A Typological Analysis of Cognate Infinitives in Lebanese Arabic Based on Comparative Semitic Evidence

Abstract

Despite the relatively scarce literature on the topic and the lack of terminological consensus among scholars, Cognate Infinitives (CI) have been identified to share formal and functional characteristics across Semitic. The present study provides a description of the formal features of Cognate Infinitives in Lebanese Arabic (LA) based on the analysis of linguistic data gathered through a participant observation method. The novelty of this description lies in its comparative approach, which has been developed in the light of the Semitic evidence available, gathered through a review of the main literature available on the topic. The results of this comparative analysis reveal that the grammatical features of Cognate Infinitives in Lebanese Arabic seem to be in line with general Semitic trends that do not, however, always find their parallel in prescriptive descriptions of Cognate Infinitives in Classical or Standard Arabic.

1. Introduction

The existence of Cognate Infinitives within the Semitic continuum has been noted for centuries by various scholars, on occasion resulting in the creation of seminal studies on the topic (Goldenberg 1971; Kim 2009).

However, these scholars’ attempts to describe the formal and functional nature of this linguistic feature did not always lead to a consensus as far as terminology is concerned. Different grammatical approaches brought about many distinct nomenclatures for one single linguistic form: mafʕūl muṭlaq mubham in Classical Arabic (Al-Zamaxšarī 1870, p. 111); Paronymous Complement in Syrian Arabic (Cowell 1964); Unmodified Cognate Complement in Rural Palestinian Arabic (Shachmon and Marmorstein 2018); Tautological Infinitive in Biblical Hebrew (Goldenberg 1971); Infinitive Absolute in Syriac (Nöldeke 2003); Paronomastic Infinitive in Akkadian (Cohen 2004), etc.

Nevertheless—and notwithstanding the lack of agreement in terminology surrounding Cognate Infinitive constructions— if we are able today to group together this myriad of grammatical labels, it is only because both formal and functional characteristics of Cognate Infinitives seem to be clear enough to be described, even across Semitic languages.

At the formal level, a Cognate Infinitive construction (CI) is formed by two essential elements: (1) a finite verbal form that functions as the lexical head of a predicate (or ‘cognate head’) and (2) an infinitive that depends syntactically on and is cognate with the verbal head and stands indefinite and unqualified (‘cognate infinitive’). This makes CIs different from Cognate Object constructions (CO), which, albeit similar, present an “infinitive” that is specified, modified, or qualified in a variety of ways.

The distinction between CIs and COs throughout most of the Semitic literature also extends to functional grounds. While COs are known to modify the verb adverbially, the function of CIs has been often described with vague notions such as ‘emphasis’, ‘asseveration’, ‘contrast’, or ‘intensification’.

Within the Arabic grammatical tradition, these two concepts have been traditionally studied as two facets of one grammatical category: al-mafʕūl al-muṭlaq. The combined analysis of CIs and COs in Classical Arabic, which strongly influenced subsequent analyses in Modern Standard Arabic (MSA), seems to neglect the abundant Semitic evidence of analogous constructions that draw a clear grammatical line between these two structures.

The present study argues for the separation of these two types of cognate structures following Semitic treatments, and takes a first step in this direction by describing the formal features of Cognate Infinitives in Lebanese Arabic (LA), based on the analysis of abundant linguistic data gathered though a participant observation method during a period of four years. The novelty of this description lies in its comparative approach, for it was developed in light of the Semitic evidence available, gathered through a review of the literature available on the subject.

Given the scarcity of descriptions of CIs in non-Standard Arabic varieties, adopting a traditional approach for the analysis of the data could misleadingly urge us to read the grammatical features of CIs in LA as ‘exceptions’ to or ‘simplifications’ of the Standard norm, since this is the only one that has been thoroughly documented. However, when analyzed along comparative Semitic evidence, the data results reveal that the grammatical features of CIs in LA seem to be in line with general Semitic trends that, interestingly enough, do not always find their parallel in prescriptive descriptions of CIs in Classical or Standard Arabic.

By using the CIs in LA as a case study, this study attempts to shed light on the general benefits of a cross-Semitic analysis in the study of Arabic linguistics. As it was the case with CIs, the comparative Semitic evidence available may very well help researchers elucidate a broader, more inclusive vision of the grammatical nature of specific linguistic features, and assist them in the challenging task of revisiting traditional classifications to ensure accuracy in typological descriptions of Arabic varieties.

2. Materials and Methods

The present paper is based on an extensive study of CIs in LA based on a corpus of 133 recorded instances collected over 4 years of participant observation in Lebanon for the author’s doctoral dissertation (Iriarte Díez, forthcoming)1 as well as on a thorough review of the literature on CIs in Semitic languages.

Participant observation is the main method used by anthropologists. Fieldwork through this method requires “active looking, improving memory, informal interviewing, writing detailed field notes, and perhaps most importantly, patience” (DeWalt and DeWalt 2002, p. vii). The author’s long stay in Lebanon (almost 10 years) facilitated the ‘prolonged engagement’ (Lincoln and Guba 1985) necessary to the creation of the CI corpus. This method was chosen for being the only one able to face the methodological challenges that the study of CIs presented, namely, (1) their marked interactional and emotional nature, (2) the variety of linguistic, social, and communicative contexts in which they occur, and (3) the extreme difficulty of eliciting them2. Furthermore, as in previous studies, participant observation considerably diminished the effects of the so-called observer’s paradox (Labov et al. 1968; Milroy 1977).

Comparison is essential for any researcher to acquire a real and profound knowledge of the true nature of a linguistic feature. In fact, comparison is not only useful “to relativize a phenomenon that we tend to consider as outstanding” but also necessary “to understand the role of the specific grammar of a dialect in leading to a type of evolution” (Ibrahim 2011, p. 128). In this spirit, I carried out a thorough review of the relatively scarce literature on CIs. The purpose of this review was to gain awareness of the formal and functional variation that CIs show along the Arabic and Semitic continuums in order to be able to better evaluate (1) the morphological and syntactical factors that effectively differentiated CIs from COs, and (2) the pragmatic and discursive factors that could hold the key to a deeper understanding of the CI’s function.

As for the data presentation, for the purposes of this paper, the LA examples gathered in the corpus will be numbered and preceded by the letters LA (Lebanese Arabic) (i.e., [LA.n]). This system will help the reader differentiate CI and CO examples in LA from instances in other Semitic languages—which have been kept as they appear in the source reference and will be numbered and preceded by their corresponding abbreviation (e.g., [BH.3]). The abbreviations used for the different Semitic languages and Arabic varieties are displayed in the following

Table 1. Data tag abbreviations.

Abbrev.	Variety	Abbrev.	Variety
AKK	Akkadian	MEH	Mehri
BH	Biblical Hebrew	MSA	Modern Standard Arabic
CA	Classical Arabic	NENA	Northeastern Neo-Aramaic
CRA	Christian Arabic3	OA	Omani Arabic
EA	Egyptian Arabic	PH	Phoenician
EB	Eblaite	RPA	Rural Palestinian Arabic
JA	Jordanian Arabic	SSA	Sason Arabic
LA	Lebanese Arabic	SYR	Syriac
MAL	Maltese	UG	Ugaritic

.

3. Results

The present section provides a preliminary description of the formal features of CIs in LA in comparison to those of analogous forms in different Semitic languages. Its aim is to illustrate the formal variation that this construction shows across varieties.

As previously mentioned, in Lebanese Arabic, Cognate Infinitive constructions are formed by a finite verbal form that functions as the lexical head of a predicate (‘cognate head’ or CH)4 and a less finite verbal form (usually an infinitive) that depends syntactically on and is cognate with the cognate head and stands indefinite and unqualified (‘cognate infinitive’ or CI).

[LA.1] Lebanese Arabic

baram-ət	barəm	əs-siyāra
PFV.circle-3FS.CH	circle.INF.CI	DET-car

‘The car [really] spun (lit. *The car circled circling)’

In contrast, Cognate Object constructions, present an “infinitive” that is specified, modified or qualified in a variety of ways (cognate object or CO).

[LA.2] Lebanese Arabic

(a)	baram-ət	barm-e	əs-siyāra
	PFV.circle-3FS.CH	circle-NSI.CO	DET-car

‘The car toured once (lit. The car toured one tour)’

(b)	əs-siyāra	baram-ət	barm-e	sarīʕ-a
	DET-car	PFV.circle-3FS	circle-NSI	fast-F.S

‘The car did a quick tour (lit. The car toured a long tour)’

(c)	əs-siyāra	baram-ət	barm-et	əl- ʕarūs
	DET-car	PFV.circle-3FS	circle-NSI	DET-bride

‘The car took a long detour (lit. The car circled the circle of the bride)’

In [LA.2], the suffix –a(t)/-e(t), which in Arabic may be used to form a noun of single instance—also called nomen vicis, or ism al-marra—modifies the CH to indicate that the action has taken place once. Cognate nouns of single instance in Lebanese Arabic are often qualified, as in [LA.2b], where the noun with the adjective ‘fast’ modifies the verb adverbially, explaining how the action took place. The noun of single instance may also be made definite by a genitive construction or iḍāfa, as example [LA.2c] illustrates.

Both CIs and COs in LA are nominal(ized) elements cognate with a verbal form and canonically stand after the verb. However, CI constructions are formed with a plain infinitive (indefinite and unqualified), expressing some sort of ‘emphasis’ while CO constructions make use of nouns of a single instance—generally marked by the suffix –a(t)/-e(t)—that appear often qualified and modify the verb adverbially. This indicates that CIs and COs are distinct grammatical forms, both on formal and functional grounds.

The following subsections will address, in more detail, some of the formal features of CIs in LA in light of Semitic evidence. The morphological features include form and pattern of CIs and pattern correspondence between CIs and CHs. The syntactic features include CIs’ syntactic case, CIs’ position in the sentence, and the presence of enclitics.

3.1. Morphological Features of CIs in LA in the Light of Semitic Evidence

3.1.1. Infinitival Form of CIs

I have previously illustrated the formal differentiation between CIs and COs in LA as far as the choice of the ‘infinitival form’ is concerned. When looking at CI instances in other Semitic languages, the data consistently show a choice of morphological infinitives,5 which are formally and functionally distinct from the cognate nouns used in CO constructions (the latter being generally identifiable by an -a/-e(t) ending)).

Table 2. CI and CO instances in different Semitic varieties.

CI Instances	CO Instances
[SYR.1] Syriac meštaq [CI] šteq-w ‘They were completely silent’ (Robinson and Coakley 2013, p. 66)	[SYR.2] Syriac mít [CH] mawtā bíšā wa-mṭarpā [CO] ‘He died an evil and painful death’ (Sim. 333, 3 from Nöldeke 2003, p. 237)
[MAL.1] Maltese joħrog [CH] ħrug [CI] ‘he goes out extensively’ (Maas 2005, p. 416)	[MAL.2] Maltese għajtu [CH] għajta ta’ ferħ [CO] ‘They shouted a shout of joy’ (Sutcliffe 1960, p. 169)
[MEH.1] Mehri yisḥōṭ [CH] ḥābū saḥṭ [CI] (lit. he slaughters people slaughter) ‘he absolutely slaughters people [with his prices]’ (Watson 2012, p. 215 )	[MEH.2] Mehri ḳaṭays [CH] mən ḳaṭāt ḳənnət [CO] (lit. he cut her a little cut) ‘he cut her lightly’ (Rubin 2010, p. 219)
[AKK.1] Akkadian [š]a ta[q]biam/ana fPN/ana ešrīšu aqbīšim-ma/apālum-ma [CI] ul īpulanni [CH] “[Wh]at you to[l]d me I told fPN ten times but answer me she did not” (AbB 10, 8:16-19 from Cohen 2004, p. 107)	[AKK.2] Akkadian mīnam ēpuškāma ḫa-lu-qám ra-bi-a-am [co] tuḫallaqanni [ch] ‘what have I done to you, that you are completely ruining me’ [lit. *that you are ruining me a big ruining] (Kouwenberg 2017, p. 653)
[BH.1] sāqôl [CI] yissāqel [CH] haššôr “the ox shall be stoned” (Exod. 21:28 from Van der Merwe et al. 1999, p. 159)	[BH.2] way-yęḥęraḏ [CH] Yiṣḥāq ḥarāḏâ gdolâ [CO] “And Isaac trembled a very great trembling” (Gen. 27:33 from Goldenberg 2013, p. 295)

illustrates this formal distinction. In the table, both CIs and COs are marked in bold and cognate heads are underlined.

In Biblical Hebrew (BH), not only is a morphological distinction made between CIs and COs, but the infinitive used for CI constructions is also a special form of infinitive called ‘Infinitive Absolute’, which contrasts with the Infinitive Construct6. The distinction between these two forms is both morphological and syntactic7.

In contrast to BH, Classical and Modern Standard Arabic seem to be the most outstanding exception to the Semitic constant that morphologically distinguishes between infinitival forms for CIs and COs. In these varieties, both COs [CA.1] and CIs [CA.2] may be formed with a maṣdar (infinitive), although the use of a noun of single instance (NSI) in CO constructions is also accepted [CA.3]:

[CA.1] Classical Arabic (CI instance)

qumtu	qiyām-an
(PFV-stand.1S	stand.INF-ACC)

[lit. I stood standing]

(Al-Zamaxšarī 1870, p. 111; my glossing)

[CA.2] Classical Arabic (CO instance with maṣdar)

qumtu	qiyām-an	ṭawīl-an
(PFV-stand.1S	stand.INF-ACC	long.M.S-ACC)

[lit. I stood a long standing]

(Al-Zamaxšarī 1870, p. 111; my glossing)

[CA.3] Classical Arabic (CO instance with NSI)

qaʕada	qaʕda-ta	al-qurfuṣāʔ
(PFV-sit.3MS	sitting-NSI	DET-squatting position)

[He squatted]

(Sībawayhi: 112; my glossing)

In fact, the shared use of maṣdar between CIs and COs was in all likelihood one of the main factors (along with syntactic case8) that led traditional Arabic grammarians to group these two phenomena under one single grammatical category, at variance with what we normally find in the descriptions of Semitic languages.9

3.1.2. Pattern Correspondence between CIs and CHs

In Semitic, there is in general a high degree of correspondence between the verbal pattern of the CH and the corresponding CI. The majority of the studies on the subject (Goldenberg 1971; Cohen 2004, 2006; Kouwenberg 2010) highlight the existence of an exact pattern correspondence between CHs and CIs in Akkadian. Kim (2006, p. 197) corroborates this fact by compiling 228 examples of CIs throughout the different stages of the Akkadian language, and Finet affirms that this is also true as well for the Mari dialect (Finet 1952, pp. 21–22). On the other hand, Syriac and BH follow different pattern correspondences, similar to those of Lebanese Arabic.

As for Classical Arabic, Talmon’s study on mafʕūl muṭlaq occurrences in the Qur’ān shows that 61/64 of the maṣādir appearing in CI constructions10 share both the root and the pattern with their governing verb, reaching an ‘almost perfect’ pattern correspondence (Talmon 1999).

CIs in Lebanese Arabic generally share their pattern with their CHs when said heads are in pattern I (faʕal), II (faʕʕal), III (fāʕal), and X (istafʕal). In the cases of those patterns that carry passive, reflexive, or reciprocal values, such as V (tfaʕʕal) VI (tfāʕal), VII (nfaʕal), or VIII (ftaʕal), CHs take the cognate infinitive of their corresponding active pattern. The following examples illustrate pattern correspondence between CHs in patterns VII and V with their CIs in their active counterpart: Patterns I and II, respectively.

[LA.3]  Lebanese Arabic

kēn-ət	mēšy-e	l-ʕilēʔa	bas	baʕdēn
was.3FS	walk.PTCP.ACT.-FS	DET-relation	but	afterwards

ma	b-aʕrəf	šu	ṣār	[pause]
neg	HAB-IPFV.I.know	what	happened.3MS

ʔənʔaṭaʕ-ət	ʔaṭəʕ
was.cut-3FS [CH in pattern VII]	cut.INF [CI in pattern I]

‘The relationship was going [well] but then, I don’t know what happened… it [suddenly] broke off’

[LA.4]  Lebanese Arabic

ʔəza	ši	nhār	šəftī-h	hōn	ma
if	some	day	PFV.3MS.see-him	here	neg

txāfe	b-yətmaššā	təmšēye
IPFV.2FS.fear	HAB-IPFV.3MS.stroll [CH in pattern V]	stroll.inf [CI in pattern II]

huwwe	w-əl-kaləb	tabaʕ-o	la-yrēʔəb	əl-bnūke	bas
he	and-DET-dog	GEN-him	to-IPFV.3MS.guard	DET-banks	only

‘If you see him some other day, don’t be afraid… He just patrols with his dog to guard the banks [of the area], nothing else’

Detailed pattern correspondence between CIs and CHs for the verbs in my LA data is illustrated in

Table 3. Pattern correspondence between CIs and CHs in LA and percentage of occurrence in the corpus.

CH’s Pattern	CI’s Pattern	Example	% of Occurrences
I	I	maʔṭūše ʔaṭəš	67.67%
II	II	msaṭṭaltəsṭīl	14.29%
III	III	tsēʕdūne msēʕade	0.75%
V	II	btətlaxxaṣ təlxīṣ	2.26%
VI	III	txēnaʔo xnēʔ	0.75%
VII	I	byənmašā maše	3.01%
VIII	I	nəštəġəla šəġəl	0.75%
X	X	staʔbalto stiʔbēl	0.75%
QI	QI QII	farfaṭ farəfṭa mʔarʔad tʔərʔod	2.26% 2.26%
QII	QI QII	tbahdalət bahdale tbahdal tbəhdol	0.75% 1.50%
Exceptions			3.76%

below.

These apparent ‘asymmetries’ seem to contrast with the ‘perfect’ or ‘almost perfect’ correspondence between CIs and CHs that is claimed to exist in certain Semitic languages.

Nevertheless, exceptions to this apparent ‘perfect correspondence’ in patterns between CH and CI have also been documented in Classical Arabic (CA) by grammarians, who do note that some finite verbs in specific patterns might govern a maṣdar of the same root but a different pattern [CA.4]:

[CA.4]   Classical Arabic

wa-tabattal	ʔilay-hi	tabtīl-an
(and-IMP.2MS.devote[CH in pattern V]	to-him	devotion.INF-ACC [CI in pattern II])

‘And devote thyself to Him whole-heartedly’

(Al-Zamaxšarī 1870, p. 111; my glossing)

Ibn Yaʿīš argues that in these cases, the two forms of the verb carry the same meaning (Al-Zamaxšarī 1870, p. 111). However, Sībawayh and Al-Mubarrad, among other grammarians, explain the lack of pattern correspondence as a consequence of the elision of the verb11.

Interestingly, however, the logic for pattern correspondence followed by LA seems to be in line with that of Biblical Hebrew and Syriac.12 In these two varieties, like in LA, the CH and the CI generally share the same pattern. However, Syriac passive verbal forms (i.e., ethpʿel; ethpaʿal; ettaphʿal) can take the infinitive of their corresponding active pattern. In the following example [SYR.3], the CH is in ethpʿel—passive counterpart of pʿal—and the CI is in pʿal pattern.

[SYR.3]  Syriac

meḥzā [CI] ʾetḥāzā [CH] ^hwāt leh ṣúr mtúm

“il n’avait jamais vu Tyr” [he had never seen Tyre]

(Ined. Syr. 2, 14 from Duval 1881, p. 332; English translation mine)

As for BH, Gesenius’ grammar specifies that with a verb of the derived conjugations, not only the infinitive absolute of the pattern can be used, but also the Qal pattern as “the simplest and most general representative of the verbal idea” (Cowley and Kautzsch 1910, p. 345). This is specifically common with verbs in the Niphʿal pattern, the passive or reflexive form of the Qal pattern. In the following example [BH.3], the main verbs are in the Niphʿal form while the CIs appear in the Qal form.

[BH.3] Biblical Hebrew

loʾ tiggaʿ bô yāḏ, kî sāqôl [CI] yissāqel [CH] ʾô yāroh [CI] yiyyāręh [CH]

“They are to be stoned or shot with arrows; not a hand is to be laid on them.

(NIV 2011, Exod. 19:13)

In light of this Semitic evidence, the logic of pattern correspondence between CI and CH in LA can be said to align with systems existing in other geographically adjacent Semitic languages, rather than simply ‘deviating’ from the idealized CA standard of ‘perfect’ correspondence.

3.2. Syntactic Features of CIs in LA in Light of Semitic Evidence

3.2.1. Case Marking on CI

Syntactic case is not marked in Lebanese Arabic; therefore, CIs are not marked with any syntactic case in this variety. This is also the case in all the spoken varieties of Arabic as well as in the great majority of Semitic languages, except for Akkadian, Ugaritic, and Classical Arabic—the only Semitic languages that are known to mark syntactic case with enough attested examples of CIs to render a syntactic analysis possible.13

Although syntactic case has no bearing on LA, looking at the case of CIs in case-bearing Semitic languages may provide further evidence on the distinct natures of CIs and COs, and it may shed light on the function of the CI across Semitic.

CIs in Classical Arabic (CA), such as [CA.1] and [CA.4], appear in the accusative case, generally marked with the indefinite accusative ending –an. In fact, the relevance of the syntactic case for the CA description of CIs is such that Sībawayhi and Al-Mubarrad decided to name this construction after its syntactic case (i.e., maṣdar manṣūb, lit. ‘accusative infinitive’).

Ibn As-Sarrāj observes that the maṣdar used to ‘strengthen’ the meaning of the action has to be in the accusative: “فاذا لم يكن فيها الّا التوكيد نصبت والرفع بعيد جدا” [if its only [function] is emphasis, then it [appears] in the accusative, for the nominative is too improbable] (Ibn As-Sarrāj 1985, p. 168; translation mine).14 Although Sībawayhi had also stated that CIs are accusative, a more detailed reading of his description reveals that he conceived both accusative and nominative as acceptable options and placed the decision in the speakers’ hands: “وكذلك جميع المصادر ترتفع على أفعالها إذا لم تشغل الفعل بغيرها” [in the same way, any maṣdar may be in nominative of their verb if (the verb is) not (syntactically) occupied with another (subject)] (Sībawayhi 1996, p. 229; translation mine).

This flexibility in the case marking of CIs falls in line with the general situation in the Semitic languages. Both –u(m)/-u(n)—generally associated with nominative in Semitic—and –a(m)/-a(n)—generally associated with accusative in the Semitic languages—appear to mark CIs in case-bearing Semitic languages. However, while accusative seems to be the norm in CA, Akkadian15 and Ugaritic present quite the opposite situation, as the following examples illustrate:

[AKK. 3]   Akkadian

[ša i]štu ṣeḫrēku lā āmuru/[am]ārum-ma [ci] ātamar [ch]

        “[That wh]ich I have not seen [si]nce I was young I have seen now”

(AbB 11, 34:5-6 from Cohen 2004, p. 108)

[UG.1]  Ugaritic16

lʾakm [ci] ʾilʾak [ch] [laʾāku-ma ʾilʾaku]

“I will surely send”

(2.30, 19-20 in Sivan 2001, p. 123)

The predominance of the -u(m) ending for CIs’ analogous forms in Akkadian and Ugaritic has triggered discussion about the ending’s function. While some interpret it as a locative-adverbial case (Rosenthal 1942; Pope 1951; Huehnergard 2012), some preferred to think of it as a nominative (Driver 1956; Finet 1952; Kouwenberg 2017, p. 659)17, and others stayed neutral on the grounds of a lack of sufficient evidence (Goldenberg 1971; Bordreuil and Pardee 2009).

Be that as it may, a wider review of the literature on CI case marking that observes Arabic varieties as part of the Semitic continuum suggests that CIs are syntactically marked as salient entities and bearers of adverbial meaning rather than objects. This supposes additional evidence of the formal and functional differentiation of CIs and COs in Semitic varieties and presents the seemingly consistent accusative marking of CIs in Classical and Standard Arabic as an ‘exception’ within the Semitic continuum.

3.2.2. Presence of Enclitics

CIs in Lebanese Arabic do not present any kind of enclitics, as is the case in the majority of the other Semitic languages. The review of the literature shows, nonetheless, that in Akkadian and Ugaritic, CIs systematically present the enclitic -ma/-m, respectively.

In CI constructions in Akkadian (i.e., what scholars referred to as the parāsum (-ma) iprus type), the enclitic particle –ma often appears attached to the infinitives [AKK.1] [AKK.3].18 However, while the enclitic -ma seems to appear very frequently in Akkadian CIs, it is almost non-existent in COs.19 This ‘emphasizing’ particle, according to Huehnergard (1997, p. 325), marks the “logical predicate of a sentence” while for Buccellati (1996, p. 387) –ma “is more often than not associated with emphasis of limitation”20—both functional features associated with CIs and not to COs.

The situation is similar in Ugaritic, where the so-called paronomastic infinitive (i.e., CI) appears with an enclitic –m. Pope (1951, p. 124) suggests that the enclitic –m attached to CIs in Ugaritic indicates “merely additional emphasis” and its omission or addition does not affect the meaning perceptibly. Scholars agree that the Ugaritic –m is related to the aforementioned Akkadian enclitic –ma.

[UG.3] Ugaritic

mtm [ci] ʾamt [ch] [mātu-ma/mūtu-ma ʾamūtu]

“verily I will die”

(1.17 VI, 38 in Sivan 2001, p. 124)

The presence of enclitics in CI constructions in Akkadian and Ugaritic—two of the oldest documented Semitic languages—points directly to the correlation between the use of CIs and the marking of logical predicates, limitation, and/or focus. Once again, a cross-Semitic analysis of formal features elucidates the functional nature of the CI that appears to be linked to information structure.

3.2.3. CIs’ Position in the Sentence

The syntactic position of CIs has been the formal feature probably more widely studied in Semitic languages. Throughout the Semitic continuum, CIs are found both in post-verbal and pre-verbal positions.

The literature suggests that CIs consistently show a post-verbal position in most Arabic varieties, including Lebanese Arabic ([LA.4] [JA.1] [EA.1] [RPA.1] [CA.1] [CA.4]), with the exception of Sason Arabic [SSA.1, SSA.2] (Akkuş and Öztürk 2017). Functionally speaking, it is worth mentioning that this position in many varieties of spoken Arabic is oftentimes reserved for focus of contrast (Brustad 2000).

[LA.4]  Lebanese Arabic

məš	maʕʔūl	šu	b-təšbah-ik...
neg	possible	what	hab-ipfv.3fs.look.like-you.2fs

bəzəʔti-a [CH]	bazəʔ [CI]
pfv.spit.2fs-her	spit.inf

‘It’s incredible how much she looks like you…like two drops of water! [lit. you spat her spitting]

[JA.1] Jordanian Arabic

il-bandora…	ynaššfū-ha [CH]	tanšīf [CI]
def-tomato	dry-sbjv.3mp-3fs	drying

‘As for tomatoes, they used to dry them properly’

(Personal communication from Bruno Herin)

[EA.1]  Egyptian Arabic

nāyim [CH]	fī	l-ʿasal	nōm [CI]
sleep.ptcp	in	the-honey	sleep

‘He is sleeping soundly’

(Woidich 2006, p. 269)

[RPA.1]  Rural Palestinian Arabic

yixinkūna [CH] xanīk [CI]

‘they suffocate us completely’

(Shachmon and Marmorstein 2018, p. 32)

A post-verbal position is also typical of CIs in South Semitic languages such as Mehri [MEH.1] [MEH.3] and Ge’ez [GE.1]:

[MEH.3]  Mehri

ḥṣūr [CH] ḥābū ḥāṣar [CI]

‘he wiped the people out’

(Watson 2012, p. 215)

[GE.1]  Ge’ez

zabṭəwwo [CH] zəbṭata [CI]

‘They whipped him heavily’

(Lipiński 2001, p. 520)

However, this does not seem to be the case in other Semitic languages belonging to the Northwest and East Semitic groups, which tend to show a pre-verbal position, in some occasions accepting CIs both before and after the CH.

Akkadian, Phoenician, and Eblaite, along with Sason Arabic, regularly place the CI in a pre-verbal position. Kim (2006) argues that the pre-verbal position is the most common order in all the extinct Semitic languages while Solà-Solé (1961, p. 191) regards the post-verbal position as the original one. In Akkadian, the CI regularly precedes the main verbal form21, following the [CI+CH] order in affirmatives and [CI+neg+CH] in negative utterances. This is also the case in the few documented instances of CI in Eblaite [EB.1], Phoenician [PH.1], and Ugaritic [UG.1] [UG.2]:

[EB.1] Eblaite

pá-kà-ru [CI] a-pá-kà-ru [CH]

“They should join firmly”

(Lipiński 2001, p. 520)

[PH.1]  Phoenician

ʾm nḥl [CI] tnḥl [CH] mgštk ʿlk wmgšt ʿly

“If you shall come into possession of it (the money), your share is yours and my share is mine”

(Krahmalkov 2000, p. 210)

[UG.2]   Ugaritic

ydʿm [CI] l ydʿt [CH] [yadāʿu-ma lā yadaʿta]

“verily you (m.s.) knew not”

(2.39, 14 in Sivan 2001, p. 123)

In the realm of currently spoken languages, an interesting case is that of Sason Arabic, where, in contrast to the post-verbal position of CIs in all other Arabic spoken varieties, both CIs [SSA.1] and COs [SSA.2] are canonically placed before the CH22:

[SSA.1]   Sason Arabic

şuşa qarf [CI] ınqaraf [CH]

‘The glass broke a breaking’

[SSA.2]   Sason Arabic

babe fadu-ma hedi [CO] ınfada [CH]

‘The door opened a slow opening’

(Akkuş and Öztürk 2017, p. 3)

However, despite researchers’ interest in classifying CIs according to their position, not all languages have clear preferences for pre-verbal or post-verbal positions—what is more, most of those where CIs are fairly well documented often exhibit both. Some examples of the latter are Biblical Hebrew, Syriac, Mandaic, and Northeastern Neo-Aramaic dialects (NENA)23 where, according to the literature, CIs may be found both pre- and post-verbally. Examples from some of these languages are shown in

Table 4. CIs in both pre-verbal and post-verbal positions in specific Semitic varieties.

Pre-Verbal CIs	Post-Verbal CIs
[SYR.5]    Syriac meḥṭā [CI] lmānā ḥṭ-ayt [CH] ‘Why hast thou then [so greatly] sinned?’ (Aphr. 270, 5 in Nöldeke 2003, p. 236)	[SYR.6]    Syriac mramrmin-an [CH] mramrāmu [CI] l-āḵ ‘We extol thee (lit. we exalt you exalting) (Psalm 30 in David and Rahmani 1896, p. 424)
[NENA.1]    North Eastern Neo-Aramaic ʾána zála [ci] har-zílən [ch] bìya ‘I have absolutely gone with it! (i.e., I am finished!)’ (Khan 2008, p. 731)	[NENA.2]    North Eastern Neo-Aramaic ’εga lanwa briθa [CH] ʾana braya [CI] ‘At that time I was not even born’ (Khan 2008, p. 732)
[BH.4]    Biblical Hebrew hę ʾaḵôl [ci] ʾaḵalnû [ch] min hamęlęḵ ‘Have we eaten at all any of the king’s provisions?’ (2Sam. 19:43. NIV translation. See also Gen. 37:8; Isa. 50.2)	[BH.5]    Biblical Hebrew šimʿû [CH] šāmôaʿ [CI] w-ʾal tāḇînû, w-rʾû [CH] rāʾô [CI] w-ʾal tedʿû ‘Be ever hearing, but never understanding, be ever seeing, but never perceiving’ (Isa. 6:9. NIV translation)

:

Given that in BH the pre-verbal (and often clause-initial) position of the CI seems to be the most frequent24 and often regarded as the “basic structure” (Goldenberg 1971, p. 65), there is a vivid scholarly debate as to whether the post-verbal position is syntactically conditioned or not.25 The situation is similar in Syriac (Hoffmann 1827, p. 341; Duval 1881, p. 333; Nöldeke 2003, p. 235)—while some believe that the post-verbal CI expresses a higher degree of emphasis (Nöldeke 2003, p. 236), others find that there is no difference in meaning between both variables (Duval 1881, p. 332 based on Barhebraeus).

Immersed in this process, both groups of scholars seem to have overlooked the possibility that pre-verbal and post-verbal CIs could be, in fact, two separate (although closely related) grammatical forms—with their corresponding functions—rather than two different manifestations of one single grammatical form. In the following section, I discuss how a comparative analysis of CIs’ position across Semitic languages such as the one I propose in this study could, in fact, shed light on the ‘evolution’ of both the formal and functional variation of cognate infinitives across Semitic languages.

4. Discussion

4.1. CI Position across Semitic Varieties: A Practical Discussion

In their attempt to understand CIs’ syntactic variation, many scholars traditionally identified the most frequent position in the varieties they studied, then proceeded to find the formal explanations of what factors may condition the occurrence of the ‘exceptions’ to the rule they had themselves drafted.

It is worth noticing that different approaches in the literature have inevitably resulted from the nature of the available data in each variety, but also from the feeding influences that diverse grammatical schools may have received and, especially, from the authors’ attitudes towards other related Semitic varieties.

Studies on the CI in most Semitic varieties have explored both positions as two different manifestations of the same form (Goldenberg 1971; Cohen 2004, 2006; Mengozzi and Miola 2018). As for Arabic, the few descriptions of CIs in Arabic varieties—where the grammatical tradition establishes the post-verbal position of the mafʕūl muṭlaq—hardly ever include examples of the so-called “extraposed” CIs, for they are considered to be, simply, a separate grammatical entity.

One of the few studies that actually includes such examples—Blau’s Grammar of Christian Arabic—rules out the possibility of extraposed CIs existing in Christian Arabic, and ascribes their occurrence to a Greek-Aramaic interference resulting, presumably, from poor translations (Blau 1967, p. 605)26:

[CRA. 1]   Christian Arabic

سمعا يسمعون ولن تفهمون وبصر تبصرون ولن تنظرون

“hear indeed and understand not, and see indeed and perceive not!”

(Blau 1967, p. 604)

Contrary to Blau’s opinion, although extraposed CIs are rarely documented, they seem to be indeed used in Spoken Arabic. The followings are some examples from Lebanese and Najdi Arabic:

[LA.ext1] Lebanese Arabic

weqraye [ext.CI] kenna neqra qimet saʿa u-nēss ben-nhār bel-qeṣaṣ wer-rwāyāt el-ġrāmiye

“Notre travail durait environ une heure et demie par jour et consistait dans la lecture d’histoires amusantes et romans d’amour” [our daily work would last around an hour and a half and it would entail reading entertaining stories and romantic novels]

(Feghali 1935, p. 10; translation mine)

[NA.1]  Najdi Arabic

hawāš	hāwaš-t-ih
rebuking.inf	rebuked-I-him

‘As far as rebuking is concerned, I have rebuked him’

(Ingham 1994, p. 43)27

Oftentimes, these extraposed CIs may appear followed by a –w that introduces the CH and contributes to the expression of topicalized enumerations in Lebanese and Egyptian Arabic:

[EA.2]   Egyptian Arabic

bōs wi-bosti, hizār wi-hazzarti, liʕb wi-liʕbti

As for kissing, you kissed. As for flirting, you flirted. As for playing, you played.28

(Movie: El nōm fī-l-ʿasal (‘Sleeping in Honey’).

[LA.ext2]

akəl	w-akalna	raʔəṣ	w-raʔaṣna
eat.inf	and-pfv.eat.1s	dance.inf	and-pfv.dance.1s

ma	fi	ši	ma	ʕamelnē
neg	there.is	thing	neg	pfv.do.1s.it.m.sg

‘We ate, we danced… there is nothing we did not do!’

As the previous examples show, at least in Spoken Arabic, extraposed CIs seem to function as regular topics. The infinitive in this case is the chosen form for the topicalization of the finite CH.29 Extraposed CIs could thus be simply considered Infinitival Cognate Topics30, as Ingham (1994, p. 43) suggested, which, as infinitives in topic positions “can be used to encode states of affairs as topics” (Maslova and Bernini 2006, p. 83).

In this sense, it is my impression that, at least in Spoken Arabic, these extraposed CIs, for which we adopt Ingham’s term “Infinitival Cognate Topics”, have a more accentuated nominal character than that of (post-verbal) CIs. For this reason, perhaps, it is common to find in extraposition those infinitives that have been almost completely nominalized (e.g., akəl, raʔəṣ, qrāye, etc.).31 This, along with the function of topic, would differentiate them (but not necessarily isolate them) from the ‘canonical’ CIs described in this study.

Were the case of Spoken Arabic applicable to other Semitic languages, there would be a possibility that the existence of both pre-verbal and post-verbal positions of the CI are simply a manifestation of two closely related grammatical forms. One would be a reduplication of the verb that has been fronted, therefore topicalized, while the other represents the reduplication of the verb that focuses on the event expressed by the CH.

In this case, although the joint analysis of Infinitival Cognate Topics and CIs that has often been adopted in the literature—originally based on an excessive concern for the ‘form’ to the detriment of function—might raise some doubts from a functional perspective, it would be also understandable given the lack of extensive and comprehensive data available in most of the studied varieties.

Another difficulty that might have added to the typological confusion is that the line between the pragmatic notions of topicalization and focus is not only thinner than what it seems, but also, practically imperceptible for scholars working with written texts and consequently deprived from any information regarding the communicative contexts of the utterances in question. With such thin a line, it is not surprising that topics and focus sometimes overlap.

In fact, the function of topicalization (normally assigned to pre-verbal CIs) could have overlapped with that of focus (normally assigned to post-verbal CIs) under the umbrella of CIs in languages such as Old Babylonian (Cohen 2004)—which exhibited two functions but only the pre-verbal extraposed position.

4.2. Data, Ideologies and Their Role in the Creation of Typologies: A Theoretical Discussion

A review of analogous phenomena along the Semitic continuum provides us with invaluable comparative evidence that reveals that the grammatical features of CIs in LA seem to be in line with general Semitic trends that, interestingly enough, do not always find their parallel in prescriptive descriptions of CIs in Classical or Standard Arabic.

Consequently, by taking the analysis of CIs in LA in light of Semitic evidence as a case study, this paper’s results invite the reader to ponder on several theoretical and methodological questions of relevance for the field of Arabic linguistics in general and for that of typology in Arabic dialects in specific.

4.2.1. Data

Scholars working with written texts, particularly those working with extinct varieties, are often deprived of essential information regarding the communicative contexts of the analyzed utterances (e.g., discursive context, communicative setting, communicative intentions and priorities, speakers’ stance, etc.), for the data generally tend to have a performative character, rather than an interactive one.

This leads to the concern that both the amount, but especially the quantity, variety, and contextualization, of linguistic data available in certain varieties may be insufficient for a functional approach. Given that most grammatical descriptions of standard norms rely on this kind of data, typologies may also be of a predominantly form-based nature and therefore neglect important functional aspects.

The joint classification of the functionally distinct categories of CIs and COs in Classical and Standard Arabic is just one example of the potential consequences of an excessive reliance on form over function for grammatical description. As the study of CIs across Semitic languages illustrates, the subsequent grammatical labels that arise from such descriptions and that are systematically attributed to analogous linguistic phenomena across varieties may hamper the typologists’ task of identifying functional cross-dialectal and cross-linguistic patterns, compromising the linguistic accuracy of typologies.

Consequently, the aforementioned considerations would compel researchers to reflect on the following methodological question: In reconsidering old typologies, what kind of data should be considered for the creation of new typologies?

4.2.2. Cross-Dialectal and Cross-Semitic Approaches for Descriptive Purposes

More often than not, the amount of available data at our disposal might be neither enough nor have the desired quality. While questioning the representativeness of the data behind the literature remains an academically healthy and necessary exercise, the reality is that oftentimes the available tools are relatively scarce and limited.

This, however, does not mean that researchers cannot optimize the worth of these resources. Adopting a comparative approach, even when the purpose of the research is not comparative per se, may be an excellent strategy to broaden the researcher’s scope—particularly when facing the scarcity of available data—thus improving the quality of the description.

The study of CIs in light of a comparative analysis of the data available in Semitic varieties has:

(1)

Formally differentiated CIs and COs as different grammatical features and thus questioned traditional Arab grammarians’ categorization of CIs and COs under the label of mafʕūl muṭlaq.

(2)

Shed light on the ways that the functioning of CIs in LA aligns with that of analogous features in other Semitic varieties.

(3)

Provided us with valuable hints and theoretical leads for a better understanding of the functional nature of CIs in LA—one that goes beyond the overused and linguistically vague notions of ‘emphasis’ and ‘intensification’ and points to information structure.

(4)

Given a more holistic understanding of the feature that allows us to build more educated theories regarding the development of this feature.

When considering the Semitic evidence for the study of a linguistic feature in a non-standard Arabic variety, both formal and functional patterns of use become more discernible. Catching sight of these patterns provides researchers with a more holistic understanding of the feature in question and enables them to create linguistic models and descriptions with potential cross-dialectal applications.

4.2.3. Ideologies

According to one characterization of the concept of ‘language ideologies’, these include “speakers’ consciousness of their language and discourse as well as their positionality (in political economic systems) in shaping beliefs, proclamations, and evaluations of linguistic forms and discursive practices” (Kroskrity 2004, p. 498; Kroskrity 2000).

Previous studies have revealed that the ideological biases in linguistic scholarship have tangible effects on practices such as linguistic mapping and/or on the interpretation of historical linguistics (Irvine and Gal 2000).

The thorough reading of the Semitic literature carried out for this study also uncovered two different types of ideologies that may have very possibly affected the accuracy of some of the descriptions of CIs, namely (1) ideologies toward the linguistic feature and (2) underlying ideologies toward linguistic varieties and their grammatical traditions:

(1)

Ideologies toward the linguistic feature: Given its reduplicative character, some scholars have often treated CIs as redundant, literally as mere “ornaments” (Guismondi 1913, p. 65) or as a “purely rhetorical” complementation (Krahmalkov 2000, p. 210). These ideological biases have been enough for some to consider CI features not worthy of systematic analysis. This functional stance is also quite present in the underlying implications of other qualifiers that have been traditionally used in the literature to name CIs—such as ‘paronomastic’, which implies some kind of pun or play on words, or tautological, which directly implies that this infinitive is not necessary and thus “syntactically and pragmatically insignificant” (Callaham 2006, p. 4).32

These ideologies have had a rather tangible effect on grammatical descriptions on CA and MSA, where, in the name of eloquence, the use of CIs is often said to be appropriate only in cases where the meaning of the action is doubtful or vague. Consequently, expressions such as أكل أكلا ʔakala ʔaklan (lit. ‘He ate an eating’) orقعد قعودًا qaʕada quʕūdan (lit. ‘he sat a sitting’), although grammatically correct, are considered by some grammarians as ‘rhetorically weak’, since the meaning of the verbs أكل ʔakala (to eat) orقعد qaʕada (to sit) are not in a situation of uncertainty or doubt (Hasan 2009, pp. 326–27).

However, in sentences such as طارت السمكة في الجوّ طيرانًا ṭārati s-samka fi-l-ʒaww ṭayarānan (lit. ‘the fish flew a flight in the air’), however, the use of the CI is justified by the bizarreness of the meaning (Hasan 2009, p. 327; translation mine). A fairly quick look at the available data, however, shows that this description of CIs is not usage-based, but rather ideology-based.

(2)

Ideologies toward linguistic varieties and their grammatical traditions: In spite of the formal and functional similar nature of CIs all across Semitic languages, certain analyses of CIs across Semitics show traces of ideological biases that can lead to typological inaccuracies.

Goldenberg (1971), for instance, wrote a seminal paper on CIs—which he referred to as Tautological Infinitives—in Biblical Hebrew and presented a classification of this feature with the help of comparative Semitic data. His rather detailed cross-linguistic classification clearly distinguishes “Tautological Infinitives (TI)” (Cognate Infinitives) from “Inner object constructions” (Cognate Object constructions).

Despite the productivity of CIs in different Arabic varieties (illustrated throughout the present study), Goldenberg seems to show some reticence at including Arabic as one of the languages where Tautological Infinitive constructions occur, for he considers the grammatical concept of ‘mafʕūl muṭlaq’ a mere synonym of his notion of ‘Inner Object’ (CO).33 Although, in absolute terms, his description does account (albeit briefly and only in the final pages) for “exceptional” examples of what could be type A and type B Tautological Infinitive constructions in different Arabic varieties, one cannot help but notice that the scanty Arabic data are treated and analyzed with certain skepticism.34

The different treatment of Arabic varieties in this classification—despite the abundant examples of CI available in CA and MSA—could be explained by either (a) an excessive reliance on the joint grammatical traditional label of mafʕūl muṭlaq35 or by (b) insufficient research on CIs in the different Arabic varieties (including the spoken varieties) and/or an overgeneralization of the scarce available data.

Be what it may, both of these factors, probably fueled by ideological biases, may have led to an excessive reliance on formal features (in this case, on syntactic order and case) that in turn, resulted in typological inaccuracies.36

The results of this study suggest that the use and functioning of CIs is fairly similar across Semitic languages with clearly identifiable patterns of formal variation. The divergence between CIs in different Semitic varieties lies, to a great extent, in the different ideological approaches used for its analysis rather than in its linguistic nature.

Once we, as Arabic dialectologists, acknowledge the profound ways in which language ideologies can shape presumably ‘‘objective’’ linguistic analyses, it thus becomes imperative to pose the question: In revising the typologies of the grammatical tradition, how can we identify and set aside language ideologies to ensure accuracy in future typological descriptions of Arabic varieties and its shared features?