Abstract
In this article the interpretation of the Armenian verb karem ‘be able’ as a borrowing from Parthian kar- ‘do’ is discussed. It is argued that the verb was borrowed into Armenian with the meaning ‘be able’ assumed by Parthian kar- in the so-called “potential” construction. Then, some reflections are offered on the theoretical issues raised by this peculiar case, claiming that it can be effectively described in the framework of Diasystematic Construction Grammar. Finally, some remarks are added on the Iranian origin of Arm. bawem ‘suffice, be enough’ and kari ‘strongly, very much’, claiming that, despite being derived from the same roots involved in the “potential” construction, these words did not follow the same borrowing path as karem.
1 Introduction
One of the verbs normally employed in Old Armenian to mean ‘to be able’ is karem (aor. karac‘i), construed with the infinitive (ex. 1–2).
(1) Gen. 44.22
oč‘
karē
manowk-n
t‘ołowl
z-hayr
iwr
not
can.pres.3sg
boy-det
leave.inf
obj-father.acc.sg
his
“The boy cannot leave his father”
(2) Mt. 22.46
ew
oč‘
ok‘
karēr
tal
nma
patasxani
ban
mi
and
not
anyone.nom.sg
can.impf.3sg
give.inf
3sg.dat
answer.acc.sg
word.acc.sg
one
“And nobody could answer him a single word”
A peculiar feature of this verb is that in the oldest texts it is exclusively employed in negative or interrogative sentences (also indirect, e.g., Gen. 44.1 orč‘ap‘ ew karic‘en baṙnal “as much as they could carry”), but never in affirmative sentences, where the periphrasis karoł em + inf. ‘be able to’ is used instead (cf. de Lamberterie 1982: 27–29, ex. 3).
(3) Dan. 5.16
ew
es
loway
z-k‘ēn
t‘ē
karōł
es
meknel
meknowt‘iwns
ew
ard
et‘ē
karic‘es
z-gir-d
ǝnt‘eṙnowl
ew
z-meknowt‘iwn
dora
c‘owc‘anel
inj,
ciranis
zgec‘c‘is
and
1sg.nom
hear.aor.1sg
about-you.abl.sg
that
able
be.pres.2sg
interpret.inf
interpretations.acc.pl
and
then
if
can.pres.subj.2sg
obj-writing-det
read.inf
and
det-interpretation.acc
3sg.gen
indicate.inf
1sg.dat
purple.acc.pl
wear.aor.subj.2sg
“And I heard of you that you can give interpretations. Then, if you are able to read this writing and explain to me its meaning, you will be clothed in purple”
This odd behaviour of karem has long remained unexplained, just like its etymology. The most relevant positions expressed in the literature concerning the latter are two. De Lamberterie (1982) proposed to connect karem to the PIE root *gwerh2- ‘be heavy’ attested in other Indo-European languages only in nominal forms (most famously, the adjective Gk.
As I set out to write this paper, I was surprised that nobody had ever thought of connecting Arm. karem with the meaning ‘be able’ assumed by the Iranian verb *kar- ‘do’ in the so-called “potential” construction (on which see the following paragraphs). However, with impeccable timing, I received confirmation that my surprise was misplaced, and that someone had indeed had this same idea. In June 2024 I had the pleasure of meeting Prof. Daniel Kölligan, who told me that he had recently proposed this explanation of Arm. karem in a forthcoming work (Kölligan 2024, still in print at the time). Had it been a regular case of lexical borrowing, I would probably have been content to see that my hypothesis had received an authoritative confirmation and I would have given up writing a paper on it. However, I believe that this specific instance of borrowing offers food for thought that goes beyond the punctual—and, to be clear, entirely convincing—discussion provided by Kölligan. In what follows, after exposing the arguments that support the connection between Arm. karem and Parth. kar-, I shall present some broader reflections on the dynamics of linguistic interference involved in this example of borrowing. In addition, I will take the opportunity to discuss the possible Iranian derivation of two further Armenian words, i.e., bawem ‘suffice, be able’ and kari ‘strongly, very much’.
While dealing with a single topic, namely the possible reflexes of the so-called “potential” construction of the Iranian verbs kar- and bav- in Armenian, this article is divided into two parts distinguished by quite different methodologies. In the first one, and especially in the final paragraphs, a decidedly theoretical approach is employed, whereas in the second one, dealing with Arm. bawem and kari, the discussion is mainly philological. This is obviously not due to an arbitrary choice of mine but responds to the different nature of the problems addressed. Indeed, I believe that the structure of this paper tells a lot about the problematic position of the study of Armenian-Iranian contact, which can be of great interest for the theory of language contact, but largely remains a domain for specialists due to the highly sectorial competence needed to collect and interpret the relevant data.
2 The so-called “potential” construction in Iranian
As has long been observed, in several Iranian languages (e.g., Old Persian, Sogdian, Khotanese, Chorasmian, Balochi etc.) verbs derived from the roots kar- ‘do’ and bav- ‘be, become’ may take on the meaning ‘be able’ when construed with a verbal adjective, mainly derived from the OIr. ppp. in -ta-.3 Let us only quote some examples of constructions with kar-, employed with transitive verbs, from Middle Iranian languages (on bav- see below):
(4) Sogdian (SCE 273, quoted from Sims-Williams 2007: 378):
r’
β k’wprtr
L’
’krtw
wnty
ill
better
not
do.vbadj.acc.sg
do.pres.3sg
“he cannot cure the sick man”
(5) Khotanese (Z 12.123, quoted from Emmerick 1987: 285):
ka-i
ne
jätu
yīndä
puṣṣo
if-3sg.acc
not
remove.vbadj.acc.sg
do.pres.3sg
completely
“if he cannot remove it completely”
(6) Chorasmian (Q 288, ed. MacKenzie 1991, quoted from Henning & MacKenzie 1971: 40a)
k-f’
m’r-ẟ’r
’y
’llh
ny
p’ry’d-kyc
for-mod
1sg.dat-to
def.m.sg
god
not
hinder.vbadj-do.pres.3sg
“for God cannot prevent me”
Despite being far better attested in Eastern Middle Iranian languages, some traces of this construction have been identified as well in Parthian, which is notoriously the language that has exerted the greatest influence on Armenian in pre-documentary times.4 The clearest examples, found by Sundermann (1981: 59) and discussed by Durkin-Meisterernst (2002: 57–58), are the following:
(7) MKG 1754–1756 (ed. Sundermann 1981):
…
a]ž
imīn
panǰ
[īr]ān
kaẟāž
andar
im
zādmurd
windād
karēnd
from
this.obl.pl
five
thing.obl.pl
ever
in
this
rebirth
find.vbadj
do.pres.3pl
“They can find none of these five things ever in this rebirth”
(8) MKG 723–726 (ed. Sundermann 1981):
ud
Ahremen
pad
hō
rask
burd:
k[ū]-š
tāž
až
kirbag
kirdag[ān
frāma]rg
kird
karān
and
pn
to
3sg.dir
envy
carry.pst.3sg
that-3sg.encl
if only
from
good
deed.obl.pl
negligent
do.vbadj
do.pres.subj.1sg
“And Ahremen was jealous of her: ‘If only I could make her negligent about good deeds’ ”
3 Arm. karem < Parth. kar-
In light of the small extent of the extant Parthian corpus and of the certainly inherited nature of the “potential” construction, it is easy to imagine that examples (7) and (8) represent the remnants of a structure which was more widespread and functional in pre-Sasanian Parthian. As should be clear at this point, I believe that the source of the Arm. verb karem lies in the usage of Parthian kar- in this specific construction, as first suggested by Kölligan 2024: 364–366. In addition to the perfect formal and semantic overlap of the pair Parth. karēd (pres.3sg) ~ Arm. karē (pres.3sg), a crucial piece of evidence in support of this hypothesis is the restriction of karem to negative and interrogative sentences, because all studies on the Iranian “potential” construction register a sharp predominance of negative examples over affirmative ones.5
I believe that the reason behind such a restriction lies in the semantic evolution that led to the development of this periphrasis in Iranian. The first stage must have been the fixation of the idiomatic meaning of kar- + ppp. as ‘make done, accomplish’, still visible in Old Persian, Sogdian, and Khotanese in the use of this construction in subordinate clauses to express anteriority.6 In fact, this is the most common meaning in affirmative sentences (ex. 9–10):
(9) Sogdian (VJ 62–63, quoted from Sims-Williams 2007: 379):
c’n’w
xwrt
’sptk
xwrt
wn’nt
when
food
completely
eat.vbadj
do.impf.3pl
“When they had eaten the food completely”
(10) Khotanese (Z 23.105, quoted from Sims-Williams 2007: 380):
cī
aysura
purrdu
yäḍāndä
when
Asura.acc.pl
defeat.vbadj.acc.sg
do.pf.2pl
“When you had defeated the Asuras”
When turned into negative, this construction means ‘fail in accomplishing’, a concept which lends itself better than its positive counterpart to be equated to the category of ‘being able’ on a merely logical basis: to say that someone is not able to do a given task necessarily implies that they will not complete the said task, whereas to say that someone is able to do a given task does not necessarily imply that they will complete the said task.7 To put it in other words, when turned into negative the construction with kar- loses its inherent telicity, thus favouring the semantic shift towards the typically non-telic category of ability.8 As for questions and other non-affirmative sentences, it is difficult to decide whether the semantic shift from accomplishment to potentiality is also due to a logical implication (to ask if someone will accomplish a given task implies also asking if he is able to do that) or is simply an extension of what happens in the negative, on the model of other negative polarity items that are both found in negative and interrogative sentences.9 Anyway, the fact that the distribution of the meaning ‘be able’ for kar- in non-affirmative sentences is motivated in Iranian but not in Armenian represents in my view a quite relevant proof of the borrowing hypothesis.
Finally, some words should be spent on the fact that the verbal form with which karem is construed in Armenian is an infinitive, whereas in Iranian a derivative of the OIr. ppp. in -ta- is found. I spoke of a “derivative of the OIr. ppp.” and chose the label of vbadj in the glosses above because, from a synchronic point of view, the element governed by kar- in Parthian cannot be unequivocally interpreted as a past participle.10 Often, and always when they have a markedly adjectival sense, past participles in Parthian show an extension in -ag (Durkin-Meisterernst 2014: 253): e.g., būdag ‘been’ from vb. baw-, āyuštag ‘agitated’ from vb. āyōz-. On the other hand, the verbal noun in -t can also express what is called a “short” infinitive beside the one in -tan (< OIr. *-tanai, OP -tanai).11 This “short” infinitive, can be used in constructions which are perfectly parallel to the “potential” construction (ex. 11–12):
(11) Andreas-Henning (1934: 868)
kē
bōxt
kāmēd
rel
save.inf
desire.pres.2pl
“who you want to save”
(12) Andreas-Henning (1934: 855)
pattūd
čār
až
kasādar
parxwadan12
endure.inf
necessary
from
inferior
abuse
“one must endure abuses from inferiors”
As already observed by Kölligan (2024: 365), the verbal noun governed by kar- in the “potential” construction could well have been interpreted as a “short” infinitive rather than a ppp. by bilingual speakers.13 To this one should add that, as is well known, the only proper participle in Armenian, i.e., the verbal noun in -eal, lacks a morphological distinction for diathesis (Jensen 1959: 103). Thus, for example, the Arm. participle bereal can either mean ‘having brought’ (active) or ‘brought’ (passive), which is not well comparable to the strictly passive/intransitive nature of the Iranian ppp. For this reason, even assuming that the nominal form in the Parthian “potential” construction was perceived as a ppp., the Arm. participle in -eal would not have represented a suitable candidate to calque the Parthian construction.14
4 Borrowing or calque?
Traditionally, in the study of language contact a clear-cut distinction is made between two types of borrowing: lexical borrowing and structural borrowing. For example, Renner (2023) opens his bibliographical survey on “Structural Borrowing” stating:
The concept of structural borrowing stands in a binary contrast to that of matter borrowing, which entails the copying of any concrete linguistic element, which prototypically corresponds to the borrowing of lexical units, but also comprises that of function words, bound morphemes, and phonemes. In contrast, structural borrowing can thus be explained as the copying of any abstract linguistic element (i.e., pattern) from one language to another.
This dichotomy was indicated with different terminologies: borrowing vs. calque, loanword vs. loan-translation, direct vs. indirect transfer, global vs. selective copying (Johanson 2023) and the very popular matter vs. pattern replication (Sakel 2007), which I will use in what follows. The idea behind this distinction—more or less explicitly stated—is that in the replication of linguistic material matter and pattern replication are mostly alternative and mutually exclusive strategies.15 This assumption turns out to be problematic as one tries to understand to which category the borrowing of Arm. karem should belong.
On the one hand, the interpretation of Arm. karem as a lexical borrowing—i.e., an instance of matter replication—is clearly unsatisfactory: the meaning ‘be able’ is only explained taking as a basis the contextual usage of Parth. kar- in the “potential” construction, and the limitation to non-affirmative sentences is clearly a point of structure, unrelated to the single lexical item. On the other hand, if structural borrowing is defined as the replication of a pattern of the model language employing categories which are native to the replica language, one is faced with the fact that, in Armenian, the phonological matter of Parthian kar- is replicated as well. Also, if one only considers structural features, I do not believe that the construction of auxiliary + infinitive only in negative sentences can be regarded as semantically marked enough to trigger borrowing. In similar contexts, the employ of the infinitive is perfectly normal in Armenian, and one finds it also with mart‘em ‘be able’, certainly not Iranian.
Apparently, neither form nor structure, taken individually, can describe effectively the mechanism by which karem has been borrowed into Armenian.
5 A Construction Grammar approach
In my opinion, the difficulties detected in the previous paragraph arise because, in the most influential branches of language contact studies, the category of “construction” has not been given the due importance.16 This shortcoming has been emphasised by the most recent studies on the application of Construction Grammar to contact linguistics. In what can be considered a manifesto for the implementation of Construction Grammar in language contact, Boas & Höder (2018: 10–11) rightly observe that the rigid partition of structural levels (lexicon, syntax, morphology, etc.) adopted by traditional models (e.g., Thomason & Kaufman 1988) proves inadequate for describing complex situations in which different levels interact. Recognising the validity of Clyne’s (2003: 76–79) enlargement of the classification of types of borrowing to include all the possible combinations of structural levels (e.g., lexicon + morphology, lexicon + syntax, morphology + syntax, etc.),17 they take one step further claiming that, rather than formulating discrete categories, “it would be more adequate to describe and analyze such contact phenomena in an integrative, non-modular approach.” Indeed, this approach can be employed in the framework of Construction Grammar and the case of Arm. karem represents a perfect example to illustrate the explanatory potential of the constructionist model.
Let us take the definition of construction given by Goldberg (2006: 6): “Any linguistic pattern is recognized as a construction as long as some aspect of its form or function is not strictly predictable from its component parts or from other constructions recognized to exist.” I believe that what makes the Iranian “potential” construction attractive for replication are precisely the aspects of form and function that cannot be predicted from its single components taken individually.18 On the one hand, the replication of the “potential” construction in Armenian can be seen as filling a semantic and pragmatic gap only inasmuch as it is restricted to non-affirmative contexts, i.e., for a structural feature. Unlike the negative meaning ‘not be able, fail’, the positive counterpart ‘be able’ could hardly have been perceived as semantically marked enough to trigger borrowing. On the other hand, the construction was recognisable as an expressive module belonging to the prestige language, and therefore worthy of imitation, due to its phonological shape kar-, which is a matter of lexicon.
Thus, it seems to me that the best way to explain the borrowing of karem is admitting that the Parthian “potential” construction appeared in the perception of the Armenian bilingual speakers as a unitary linguistic item, subject to replication as such.19 The mechanism of replication of a construction is effectively described by the concept of “diasystematic construction” introduced by Steffen Höder.20 A diasystematic construction (or diaconstruction) is defined as a construction in the repertoire of a bilingual speaker that does not belong to only one linguistic system (in our case, only to Parthian or Armenian) but features some language-unspecified elements that can be filled differently according to the context. In Tab. 1 I have represented as combinations of form + meaning respectively the Parthian “potential” construction, the corresponding diasystematic Parthian-Armenian construction, and the construction of karem in Classical Armenian (assuming a no longer bilingual speaker).
Table 1
Borrowing of karem in a constructionist perspective
|
Language |
Parthian |
Diaconstruction |
Armenian |
|
|---|---|---|---|---|
|
FORM |
Syntax |
neg/int sentence aux ‘do’ + vb |
neg/int sentence aux ‘_’ + vb |
neg/int sentence aux ‘able’ + vb |
|
Morphology |
aux.inflection + vb.vbadj |
aux.inflection + vb.non-finite |
aux.inflection + vb.inf |
|
|
Phonology |
kar-ām, -ē, -ēd … + vb-t, -d |
kar-[_] + vb-[_] |
kar-em, -es, -ē … + vb-el, -al, -owl |
|
|
link |
||||
|
MEANING |
Semantics |
(not) ‘be able’ + vb |
(not) ‘be able’ + vb |
(not) ‘be able’ + vb |
|
Pragmatics |
inability, impossibility, failure |
inability, impossibility, failure |
inability, impossibility, failure |
|
In this perspective, we must assume that the Parthian “potential” construction, for the reasons pointed out above, triggered in Armenian bilingual speakers the development of a diaconstruction that could be used in Armenian as well (this is what Höder calls a “pro-diasystematic change”). This happened by selecting some pivotal features to be preserved and replicated in the target language while leaving some less prominent elements as unspecified, so as to allow a more natural implementation in a different system. In particular, the constant elements are obviously the conventional meaning of the construction ‘(not) be able’, but also the phonological shape of the root kar-, the restriction to non-affirmative sentences and the pattern of auxiliary + non-finite form of the verb. The language-unspecified elements (marked as _ in Tab. 1) are the inflection of verbs (personal endings, ending of the non-finite verbal form) and the meaning of the auxiliary kar-, which in Parthian is ‘do’ whereas in Armenian, at the time of the introduction of the construction (column “Diaconstruction” in Tab. 1), was not defined outside the construction itself. The attribution of a lexical meaning ‘be able’ to Arm. karem as an independent verb must have been a later phenomenon, happened when Armenian-Parthian bilingualism was not so widespread anymore and speakers could no longer perceive the link with the Parthian verb (column “Armenian” in Tab. 1).
I believe that this is a good example of the potentiality of the constructionist approach not only to the description of the results of language contact, but also to the clarification of the mechanisms that trigger and bring to completion borrowing.21 I would like to take the opportunity to mention another comparable case of borrowing between Armenian and Parthian that can be explained in the same framework, as it involves the interaction of morphology and lexicon. This is the correspondence between Iranian and Armenian inflectional classes in nominal borrowings, which I have dealt with in a recent paper (Fattori 2023). In my paper it is shown that a regular correspondence between Armenian and Parthian nominal paradigms was established based on the oblique plural form of Parthian nouns, which was the only case-form with a specific ending, the other being morphologically unmarked. For pre-Sasanian Parthian, one can reconstruct three nominal classes with an endingless direct singular case and an oblique plural case with ending -ān, -īn or -ūn respectively. Armenian-Parthian bilingual speakers must have matched this paradigm with the native Armenian -a-, -i- and -ow- declensions that, despite having more case forms, still had an endingless nominative/accusative singular and a genitive/dative/ablative plural in -ac‘, -ic‘ and -owc‘ (e.g., pre-Parth. *mog ‘Magian’, *mogūn → Arm. mog, mogowc‘).22 Such superimposition of paradigms can be seen as the application of a diaconstruction, as represented in Tab. 2 (only the relevant levels of analysis are indicated). The noun chosen as an example is Arm. dat, -ac‘ ‘law, judgement’ < pre-Parth. *dāt, -ān ‘id.’.
Table 2
Borrowing of dat in a constructionist perspective
|
Language |
Parthian |
Diaconstruction |
Armenian |
|
|---|---|---|---|---|
|
FORM |
Morphology |
n.a-stem.obl.pl |
n.a-stem.[_] |
n.a-stem.gdabl.pl |
|
Phonology |
dāt-ān |
dat-a[_] |
dat-ac’ |
|
|
link |
||||
|
MEANING |
Semantics |
‘law’ + genitive, dative, agent etc. |
‘law’ + genitive, dative, agent etc. |
‘law’ + genitive, dative, agent etc. |
Here, the elements common to both systems, leaving aside the form and meaning of the borrowed lexical item, are the category of a-stem, which exists both in Parthian and Armenian,23 its inflectional pattern consisting of root + stem-vowel + ending, and the meaning of the matching case (genitive, dative, agent, etc.). The language-unspecified elements are the actual morphological category, i.e., the oblique case in Parthian and the genitive/dative/ablative in Armenian, and the phonological shape of the ending -n in Parthian and -c‘ in Armenian. Such an analysis, in my view, duly emphasises that in the dynamic of borrowing multiple concurring factors are at work: what allows speakers to establish a correspondence between Parthian and Armenian nominal classes is both the phonological similarity between the inflectional patterns (-ān, -īn, -ūn vs. -ac‘, -ic‘, -owc‘) and the semantic similarity of the cases associated with the endings.
Not being a general linguist, let alone a specialist in Construction Grammar, I thought it could be useful to propose this digression precisely to show the usefulness of introducing the category of “construction” in language contact studies, also beyond the specific interests of Construction Grammar. This seems to me a good way to adequately frame some peculiar instances of borrowing that, though not very frequent, are still possible and found in cases of close contact between two languages. Such circumstances are likely to be found especially in the study of ancient languages of Europe and Asia in contact, a field which, quite understandably, is not often taken as a source for developing linguistic theory. Languages such as Greek and Latin, Greek and Old Iranian, Aramaic and Akkadian are both—more or less closely—genealogically related and typologically quite similar, i.e., they share some common features in the phonological shape of lexical and morphological items and in their structural behaviour, which strongly favours phenomena of multi-layered interference as the ones discussed above.24 As I hope to have contributed to show, in line with other recent studies,25 a particularly interesting case from this point of view is the contact between Armenian and Parthian.
6 Arm. bawem ‘suffice, be enough, be able’, baw ‘term, end, limit’
In Iranian, the “potential” construction with intransitive and passive verbs is expressed by derivatives of Ir. *bav- ‘be, become’ (ex. 13, cf. the literature cited above in n. 3). Since above it was assumed that Arm. karem is borrowed from Parth. kar- in the “potential” construction, one might as well wonder whether the same is true for Arm. bawem ‘suffice, be enough’, sometimes also translated as ‘be able, have strength’,26 and Parth. baw-. In my view, even though this verb is almost certainly borrowed from Iranian, one cannot reconstruct a borrowing pattern similar to that of karem.
To begin with, there is a formal obstacle to that. The phonological shape of Arm. bawem implies an Iranian prototype with long vowel MIr. *bāw-, whereas from MIr. *baw- an outcome Arm. *bovem would be expected.27 On the contrary, in Iranian languages where the “potential” construction with *bav- is attested, the verbs employed all have a present stem with short vowel:28
(13) Sogdian (Henning 1948: 307.22), with MSogd.
β wṭ < Ir. *bavati
’rṭyšyy
xww
n’ywk’wy
nyy
’’p’ṭ
β wṭand-3sg.obl.encl
3sg.nom
depth
not
perceive.vbadj
become.pres.3sg
“and its depth cannot be perceived”
Also, unlike what happens in Iranian, Arm. bawem is mainly employed with the meaning ‘be enough, be able’ in active constructions, not passive or intransitive. If the Iranian construction were calqued, the sentence in ex. (14) should translate as “I cannot be told”:
(14) Agath. 4.17:
oč‘
bawem
patmel
not
suffice.pres.3sg
narrate.inf
“I am not able to tell”
On the other hand, in Sogdian there is a verbal base
Related and yet distinct is the issue of the Arm. noun baw, about which Benveniste (1945: 71) observes: “bav “satiété”, attestait déjà indirectement en moyen-iranien le mot qui reparait dans sogd.
(15) Ps. 146.5
imastowt‘ean
nora
oč‘
goy
baw
wisdom.gen.sg
3sg.gen
not
be.pres.3sg
end
“There is no end to his wisdom”
(Gk.
τῆς συνέσεως αὐτοῦ οὐκ ἔστιν ἀριϑμός “there is no counting of his wisdom”)
(16) Bar. 3.18
ew
oč‘
goyr
baw
stac‘owacoc‘
noc‘a
and
not
be.impf.3sg
end
possession.gen.pl
3pl.gen
“And there was no end to their possessions”
(Gk.
καὶ οὐκ ἔστιν ἐξεύρεσις τῶν ἔργων αὐτῶν “there is no counting of their possessions”)
(17) Bar. 3.25
mec
ē
ew
oč‘
goy
baw
big
be.pres.3sg
and
not
be.3sg
end
“it is big and there is no end (to it)”
(Gk.
μέγας καὶ οὐκ ἔχει τελευτήν “it is big and does not have an end”)
Such meaning is confirmed by the existence of the adjective anbaw, which means ‘endless’ (Gk.
To sum up, we can establish that in Armenian two different idiomatic meanings of the Iranian root *bav- ‘be, become’ are reflected, both directly witnessed in the extant Middle Iranian material: the verb bawem ‘suffice’ lines up with MSogd.
7 Arm. kari ‘very, strongly’ and Sogd. k’ẟy ‘very much’
For the sake of completeness, I shall take the opportunity to briefly discuss the Sogd. word k’ẟy ‘very much’ and to show that it probably has little to do with Arm. kari ‘strongly’, despite having been connected therewith since Meillet apud Gauthiot (1916: 128–129).
Sogd. k’ẟy is constantly referred to in the literature discussing possible Eastern MIr. loanwords in Armenian (see n. 4), but is barely mentioned at all in recent literature dealing with the Sogdian language. This is because it is an extremely rare word in Sogdian, occurring only four times in Buddhist texts and two times in the so-called Ancient Letters in the variant k’ẟyw (ex. 18–23):32
(18) VJ l. 10 (ed. Benveniste 1946)
rty
’xw
k’ẟy
wɣšy
’
β ’and
3sg.nom
very
joyful.nom.sg
become.impf.3sg
“and he became very joyful”
(19) VJ l. 3a (ed. Benveniste 1946)
rty
ZKH
k’ẟy
wɣš’
β ntand
3sg.pl
very
joyful.nom.pl
become.impf.3pl
“and they became very joyful”
(20) TS 5, 71 (ed. Benveniste 1940)
rty
č’wn
wɣšy
k’ẟy
š’t’wx
β ’and
from-3sg
joy.abl.sg
very
happy
become.impf.3sg
“And he became very happy with joy”
(21) TS 5, 87 (ed. Benveniste 1940)
’PZY
β ɣ’yšt‛M
mrtxmyt
’’m’rẟ’n
wyspw
k’ẟy
wɣšy
β ntand
gods.pl
with
man.pl
together
all.indecl
very
joyful.nom.pl
become.impf.3pl
“And the gods together with the men all became very joyful”
(22) AL 1.3 (ed. Sims-Williams 2023)
(’)[ḤR](Z)Ym
’zm’’nk
k’ẟyw
’PZY
β [w]yn’ym(n)
and-1sg.encl
anxiety
very
that-2sg.encl
see.opt.1pl
“And I am very anxious that we might see you”
(23) AL 5.19 (ed. Sims-Williams 2023)
k’ẟyw
’
β zβ r’k’ym
very
wretched
be.pres.1sg
“I am very wretched”
In Buddhist texts, k’ẟy always appears together with wɣš- /wiɣaš/- ‘joy’ or ‘joyful’ as an adjective, and in the Ancient Letters it accompanies as well words indicating a state of mind. Since the postulation of a meaning ‘very much’ is only based on the Chinese translation of the Dīrghanakhasūtra (ex. 20–21)33 and on the alleged equation to Arm. kari, in principle it cannot be excluded that the word had a narrower meaning, restricted to the field of emotions.
The problematic status of this word emerges when trying to find an etymology for it. The first attempt was made by Benveniste (1929: 160), who assumed an equivalence with Arm. kari and postulated an Ir. base *kā̆ẟiya- (his notation) “probablement dérivé de *kā̆ẟa- ‘force’?”, for which no sure connection within Iranian is found. Recognizing this difficulty, Klingenschmitt (1982: 138–139) proposed to set aside the comparison with Arm. kari and to compare Sogd. k’ẟy, read as /kā̆
Even accepting either of these two etymologies, one is faced with a phonological problem. Both Ir. *kādya- and *kā
For all these reasons, and since it only ever occurs in texts written in Sogdian script, I regard as rather likely that the spelling ⟨k’ẟy⟩ might hide an Aramaic heterogram ⟨K’LY⟩, with the phonetic complement -w in the Ancient Letters.34 However, I cannot find an obvious Aramaic antecedent for such word, nor I can propose a plausible candidate for a phonetic reading in Sogdian. No doubt, it should be a word that has a variant with final -w (e.g., BSogd. ɣr
Be that as it may, my purpose here was only to show that the word written as ⟨k’ẟy⟩/⟨K’LY⟩ in Sogdian script is poorly documented and uncertain in its exact meaning and etymology. Since plausible cognates in other Iranian languages are unknown and there is no evidence of direct borrowings from Sogdian into Armenian (see above n. 4), the assumption that this word is the source of Arm. kari does not seem to me to be sufficiently well-founded. This is all the more true when a different explanation is at hand. Indeed, Arm. kari ‘strongly’ can easily be interpreted as an adverbial derivative in -i (Olsen 1999: 434) from the noun kar ‘strength, power’, well attested in Old Armenian as an independent word. In turn, in light of the discussion proposed above, Arm. kar should be seen as a deverbal noun derived from the verb karem.
Abbreviations
| Agath. |
Agathangelos |
| AL |
Ancient Letters |
| Arm. |
Armenian |
| Bar. |
Baruch |
| BSogd. |
Buddhist Sogdian |
| Chor. |
Chorasmian |
| CSogd. |
Christian Sogdian |
| Dan. |
Daniel |
| Gen. |
Genesis |
| Gk. |
Greek |
| Ir. |
(Proto)-Iranian |
| Lat. |
Latin |
| MIr. |
Middle Iranian |
| MKG |
Mitteliranische manichäische Texte kirchengeschichtlichen Inhalts (Sundermann 1981) |
| MP |
Middle Persian |
| MParth. |
Manichaean Parthian |
| MSogd. |
Manichaean Sogdian |
| Mt. |
Matthew |
| NP |
New Persian |
| OAv. |
Old Avestan |
| OIr. |
Old Iranian |
| OP |
Old Persian |
| Parth. |
Parthian |
| PIE |
Proto-Indo-European |
| Ps. |
Psalms |
| Q |
Qunyat al-munya |
| SCE |
Sūtra of the Causes and Effects of Actions |
| Sogd. |
Sogdian |
| TS |
Textes sogdiens |
| Ved. |
Vedic |
| VJ |
Vessantara Jātaka |
| Z |
The Book of Zambasta |
Acknowledgments
In writing this article I received precious advice and criticism by Marco Mancini, Marco Bais, Maria Carmela Benvenuto, Flavia Pompeo, Alessandro Del Tomba, Steffen Höder, Beatrice Grieco, Carmelina Toscano and Alessandro Rossi. To all of them goes my most sincere gratitude. It goes without saying that the opinions expressed in the text are my sole responsibility, including errors and omissions. I also want to thank Daniel Kölligan, who kindly sent me his forthcoming paper and encouraged me to keep on working on Arm. karem in addition to his contribution.
An argument to which de Lamberterie gives much weight is the unusual aorist stem kar-ac‘- instead of the productive type in -eac‘-, expected in the case of a borrowing. However, as demonstrated by Kim (2018: 97–100), there is little ground to postulate that the distribution of the irregular weak aorist suffix -ac‘- reflects an inherited category (specifically old perfects, as some scholars proposed): “In sum, while gitem goes back to the archaic PIE unreduplicated perfect *wóyd- ~ *wid-́, asem probably continues a root present, and karem and mart‘em are not even securely assignable to the inherited lexicon of Armenian [these are the only four verbs having a weak aorist in -ac‘-, M.F.]. Given this uncertainty, it is unlikely that the exceptional morphology of this small group can be motivated on etymological grounds.” Most recently, Kölligan (2024: 365) proposes that, if karem is a loanword, the aorist stem karac‘- could have been levelled on mart‘ac‘- and gitac‘-, both of which can be employed in the exact same contexts (gitem ‘know’ can mean ‘know how to do, be able to do something’).
So Kim 2018: 97–100 and, most recently, Meyer 2023: 265 fn. 26.
Cf. Benveniste (1954), Sims-Williams (2007), Korn (2013), and (Filippone) 2015 with further literature.
It might be useful to remark that, while an additional influence of MP (in Sasanian times) and NP (in later centuries) is certain, the evidence traditionally adduced for an influence of Eastern MIr. languages—essentially Sogdian—on Armenian are all questionable (cf. Henning 1958: 93, Bolognesi 1966: 57418, Schmitt 1983: 85, Livshits 2006: passim, Meyer 2023: 23–24). Arm. margarē ‘prophet’ has been shown to derive from Ir. *mr̥ga-daya- ‘augur, lit. bird-watcher’ (Gippert 2005) and not from BSogd. m’rk’r’k /mārkarē/ ‘soothsayer’ as previously thought. Also, Arm. bargawač ‘illustrious’ can hardly derive from the word for ‘rich, happy’ written as
Cf., for example, Filippone (2015: passim) and Sims-Williams (2007: 3783).
Cf. Benveniste (1954: 63), Sims-Williams (2007: 381), and Filippone (2015: 29).
In previous literature (see n. 6 above), it is only stated that the development of the “potential” construction from the idea of accomplishment is due to the “close relation between successful completion and ability” (Filippone 2015: 29), which does not explain the prevailing use in negative sentences.
Cf. Kagan (2020: 110).
The most obvious example is the distribution of English some (affirmative) vs. any (negative and interrogative), but the same pattern is observed in several other languages including New Persian, with hič ‘anybody’ used only in negative and interrogative sentences (cf. Auwera & Koohkan 2022), and Classical Armenian (cf. Klein 1997), with pronominal and adverbial forms in -mn (affirmative) vs. -k‘ (negative and interrogative).
This was rightly emphasised by Sims-Williams (2007: 377) with reference to Sogdian: “It is important to note that the form of the past stem used in this construction is distinct from the past participle.”
Cf. Durkin-Meisterernst (2014: 260), Ghilain (1939: 122): “L’infinitif en -tan ne se rencontre pas très fréquemment; on utilise aussi, au lieu de l’infinitif en -tan, une forme brève en -t, qui se confond avec la forme du participe passé passif.”
This transcription, different from that of Durkin-Meisterernst (2004: /parxūdan/ or /fraxūdan/), is based on the comparison with BSogd. prxwn ‘Tadel, Schmähung’ proposed by Henning (1937: 76). If this and related forms in Sogdian are from Ir. *hvan- ‘to call’ (so Cheung 2007: 144), then the Parthian word is probably a substantivized tan-infinitive based on the ppp. stem *hvata- (compare OKh. hvata-).
In this respect, it might be worth noting that, depending on the dialect, in Balochi the “potential” construction can feature the bare past stem, the perfect participle in -a(g) and even the infinitive (Korn 2013: 36).
This does not mean that the Arm. participle could not be used to calque a construction involving the ppp. in Iranian (this the explanation of the Arm. periphrastic perfect given in Meyer 2023), but simply that the comparison between the two was not obvious.
See, e.g., Matras & Sakel (2007: esp. 857–858): “This procedure [i.e., structural borrowing], which we have called pattern replication or PAT, thus operates under the constraint of the exclusion or avoidance of direct replication of matter (or MAT) from the model language.”
It would be incorrect to state that the concept of “construction” is totally absent from the relevant literature: e.g., Matras (2020: 255) speaks of “constructions” in defining “pattern replication”, but he does not envisage the possible application of such category in complex contact phenomena.
According to Clyne’s terminology, the borrowing of karem would be described as a “lexico-syntactic transfer.”
A good synthesis of the features that can make a linguistic item attractive for borrowing is provided by Johanson (2023: 7–8).
On the importance of the bilingual speaker’s perception in the mechanism of borrowing cf. Matras (2020: 3): “The relevant locus of contact is the language processing apparatus of the individual multilingual speaker and the employment of this apparatus in communicative interaction. It is therefore the multilingual speaker’s interaction and the factors and motivations that shape it that deserve our attention in the study of language contact.”
Cf. Höder (2012, 2018, and esp. 2021), where this framework is applied to linguistic change resulting from contact.
Cf. Boas & Höder (2018: 26) and again Höder (2021).
The pre-Sasanian Parthian forms are marked with an asterisk as they belong to a linguistic stage that is not directly attested. However, their existence can be reconstructed based on reliable evidence (see Fattori 2023).
Vowel length, which is phonologically distinctive in Parthian but not in Armenian, cannot be considered as a relevant feature in a construction common to both systems.
The linguistic landscape just described would offer plenty of material to assess how effective the application of Diasystematic Construction Grammar is as the genealogical distance between two languages increases, a problem posed by Höder (2018: 63–64).
Cf., for example, the synthesis offered by Mancini (2008) on the contribution of G. Bolognesi to the study of Armenian-Iranian contact, and the works by R. Meyer (2023 and earlier studies), exploring the replication of Iranian syntactic patterns in Armenian.
In Awetik‘ean, Siwrmēlean & Awgerean (1836–1837) s.v., bawem is glossed by Arm. karem, Lat. possum.
A variant bovem for bawem is actually mentioned in dictionaries, but it only occurs in a few late texts (see Ačaṙean 1971–1979/1: 432). It should probably be interpreted as a secondary variant developed in post-classical Armenian by influence of bovandak ‘complete’ and similar forms (on which see below).
In Parthian there are only a couple of very doubtful examples discussed in Durkin-Meisterernst (2002: 58–59).
Cf. Sims-Williams (1976: 58) and Gharib (1995: 99).
I owe one of the anonymous reviewers the reference to Yoshida (1995: 67–70), where this periphrasis is explained as an inflected form of the verb
J̌axǰaxean (1837 s.v. bawem) includes a meaning ‘arrivare, giungere’, but it is probably based on an idiosyncratic translation choice: the only passage quoted as an example is Agath. 4.19 orč‘ap‘ akn baweac‘ for which a translation ‘as far as the eye arrived’ (hasanel) is not better than ‘as far as the eye had strength’ (zawrel) or the like. Therefore, one can exclude that this verb is directly connected to Sogd.
In Sims-Williams & Durkin-Meisterernst (2022: 112a) the sequence ⟨q’ẟyy⟩ is registered under a lemma MSogd. q’ẟy ‘very’ and described as “uncertain, possibly not a complete word.” This unique attestation in Manichaean Sogdian seems too doubtful to be taken into consideration in the following discussion.
Cf. Gauthiot (1911–1912), who translates ex. 20 as “il fut tout comblé de joie”, noting that “[l]e texte sogdien ne parle pas du “trépignement” de joie caractéristique du chinois” (p. 3662), and ex. 21 as “les dieux avec les hommes et immortels [erroneous interpretation of ’’m’rẟ’n] tous furent très joyeux”, adding: “il n’y a ici dans la version chinoise qu’une seule proposition … “se réjouissant grandement, reçurent avec foi” ” (p. 3673).
As is customary, I transcribe the letter deriving from Aramaic lāmadh as ⟨ẟ⟩ in Sogdian words spelled phonetically, where it represents the phonemes /ẟ/ and /
References
Ačaṙean, Hrač‘eay. 1971–1979. Hayeren armatakan baṙaran2. Erevan: Erevani Hamalsarani Hratarakč‘owt‘yown.
Andreas, Friedrich C., & Walter B. Henning. 1934. Mitteliranische Manichaica aus Chinesisch-Turkestan III. Sitzungsberichte der Preussischen Akademie der Wissenschaften: 848–912.
Auwera, Johan van der, & Sepideh Koohkan. 2022. Extending the typology. Negative concord and connective negation in Persian. Linguistic typology at the crossroads 2: 1–36.
Awetik‘ean, Gabriēl, Xač‘atowr Siwrmēlean, & Mkrtič‘ Awgerean. 1836–1837. Nor baṙgirk‘ haykazean lezowi. Venetik I, Tparani srboy Lazarow.
Benveniste, Émile. 1929. Essai de grammaire sogdienne. Part 2. Paris: Geuthner.
Benveniste, Émile. 1937. Notes sogdiennes. Bulletin of the School of Oriental and African Studies 9: 495–519.
Benveniste, Émile. 1940. Textes sogdiens, édités, traduits et commentés. Paris: Geuthner.
Benveniste, Émile. 1945. Études iraniennes. Transactions of the Philological Society 44: 39–78.
Benveniste, Émile. 1946. Vessantara Jātaka. Texte sogdien édité, traduit et commenté. Paris: Geuthner.
Benveniste, Émile. 1954. Expression de “pouvoir” en iranien. Bulletin de la Société de Linguistique de Paris 50: 56–67.
Boas, Hans C., & Steffen Höder. 2018. Construction Grammar and language contact. An introduction. In Constructions in contact. Constructional perspectives on contact phenomena in Germanic languages, ed. Hans C. Boas & Steffen Höder, 5–36. Amsterdam: John Benjamins.
Bolognesi, Giancarlo. 1962. Rapporti lessicali tra l’armeno e l’iranico. Rendiconti dell’Istituto Lombardo di Scienze e Lettere, Classe di lettere e scienze morali e storiche96: 235–258.
Bolognesi, Giancarlo. 1966. La tradizione culturale armena nelle sue relazioni col mondo persiano e col mondo greco-romano. In Atti del Convegno sul tema: La Persia e il mondo greco-romano, Roma: 11–14 aprile 1965, 569–603. Roma: Accademia Nazionale dei Lincei.
Cheung, Johnny. 2007. Etymological dictionary of the Iranian verb. Leiden: Brill.
Clyne, Michael. 2003. Dynamics of language contact. English and immigrant languages. Cambridge: Cambridge University Press.
De Lamberterie, Charles. 1982. Poids et force. Reconstruction d’une racine verbale indo-européenne. Revue des études arméniennes N.S. 16: 21–55.
Durkin-Meisterernst, Desmond. 2002. The uses of baw- in Parthian. In Iran. Questions et connaissances. Actes du IVe Congrès européen des études iraniennes organisé par la Societas Iranologica Europaea. Paris, 6–10 Septembre 1999, ed. Philip Huyse, 47–62. Paris: Association pour l’avancement des études iraniennes.
Durkin-Meisterernst, Desmond. 2004. Dictionary of Manichaean Middle Persian and Parthian. Turnhout: Brepols.
Durkin-Meisterernst, Desmond. 2014. Grammatik des Westmitteliranischen (Parthisch und Mittelpersisch). Wien: Österreichische Akademie der Wissenschaften.
Emmerick, Ronald E. 1987. Auxiliaries in Khotanese. In Historical development of auxiliaries, ed. Martin Harris & Paolo Ramat, 271–290. Berlin: De Gruyter Mouton.
Fattori, Marco. 2023. Armenian stem classes and the Western-Middle Iranian oblique plural case. Studi e saggi linguistici 61: 107–122.
Filippone, Elina. 2015. The so-called Old Persian “potential construction”. In From Aṣl to Zāʾid. Essays in Honour of Éva M. Jeremiás, ed. Iván Szántó, 27–55. Piliscsaba: The Avicenna Institute of Middle Eastern Studies.
Gauthiot, Robert. 1911–1912. Le sūtra du religieux ongles-longs. Mémoires de la Société de Linguisique de Paris 17: 357–367.
Gauthiot, Robert. 1916. Iranica. Mémoires de la Société de Linguisique de Paris 19: 125–132.
Gershevitch, Ilya. 1954. A grammar of Manichean Sogdian. Oxford: Blackwell.
Gharib, Badr al-Zaman. 1995. Sogdian dictionary. Sogdian, Persian, English. Tehran: Farhangan Publications.
Ghilain, Antoine. 1939. Essai sur la langue parthe. Son système verbal d’après les textes manichéens du Turkestan oriental. Louvain: Institut Orientaliste, Université de Louvain.
Gippert, Jost. 2005. Armeno-Albanica. In Indogermanica. Festschrift Gert Klingenschmitt, ed. Günther Schweiger, 155–165. Taimering: Schweiger VWT-Verlag.
Goldberg, Adele. 2006. Constructions at work. Oxford: Oxford University Press.
Henning, Walter B. 1937. Ein manichäisches Bet- und Beichtbuch. Berlin: Verlag der Akademie der Wissenschaften.
Henning, Walter B. 1948. A Sogdian fragment of the Manichaean Cosmogony. Bulletin of the School of Oriental and African Studies 12: 306–318.
Henning, Walter B. 1958. Mitteliranisch. In Handbuch der Orientalistik, ed. Bertold Spuler & Hermann Kees, 20–130. Leiden & Köln: Brill.
Henning, Walter B., & David N. MacKenzie. 1971. A fragment of a Khwarezmian dictionary. London: Lund Humphries.
Höder, Steffen. 2012. Multilingual constructions. A diasystematic approach to common structures. In Multilingual individuals and multilingual societies, ed. Kurt Braunmüller & Christoph Gabriel, 241–257. Amsterdam: John Benjamins.
Höder, Steffen. 2018. Grammar is community-specific. Background and basic concepts of Diasystematic Construction Grammar. In Constructions in contact. Constructional perspectives on contact phenomena in Germanic languages, ed. Hans C. Boas & Steffen Höder, 37–72. Amsterdam: John Benjamins.
Höder, Steffen. 2021. Grammatical arealisms across the Danish-German border from a constructional perspective. In German(ic) in language contact. Grammatical and sociolinguistic dynamics, ed. Christian Zimmer, 11–42. Berlin: Language Science Press.
Humbach, Helmut, John H. Elfenbein, & Prods O. Skjærvø. 1991. The Gāthās of Zarathushtra and the other Old Avestan texts. Heidelberg: Winter.
J̌axǰaxean, Manuēl. 1837. Baṙgirk‘ hay ew italakan. Dizionario armeno-italiano. Venezia: Tipografia Mechitaristica.
Jensen, Hans. 1959. Altarmenische Grammatik. Heidelberg: Winter.
Johanson, Lars. 2023. Code copying. Leiden: Brill.
Kagan, Olga. 2020. The semantics of case. Cambridge: Cambridge University Press.
Kellens, Jean, & Éric Pirart. 1988–1991. Les textes vieil-avestiques. Wiesbaden: Reichert.
Kim, Ronald. 2018. The prehistory of the Classical Armenian weak aorist. Acta linguistica petropolitana 14: 86–136.
Klein, Jared S. 1997. Indefinite pronouns, polarity, and related phenomena in Classical Armenian. A study based on the Old Armenian gospels. Transactions of the Philological Society 95: 189–245.
Klingenschmitt, Gert. 1982. Das altarmenische Verbum. Wiesbaden: Reichert.
Kölligan, Daniel. 2024. Complex predicates and light verb constructions in Classical Armenian. In Berthold Delbrück, Historical and Comparative Indo-European Syntax 1922–2022. Proceedings of the International Conference Delbrück Colloquium on Historical and Comparative Syntax of Indo-European, held in Verona on November 9–12, 2022, ed. Paola Cotticelli-Kurras & Filip De Decker, 351–370. Wiesbaden: Reichert.
Korn, Agnes. 2013. Looking for the middle way. Voice and transitivity in complex predicates in Iranian. Lingua 135: 30–55.
Livshits, Vladimir A. 2006. Armeno-Partho-Sogdica. Iran and the Caucasus 10: 177–186.
MacKenzie, David N. 1991. The Khwarezmian element in the Qunyat-al-munya. London: School of Oriental and African Studies.
Mancini, Marco. 2008. Contatto e interferenza di lingue nei lavori orientalistici di G. Bolognesi. In Dall’Oriente all’Occidente. Itinerari linguistici di Giancarlo Bolognesi, Giornata di studio (4.5. 2007), ed. Rosa B. Finazzi & Paola Tornaghi, 23–52. Milano: Univ. Cattolica del Sacro Cuore.
Matras, Yaron. 2020. Language contact2. Cambridge: Cambridge University Press.
Matras, Yaron, & Jeanette Sakel. 2007. Investigating the mechanisms of pattern replication in language convergence. Studies in language 31: 829–865.
Meyer, Robin. 2023. Iranian syntax in Classical Armenian. The Armenian perfect and other cases of pattern replication. Oxford: Oxford University Press.
Olsen, Birgit A., 1999. The noun in Biblical Armenian. Origin and word-formation—with special emphasis on the Indo-European heritage. Berlin: De Gruyter.
Renner, Vincent. 2023. Structural borrowing. In Oxford bibliographies online. doi: 10.1093/obo/9780199772810-0302.
Sakel, Jeanette. 2007. Types of loan. Matter and pattern. In Grammatical borrowing in cross-linguistic perspective, ed. Yaron Matras & Jeanette Sakel, 15–30. Berlin: De Gruyter Mouton.
Schmitt, Rüdiger. 1983. Iranisches Lehngut im Armenischen. Revue des études arméniennes N.S. 17: 73–112.
Sims-Williams, Nicholas. 1976. The Sogdian fragments of the British Library. Indo-Iranian journal 18: 43–82.
Sims-Williams, Nicholas. 2007. The Sogdian potentialis. In Iranian languages and texts from Iran and Turan. Ronald E. Emmerick memorial volume, ed. Maria Macuch, Mauro Maggi, & Werner Sundermann, 377–386. Wiesbaden: Harrassowitz.
Sims-Williams, Nicholas. 2015. The Life of Serapion and other Christian Sogdian texts from the manuscripts E25 and E26. Turnhout: Brepols.
Sims-Williams, Nicholas. 2023. The “Ancient Letters” and other early Sogdian documents and inscriptions. London: School of Oriental and African Studies.
Sims-Williams, Nicholas, & Desmond Durkin-Meisterernst. 2022. Dictionary of Manichaean Sogdian and Bactrian2. Turnhout: Brepols.
Sundermann, Werner. 1981. Mitteliranische manichäische Texte kirchengeschichtlichen Inhalts. Berlin: Akademie-Verlag.
Thomason, Sarah G., & Terrence Kaufman. 1988. Language contact, creolization, and genetic linguistics. Berkeley: University of California Press.
Yoshida, Yutaka. 1995. ソグド語雑録 (IV) [Sogdian miscellany (IV)]. Studies on the Inner Asian languages 10: 67–70.
