google Scholar]
  • Clauson, G. 1965. An Eastern Turki-English dictionary . By Gunnar Jarring. Lunds Universitets Årsskrift, N. F. Avd. 1, Bd. 56, Nr. 4. Pp. 338. C. W. K. Gleerup, Lund. 1964. Kr. 50. Journal of the Royal Asiatic Society of Great Britain and Ireland, no 97/1: 57. [CrossRef] [Google Scholar]
  • — 1972. An etymological dictionary of pre-thirteenth-century Turkish. Oxford: Clarendon Press. [Google Scholar]
  • Dankoff, R. and Kelly, J., eds. 1982-1985. Maḥmūd al-Kāšγarī, Compendium of the Turkic dialects (Diwān Luγāt at-Turk). 3 vols. Sources of Oriental languages and literatures; Turkish sources 7. Cambridge MA: Harvard University. [Google Scholar]
  • Ermers, R. 1999. Arabic grammars of Turkic: The Arabic linguistic model applied to foreign languages & translation of ˀAbū Ḥayyān al-ˀAndalusī’s Kitāb al-ˀIdrāk li-Lisān al-ˀAtrāk. Studies in Semitic languages and Linguistics 28. Leiden/Boston/Cologne: Brill. [Google Scholar]
  • — 2007. The use of morphological patterns in Arabic grammars of Turkic. In Approaches to Arabic linguistics presented to Kees Versteegh on the occasion of his sixtieth birthday, ed. by E. Ditters & H. Motzki, 435-453. Studies in Semitic Languages and Linguistics 49. Leiden/Boston: Brill. [Google Scholar]
  • Girod, A. 2007. Impersonal Verb. Encyclopedia of Arabic language and linguistics 2, 315-318. Leiden: Brill. [Google Scholar]
  • Johanson, L. 1998. The structure of Turkic. In The Turkic languages, ed. by L. Johanson and É.Á. Csató, 30-66. London: Routledge. [Google Scholar]
  • Khawla, M.F. 2012. Al-muṭāwaˁa wa-taˀṣīṣu-hā fī al-ˁArabiyya . Majallat Jāmiˁat Tikrīt Li-l-ˁUlūm al-Insāniyya, no 19/10: 122-135. [Google Scholar]
  • Larcher, P. 2003. Le système verbal de l’arabe classique. 1st ed. Didactilangue. Aix-en-Provence: Publications de l’université de Provence. [Google Scholar]
  • — 2009. Verbs. Encyclopedia of Arabic language and linguistics 4, 638-645. Leiden/Boston: Brill. [Google Scholar]
  • — 2012. Le système verbal de l’arabe classique. 2nd ed. Manuels. Aix-en-Provence: Presses universitaires de Provence. [Google Scholar]
  • Maalej, Z. 2008. Middle verbs. Encyclopedia of Arabic language and linguistics 3, 224-227. Leiden/Boston: Brill. [Google Scholar]
  • Maróth, M. 2009. Qiyās. Encyclopedia of Arabic language and linguistics 4, 11-14. Leiden/Boston: Brill. [Google Scholar]
  • An-Nādirī, M.A. 1995. Naḥw al-luġa al-ˁarabiyya: Kitāb fī qawāˁid an-naḥw wa-ṣ-ṣarf. Beirut: Al-Maktaba al-ˁAṣriyya. [Google Scholar]
  • Owens, J. 1988. The foundations of grammar: An introduction to Medieval Arabic grammatical theory. Amsterdam studies in the theory and history of linguistic science. Series 3, Studies in the history of the language sciences 45. Amsterdam/Philadelphia PA: J. Benjamins. [CrossRef] [Google Scholar]
  • — 1990.Early Arabic grammatical theory: Heterogeneity and standardization. Amsterdam studies in the theory and history of linguistic science. Series 3, Studies in the history of the language sciences 53. Amsterdam/Philadelphia PA: J. Benjamins. [Google Scholar]
  • Redhouse, J.W. 1890. A Turkish and English lexicon. Shewing in English the significations of the Turkish terms (Re-pr. 1978). Istanbul: Çaǧrı Yayınları. [Google Scholar]
  • Róna-Tas, A. 1998. Proto-Turkic and the Genetic question. In The Turkic languages, ed. by L. Johanson and É.Á. Csató, 67-80. London: Routledge. [Google Scholar]
  • Saad, G.N. 2019 [1982]. Transitivity, causation and passivization: A semantic-syntactic study of the verb in Classical Arabic. Library of Arabic Linguistics 4. New York: Routledge. [CrossRef] [Google Scholar]
  • Şahan Güney, F. 2006. Functions of the so-called passive morpheme −(I)l- in Kazakh . Bilig, no 36: 125-137. [Google Scholar]
  • Sheyhatovitch, B. 2018. The distinctive terminology in Šarḥ al-Kāfiya by Raḍī l-Dīn al-ˀAstarābāḏī. Studies in Semitic languages and linguistics 96. Leiden: Brill. [Google Scholar]
  • Simeone-Senelle, M.‑Cl. 1997. The Modern South Arabian languages. In The Semitic languages, ed. by R. Hetzron, 378-423. London: Routledge. [Google Scholar]
  • Soltan, U. 2009. Transitivity. Encyclopedia of Arabic language and linguistics 4, 535-542. Leiden/Boston: Brill. [Google Scholar]
  • Versteegh, K. 1995. The explanation of linguistic causes: Al-Zaǧǧāǧī’s theory of grammar: introduction, translation, commentary. Amsterdam studies in theory and history of linguistic science 75. Amsterdam/Philadelphia PA: John Benjamins. [Google Scholar]
  • — 1997. Landmarks in linguistic thought III: The Arabic linguistic tradition. Routledge History of Linguistic Thought Series. London/New York: Routledge. [Google Scholar]
  • — 2014. The Arabic Language. 2nd ed. Edinburgh: Edinburgh University Press. [Google Scholar]
  • Yavrumyan, M. 2006. Das System der Verbalstämme in der Arabischen linguistischen Tradition. Elemente der morphologischen und semantischen Analyse. Bayreuth: Universität Bayreuth. [Google Scholar]

1

The author wishes to thank the anonymous reviewer for his critical remarks.

2

The importance of Dīwān is so great, that most Turkic peoples, from Turkey to Kazakhstan, claim it as part of their cultural heritage. In addition, Dīwān formed the basis for Clauson’s etymological dictionary of Turkic languages (Clauson 1972).

3

See, e.g. Yavrumyan (2006).

4

/”/ represents the ˀalif, which is preceded by /a/ (fatḥa), which then can only be realised as a long vowel (see Bohas and Guillaume 1984: 256‑259).

5

It remains to be investigated whether they discerned new functions that were unknown to the Arabic system.

6

Khawla (2012: 129‑130) lists eleven verbal patterns for the muṭāwiˁ, one of them, in fuˁila, as in ǧadaˁahu fa-ǧudiˁa ‘he deceived him, so he was deceived’.

7

Clauson (1972) notes the rare form al-ıl- as a passive of al-; also bil-il- ‘be known’ (< bil-) (Berta 1998: 160).

8

In the sources, the vowels in Turkic morphemes are occasionally not explicitly written; in those cases, a period is used in the transcription, e.g. –.l- and –.n-.

9

On al-nāˀib ˁan al-fāˁil, cf. Bazzi-Hamzé (2007a: 82).

10

Correction for ‘noun’: R.E.

11

Clauson (1972: 57).

12

Em. by the editor.

13

The variant ye-n- does exist in the same meaning. The Turkic verb yaŋ- (pronounced with front vowels as [yeŋ-] ‘to beat, conquer’, yeŋ- [Clauson 1972]) has a regular passive form, i.e. yaŋ-il- ‘be conquered’. In Ottoman Turkish, the two verbs yen- and yeŋ- have merged into yenmek ‘to overcome’, ‘to be eaten’ (Redhouse 1978 [1890]).

14

Qawānīn repeats this same text almost literally when describing how the passive participle is formed (p. 51; see discussion in 3.2.1).

15

K represents the morphemes k or q depending on whether the word is pronounced back or front.

16

Note that in Clauson’s transcription of Turkic −q is not used. Whether a word is pronounced back or front is to be inferred from the vowels.

17

Cf. Clauson (1972: 22, 962, 270 & 928).

18

Cf. Clauson (1972: 22, 962 & 738).

19

According to Clauson (1972: 337), Kāšġarī here mistakes the -q for the suffix -siq.

20

Cf. ibid.

21

If we take this remark to also be valid for verbs with −il-, this is not entirely true, because there are verbs without ‑l‑ that are transitive, e.g. igle-l-, (< igle: ‘be sick’); a disease igle-l-di ‘was suffered’ (Clauson 1972: 107).

22

Kāšġarī elsewhere remarks that the n in general causes intransitivity, e.g. for the medio-passive or reflexive verbs (596‑597).

23

Correction for yaz-li-n-dī, which Kāšġarī mentions as an alternative later on in the text.

24

A step further would be the suggestion that Kāšġarī believed that -n has a cross-linguistic signification and that it is the same morpheme in the two languages.

25

Numbers added: R.E.

26

Note that taṭahhara is the passive form to ṭahhara ‘cleanse’, not ṭahura, as Ibn Muhannā seems to be asserting here.

27

Hence its causative form tepre-t-, cf. Clauson (1972).

28

Cf. ibid.: 987.

29

On yuv- cf. ibid.: 873.

30

Contrary to what one would expect here, a second aspect is not mentioned.

31

On maǧrā, see Maróth (2009: 13).

32

Kāšġarī adds in a subsequent passage on the same page that tetra-radical (rubāˀī) verbs that are the result of a procedure of combining, such as yuv-lu-n-dī, are transferred (manqūla) from a bi-radical (ṯunāˀī) verb (yuv-) to tri-radical (ṯulāṯī), and from tri-radical to tetra-radical.

33

On ˀaṣl, cf. Baalbaki (2009: 191); Bohas and Guillaume (1984).