Metin Barak, Angela Ralli and Dimitra Melissaropoulou

Borrowing verbs from Oghuz Turkic: two

linguistic areas

1Introduction

Recent studies have shown that contrary to the traditional view (Dixon 1997:
20) verbs are as prone to borrowing as the members of other lexical categories
(cf. Mifsud 1995; Wichmann and Wohlgemuth 2008; Wohlgemuth 2009; among
others).1 Moreover, similarly to borrowing of other lexical items, certain stra-
tegies have been shown to be involved in the transfer of a verb from one lan-
guage to another (Heine and Kuteva 2005; Matras and Sakel 2007; Matras 2007;
Wohlgemuth 2009; among others). More specifically, Wichmann and Wohlge-
muth (2008: 92102) and Wohlgemuth (2009: 87123) show that languages can
both insert a verbal root/stem in their morphology and adapt verbs with the
use of a (native) light verb, whose function is to integrate the loanverb (see also
Jger 2004). In addition, in certain rare cases of unadapted loanverbs, language
speakers borrow the entire inflectional paradigm along with the verb (e.g. Agia
Varvara Romani, as reported by Igla 1996 and Sechidou 2011). There are two ways
that a verbal root can be inserted in the targets morphology: either by direct or
by indirect insertion (cf. Wichmann and Wohlgemuth 2008: 9799). In direct
insertion, verbal roots of the donor language are plugged directly into the verbal
morphology of the recipient language, and there may be only slight phonological
modifications. In indirect insertion, a verbalizer is usually required in order for
the verb to inflect according to the inflectional pattern of the recipient. This ver-
balizer may be an affix, which flags the part-of-speech membership or defines the
class of the verb.
What is missing, however, is agreement on what exactly determines the type
of strategy(ies) that a specific language adopts while borrowing and accommoda-
ting verbs from other language systems. The prevailing view is that languages may

1This research has been co-financed by the European Union (European Social Fund ESF) and
Greek national funds through the Operational Program Education and Lifelong Learning of the
National Strategic Reference Framework (NSRF) Research Funding Program: THALIS, Investing
in knowledge society through the European Social Fund. The paper is the result of a close colla-
boration and discussion by the three authors. However, for academic purposes, A.R. is mainly
responsible for Section 2.1., M.B. for Section 2.2, while the remaining sections are the outcome
of fully joint work.

Brought to you by | University of Sydney Library

Authenticated
Download Date | 3/30/17 10:41 PM
110 Metin Barak, Angela Ralli and Dimitra Melissaropoulou

freely adopt one or more strategies, and if there is some factor at stake defining or
constraining the borrowing strategy, it is the degree of contact between the donor
and the recipient languages (cf. Wohlgemuth 2009: 263) a factor which, unless
determined quantificationally, does not add up to a satisfactory explanation.
Structural compatibility is another factor that could contribute to the discus-
sion on the direction of borrowing and the selection of a specific accommoda-
tion strategy. Since Meillet (1921), this notion has been a hot topic of debate with
respect to its role in contact-induced grammatical change. The existing approa-
ches vary from the assertion that grammatical borrowing is unrestricted, suppor-
ted by the extreme diffusionists (Schuchardt 1884, 1922; Wackernagel 19268: 8;
Thomason 2001: 11, 63, 68; among others), to the thesis that it is not possible
at all if there is no structural compatibility, supported by the extreme retentio-
nists (e.g. Sapir 1921: 203). An intermediate position is voiced by scholars who
argue that grammatical borrowing is possible, provided that the donor and the
recipient language display tendencies to structural compatibility (Jakobson 1938;
Weinreich 1968: 25; Johanson 2002: 306).2 Elaborating on this, Field (2002: 4142)
proposed the Principle of System Compatibility, which predicts borrowability to
be conditioned by the type of morphological structure of the languages involved
in a language-contact situation (see also Johanson 1999 for a similar proposal).
This paper aims to present a comparative case-study of two linguistic areas,
in which verbs borrowed from Oghuz Turkic to a variety of languages are marked
with two borrowed morphological elements. The first area includes the Balkan
peninsula and Western Asia Minor (Greek vernaculars, various Slavic languages,
Albanian, Bulgarian Romani) and the second covers Eastern Asia Minor, Trans-
caucasia and the region called Transoxiana that corresponds to modern day
Uzbekistan and Tajikistan (various Iranian, Circassian and Lezgic languages as
well as Armenian dialects). With the use of data from these two linguistic areas,
which involve typologically distinct languages, we elaborate on the postulation
of structural compatibility, as a theoretical primitive enabling structural borro-
wing from one language to the other. More specifically we will seek an answer to
the following questions:
i. How did the same morphological markers come into use in such typologi-
cally distinct languages?

2Gardani (2008: 29), examining direct grammatical borrowing (transfer), argues that typologi-
cal divergence does not impede grammatical transfer, although he acknowledges that the lack
thereof favors grammatical borrowing.

Brought to you by | University of Sydney Library

Authenticated
Download Date | 3/30/17 10:41 PM
 Borrowing verbs from Oghuz Turkic: two linguistic areas 111

ii. Can the structural compatibility principle be related to the different accom-
modation strategies and the different markers found in use among the
different recipient systems?

The paper is organized as follows: in Sections 2.1 and 2.2, we define the limits of
the two linguistic areas, with examples from various typologically distinct lan-
guages, and answer the two questions above for the respective areas. Section 2.3
presents an exceptional area to both the first and the second areas and provides
an account for its exceptionality. Section 3 concludes the paper with an outlook.

2Defining two linguistic areas

2.1D(I)-type languages

It has been noted since Sandfeld ([1921] 1930) that verbs from Turkish have
most saliently been borrowed into various languages of the Balkan penin-
sula along with a specific marker -D(I)-.3 Peninsular Greek vernaculars (Indo-
European: Hellenic), Bulgarian (Indo-European: Slavic), Bulgarian Romani
(Indo-European: Indo-Aryan), Pomak (Indo-European: Slavic), Albanian
(Indo-European: Albanian), Slavo-Macedonian4 (Indo-European: Slavic)
and Serbo-Croatian (Indo-European: Slavic) vernaculars are such exemplar
languages and the list can be proliferated:5,6

(1) a. Peninsular Greek

kazadzo become rich <kazan-
kavurdzo roast <kavur-

3The letter in capital denotes an underlying phoneme, and its surface value is determined by
assimilatory rules depending on the language. <I> in parentheses is not overtly expressed when
followed by another vowel (see (1a) and (2)).
4We introduce the term Slavo-Macedonian instead of the widely used Macedonian in order
to denote the language of Slavic origin and distinguish it from the Greek-based Macedonian
dialect.
5A quick note on the convention for transliteration employed throughout the paper: Greek
examples and languages that have established Armenian, Cyrillic or Kartvelian alphabets are
transliterated in broad phonetic transcription. Other languages, e.g. Kurmanji and Zaza, Kabar-
dian are exemplified as they appeared in the sources.
6Data from the Greek varieties are drawn from the databases of the Laboratory of Modern
Greek dialects, University of Patras, Greece (http://lmgd.philology.upatras.gr/index.php/en/).

Brought to you by | University of Sydney Library

Authenticated
Download Date | 3/30/17 10:41 PM
112 Metin Barak, Angela Ralli and Dimitra Melissaropoulou

b. Serbo-Croatian7
kartisati mix, stir <kar-
kontisati converse <konu-
(Friedman 1989: 262)

c. Slavo-Macedonian
bitisuva end <bit- (Breu 1991a: 45)
bendisa please <been-
(Friedman 1996: 134)
d. Bulgarian Romani
ujdisajlo agree <uy-
alastisajlo get used to <al-
 (Kyuchukov 1998: 107)

e. Albanian, Tosk dialect

bastis raid <bas-
gjezdis wander <gez-
(Husi 2002: 1819)
f. Bulgarian
aardisvam be surprised <ar-
kandrdisvam convince <kandr-
(Bernstein 1953: 309, 829)

g. Pomak (Xanthi area)

kazandisavom I win <kazan-
hazrladisavom I prepare <hazrla-
(Aivaz Osmantsa, p.c)

The -D(I)- marker is not exclusively confined to the Balkan peninsula. In the
Greek varieties of Asia Minor (Modern Turkey), the island of Lesbos and Cyprus
(excluding Pontic), verbs from Turkish are systematically followed by the -D(I)-
marker (cf. Melissaropoulou 2010, 2011; Ralli 2012):

(2) a. Cappadocian
juvarladzu roll <yuvarla-
jraatladzu lay down <rahatla-
arad seek, look for <ara-

b. Pharasiot
talatzo stone <tala-
ifletzo double <iftle-

7In Serbo-Croatian, the -D(I)- marker is subject to devoicing. The same holds for Pharasiot and
Cypriot Greek (see also examples (2b, 2d)).

Brought to you by | University of Sydney Library

Authenticated
Download Date | 3/30/17 10:41 PM
 Borrowing verbs from Oghuz Turkic: two linguistic areas 113

c. Lesbian, Aivaliot
savurd throw <savur-
zurladzu stretch <zorla-
atard/atardzu overthrow <aktar-

d. Cypriot Greek
vazkestzo abandon <vazge-
kistzo/kist be angry <kz-

In all cases, -D(I)- originates from the Turkish definite past tense suffix DI- (cf.
Kornfilt 1997: 337; Gksel and Kerslake 2005: 285) but it seems to be devoid of
any morphosyntactic and semantic functions. As suggested by Ralli (2012) for
the Aivaliot verbal loans, the concatenations loan verbal root+D(I) have under-
gone a reanalysis that rendered their internal structure opaque and eliminated
the tense function. In fact, following Aikhenvald (2006: 23) and Aikhenvald and
Dixon (2006: 335336), morphosyntactic elements, when entering from one lan-
guage to another, can acquire a different structure from the one they originally
had, without a change in form being a precondition. In this spirit, root+D(I) forms
do not realize past tense but are reanalyzed in the recipient systems as simple
stems. Ralli (2012) has shown that this is also the reason why -D(I)- in Aivaliot
verbal loans can appear in the present tense or in the future and is not limited
to the past. Crucially, as illustrated by the examples in (3), this reanalysis has
affected the loans of all Greek-based dialects: Cappadocian, Aivaliot, Lesbian,
and Cypriot:

(3) a. Cappadocian
[juvarla -d(i)] -()z -u8
yuvarla -d(i) -vrbz -infl
I roll

b. Lesbian, Aivaliot
[zurla -d(i)] -()z -u
zorla -d(i) -vrbz -infl
I stretch

c. Cypriot
[vazkes -t(i)] -()z -o
vazge -d(i) -vrbz -infl
I abandon

8Phonologically, two adjacent /i/ s are reduced into one; /o/ in Cappadocian becomes /u/ in
word-final position and in Aivaliot, /o/ becomes /u/ in un-stressed position.

Brought to you by | University of Sydney Library

Authenticated
Download Date | 3/30/17 10:41 PM
114 Metin Barak, Angela Ralli and Dimitra Melissaropoulou

As -D(I)- has lost its function as a tense marker, the question that arises now is
why the recipient language borrows the forms in -D(I)- and not just the bare stem.
According to Ralli (2012), this is because verbal word-formation in Greek is usually
based on the so-called aorist stem, i.e. the stem that is employed in the past
(perfective) context. As verbal loan integration may be considered to be a word-
formation process, speakers of the recipient language choose to borrow from the
donor those forms that are used in the past tense.9 However, in a second stage,
they undergo the reanalysis procedure mentioned above, which makes them lose
the tense function, and thus stems in -D(I)- can also appear in non-past contexts,
that is, in the present and the future tenses. We would like to suggest that other
Balkan languages (e.g. Serbo-Croatian (1b), Slavo-Macedonian (1c), Bulgarian
Romani (1d), Albanian (1e), Bulgarian (1f), Pomak (1g)), which also borrow the
verbal forms in D(I)-, display the same pattern to assign a prominent role to the
past stem as far as loan formation is concerned. This is not surprising as in
the Balkan sprachbund, the influence of Greek on the other languages is well-
known and has been manifested in several instances and on several occasions
(see especially Sandfeld 1930: 165).
Moreover, the selection of the particular past stem instead of the bare verbal
stem, which is used in the infinitive shows that, while external factors triggered
by high exposure to another language may lead to verb borrowing, the shape of
these loanverbs decisively depends on language-internal structural factors (see
Ralli 2012). In our case, the crucial factor is the type of the base (i.e. the particular
stem allomorph) that is operative in the recipient language for word-formation
purposes.10

9The reader should note that our aim here is to find an answer to the ostensible randomness
in choice of the definite past tense suffix -D(I)- in loanverbs over any other verbal inflectional
suffixes in Turkish. According to the usage-based model of borrowing (Matras 2009, this volu-
me), borrowing of certain structural categories is not accidental, but inherently bound to their
function and to how these categories support language-processing in discourse.
10Asl Gksel most correctly asks whether the general assertion that verbs in Turkish (and in
Turkic) are bound forms (e.g. Uygun 2009: chapter 4) has any consequence on the limitedness of
borrowing bare forms. When the data in this paper (excluding those in Section 2.3) are taken into
consideration in isolation, a positive answer is at stake. However, there are certain cases where
bare Turkish/Turkic verbs are borrowed into other languages as well. Pontic and Laz (Section
2.3) are already exemplar languages. Besides, Serbo-Croatian has numerous bare verbal borro-
wings from Turkish, perhaps numerically more than those with -D(I)- (cf. (1b)), although in these
cases, they are marked with the Greek aspectual marker -s- (see discussion below for -s- and
Breu 1991a: 4445 for a discussion on -(i)s- in Serbo-Croatian): bater-is-ati go bankrupt <T.
batr-; kajna-is-ati boil <T. kayna- (kalji 1966: 41; see the three lists on pp. 4142: one with
-D(I)-marked verbal borrowings, one without, and the last with verbs that occur both with and
without -D(I)-). When these cases are taken into consideration, borrowing of bound verb roots
from Turkish seems not to be impossible, although further research is required to confirm this.

Brought to you by | University of Sydney Library

Authenticated
Download Date | 3/30/17 10:41 PM
 Borrowing verbs from Oghuz Turkic: two linguistic areas 115

Examples 13 raise an important question pertinent to the type of strategy

which is followed in the accommodation of the loanverbs: they show that, at
least in Aivaliot/Lesbian and Cypriot, loanverbs may follow either the direct
or the indirect strategy, or even both without any difference in the meaning.
Since Hatzidakis (1905), a number of authors (among others, Janse 2001;
Melissaropoulou 2011; Ralli 2012) have observed that the past stem of Greek verbs
belonging to the second inflection class (ICII verbs, e.g. nik I win versus past
nki-sa I won)11 and verbs belonging to the first inflection class, bearing the ver-
balizer -iz- (ICI verbs, e.g. sapzo I rot, putrefy versus past spisa I putrefied)
share with some Turkish past stems the same stem-final vowel, i.e. /i/ (e.g. bitti
s/he/it finished).12,13 The formal phonological correspondence has triggered the
process for the accommodation of verbal borrowings either according to those of
ICII verbs: verbs in o (e.g. Aivaliot atard overthrow); or according to those in
iz (e.g. kazadzu to become rich). According to Ralli (2012), in Aivaliot/Lesbian,
the preference for one particular strategy over the other is ad hoc. In fact, in these
dialects, free alternation between the two strategies is often observed, as the pair
atard/atardzu (2c) illustrates. However, this is not the case for the other dia-
lects, where one particular choice prevails over the other. For instance, the ICI
iz(o) verbs are more frequent in Cypriot, while the ICII -o verbs appear to be the
only choice in the Ulaghats variant of Cappadocian (see Dawkins 1916: 129, 136,
664688; Melissaropoulou 2011: 170175). The productivity, or non-productivity,
of a specific inflection class seems to be a dialect-dependent tendency. As a con-
sequence, loanverbs are easier to be accommodated with suffix(es) denoting
an inflection class that is productive in the specific dialectal system (cf. Melis-
saropoulou 2010, 2011; Ralli 2012). If verbs of ICI are very productively formed
in one specific dialect, then, verbal loans should appear with the verbalizer -iz-,

11The reader is referred to Ralli (2005) for details on Greek inflection classes.
12In this paper, we talk about stems because Greek word-formation is stem-based (Ralli 2005).
However, because in Turkish the third person singular in the past tense (e.g. gel-di-, come-
past-3s s/he came) coincides with word-forms, we suppose that the entire inflected word is
adopted and that this is ultimately reanalyzed as a stem, following the requirements of Greek
word-formation (see also Ralli 2012).
13Note that this is not the complete picture. In Turkish, the past tense marker -DI- is subject
to vowel harmony whereby the final vowel can equally appear as //, u /u/ or /y/ as well,
depending on the preceding vowel. Therefore, the phonological correspondence between the
stem-final vowels is not established in all cases of verbal borrowing between Turkish and Greek
but one should also consider that in Greek, Turkish loans containing <> // usually appear
with /i/ (e.g. kalabalki <T. kalabalk, ari charity <T. hayr). Note that in cases where contact
was rather intense, e.g. in Cappadocian, some -D(I)- marked verbs are (partially) subject to Tur-
kish vowel harmony: aladzo understand <T. anla-, cf. anlad; dndzo think <T. dn-,
cf. dnd; oturduzo sit <T. otur-, cf. oturdu (Dawkins 1916: 6768).

Brought to you by | University of Sydney Library

Authenticated
Download Date | 3/30/17 10:41 PM
116 Metin Barak, Angela Ralli and Dimitra Melissaropoulou

i.e. they will be accommodated according to the indirect strategy. On the contrary,
if ICII verbs are equally productively formed as ICI verbs as happens in Aivaliot
and Lesbian (Ralli 2006) the direct strategy will also be used for the integration
of Turkish verbs.
It is worth noticing that the adaptation of loanverbs following the direct stra-
tegy, that is, those which do not bear a verbalizer and appear only with a person/
number inflectional ending o (e.g. Cappadocian arad and Aivaliot/Lesbian
savurd, in (2a) and (2c), respectively), shows that in the recipient language, the
Turkish complex root+D(I) is still marked as a verb and could not be accounted
for as being underspecified (cf. Moravcsik 1978: 111). That is why the suffix -iz- is
not required for the assignment of the verbal category, at least in those dialects
where ICII verbs are productively formed, as, for instance, in Aivaliot, Lesbian,
and Ulaghats Cappadocian. As the base is the only item which can provide the
loan word with a category, the use of -iz- for the accommodation of Turkish verbs
clearly relates to the formal (i.e. phonological) similarity of the loan and the -iz-
suffixed past verbal forms.
Finally, the hypothesis according to which verb borrowing occurs on the basis
of the past (perfective) stem gets additional support from other Balkan languages
(1bg) which have also been influenced by Greek within the Balkan sprachbund.
In all these languages, the root+D(I) concatenation is further affixed with the
Greek aspectual (perfective) marker -s- (Breu 1991a: 4445, 1991b: 19; Adamou
2012: 160161), which originates from the Greek verbal forms in the perfective
context. Consider the case of Pomak:14

(4) kazan -di -s -av -om15 [=(1g)]

kazan -pst t -npg -ipfv -1sg
I won
hazrla -di -s -av -om
hazrla -pst t -npg -ipfv -1sg
I prepared

14The observation is in fact owing to Breu (1991a: 45), who states that [t]he formerly Greek -s-
has [] become a common Balkanic (gemeinbalkanischen) integration suffix (our translation).
15Adamou (2012: 154, ex. 14; 161, ex. 1920) cites Turkish verbs borrowed into Pomak, spo-
ken in the Xanthi area, with the endings -va-m, contrary to the -av-om endings of our examples
(1g, 4):
i. ben -di -s -va -m <T. been- like
been -lvm1 -lvm2 -pfv -1s
I like
(Adamou 2012: 161, ex.19)
Our informant acknowledges that this -vam/-avom discrepancy is most probably a dialectal one.

Brought to you by | University of Sydney Library

Authenticated
Download Date | 3/30/17 10:41 PM
 Borrowing verbs from Oghuz Turkic: two linguistic areas 117

As has been argued by Breu (1991a: 4445, 1991b: 19), the presence of -s- in langu-
ages exemplified in (1bg) is due to contact with Greek. Greek has transferred to
other Balkan languages not only its tendency to build verbal loans on the basis of
the past (perfective) stem, but also its own perfective marker, which occurs even
when the verbs are directly borrowed from Turkish, and not via Greek. In Breus
terms (1991b: 19, 2526), this is called exchange of donor language (Wechsel der
Gebersprache).16 We would like to suggest that the root+D(I) template in Pomak
(and other languages in 1bf), which comes from Greek, is already structurally
opaque, that is, it is not marked for tense. Moreover, similarly to the -D(I)- case in
Greek, we suppose that a reanalysis applies to the concatenation [stem loan+s],
which renders the new structure opaque as well, and the -s- devoid of any perfec-
tive value. This is why the entire structure [verb root-DI-s] accepts the attachment
of another aspectual marker, e.g. the native imperfective -av- in Pomak (see (4)).

2.2mI-type Languages

It should be noticed that -D(I)- is not present in all Asia Minor languages that are
influenced by Turkish. In fact, in other languages, another Turkic marker, -mI,
is attached to the root, which, in Turkish, marks evidentiality and perfectivity,
but not necessarily in the same context (Gksel and Kerslake 2005: 75), as the
following examples illustrate:

(5) Turkish
oku -mu -
read -mI -3sg
apparently s/he (has) read (evidential+perfective)

oku -r -mu -
read -aor -mI -3sg
s/he (reportedly) used to read (evidential/hearsay)

oku -mu -tu -k

read -mI -pst -1pl
we had read (perfective)

16This case, where initial loanverbs are subject to further affixation with another loan verbal
marker, is termed forward diffusion by Matras (2009: 209).

Brought to you by | University of Sydney Library

Authenticated
Download Date | 3/30/17 10:41 PM
118 Metin Barak, Angela Ralli and Dimitra Melissaropoulou

Relatively well-known cases in the literature are the Turkish verbs borrowed into
spoken informal Kurmanji (Iranian: Kurdish) (among others, Dorleijn 1996: 65;
Haig 2001: 213; Bulut 2006: 108112) and Zaza (Iranian: Zaza-Gorani) (Paul 1998:
100):

(6) a. Kurmanji
tanimi bn get to know <tan-
smrm kirin exploit <smr-
(cited in Haig 2001: 213)

b. Zaza
dmi biyyi endure <dayan-
dimi biyyi/kerdi think <dn-
(cited in Haig 2001: 213)

Turkish verbal roots accompanied by the -mI suffix are integrated in both langu-
ages with the use of a light verb; bn be and kirin do in Kurmanji, biyyi be
and kerdi do in Zaza. The loan verbal root+mI+light verb template for accom-
modating the Turkish verbs is not exclusive to these two languages of Asia Minor.
Armenian dialects of Hemin (Homshetsi, see Vaux 2001a, 2007), Aslanbeg (now
extinct, Aaryan 1898; Vaux 2001b), and Istanbul (Bolsahayeren) also exploit the
same template to some extent:

(7) a. Homshetsi
duunmi ellu think <dn-
tanm ellu move <tan-
(Vaux 2001a: 8)

b. Aslanbeg
rslanmi glla s/he is angry <hrslan-
aalami gnin they punish <aala-
(Vaux 2001b: 55)

c. Bolsahayeren
patlam llal burst <patla-
kprm llal be angry <kpr-
(Sarven Akelyan, p.c)

In (7), the loan verbal root-mI concatenation is accompanied by the Armenian

light verb (Standard Western Armenian: llal) be or (Standard Western Arme-
nian: nel) do/make to surface as a verb. In Kabardian (Northwest Caucasian:
Circassian), spoken also in the Uzunyayla region of Turkey, Turkish verbs are

Brought to you by | University of Sydney Library

Authenticated
Download Date | 3/30/17 10:41 PM
 Borrowing verbs from Oghuz Turkic: two linguistic areas 119

borrowed with the -mI suffix and accommodated with the light verb strategy,
too (cf. Alagozlu 2002, 2007):

(8) Kabardian
a. belirtmi b-ji-n-w- <belirt-
belirtmi 2sg.erg-make-fut-def-aff
you will specify

b. takla atm y-j-a- <takla at-

takla atm 3sg.erg-make-pst-aff
s/he somersaulted
(Alagozlu 2007: 4, our glossing)

Crucially, in all the examples in 68, mI is devoid of its evidentiality/perfectivity

function. Thus, its borrowing is similar to the -D(I)- case, which is used in the lan-
guages of the Balkan peninsula, in Cypriot and Asia Minor Greek. In other words,
mI together with the Turkish root may also be considered as being subject to a
reanalysis which has rendered the structure opaque.
Yet, it should be noticed that the limits of borrowing with -mI by far exceeds
the Asia Minor geographic area and expands through Caucasus and Khorasan
to Transoxiana. Doerfer (1993) gives a neat survey of the presence of -mI in the
Iranian languages of the area. Persian texts from as early as the 13th century attest
loanverbs from Turkic with -mI and an abstract nominal suffix -i (Menges 1956:
9091; Doerfer 1993), which is ultimately combined with a light verb numdan
show or with the verb kardan do:

(9) Persian
agirlm numdan respect <arla-
blm kardan lead <bala-

Doerfer also states that the loan verbal root+mii template was in use for a
long time in Persian, until the end of the Mongolian dominion with the fall of
the Timurid Dyntasty in the 16th century, and was also in use to accommodate
Mongolian loanverbs, as the following examples illustrate:

(10) Persian
jirgm kardan enjoy <W.M jirgaqu enjoy, M.
glm kardan think <W.M. aqlaqu consider, M.
tarqm kardan diverge <W.M. tarqaqu scatter, M.
slgm kardan select, divide <W.M. salgaqu divide, M.
(Doerfer 1963: 32, our translation)

Brought to you by | University of Sydney Library

Authenticated
Download Date | 3/30/17 10:41 PM
120 Metin Barak, Angela Ralli and Dimitra Melissaropoulou

The same template, although without the abstract nominal suffix -i, survived
in Northern Tajik dialects (Iranian: Persian) and has been extensively used to
accommodate verbs from neighbouring Uzbek (Doerfer 1967: 6770):

(11) Northern Tajik

blmi kardan find <U. bul-
salqinlami kardan cool oneself <U. salqinla-

The root-mI template with a light verb is also in use in the Iranian languages
around modern day Azerbaijan. Tati (Iranian: Tat) in Northern Azerbaijan, and
Talysh (Iranian: Talysh) in Northern Iran and Azerbaijan are two such languages:

(12) a. Tati
injitmi sxtn hurt <A. incit-
utanmi bivn be ashamed <A. utan-
(Doerfer 1993: 92)

b. Talysh
bagilami karde donate <A. bala-
azmi be err <A. (yolunu) az-
(Doerfer 1993: 92)

Yet, in these languages, similarly to Kurmanji and Zaza (6ab), and contrary to
Northern Tajik (11), transitives and intransitives differ in terms of the light verb
employed. While the transitive verbs are expressed with the do verbs, intran-
sitives are expressed with be. This difference maintains even in non-Iranian
languages around Azerbaijan. Lezgian (Northeast Caucasian: Lezgic) makes use
of the light verb qun be, become for the intransitive verbs, whereas transitive
verbs are not expressed with a light verb. Here they are cited with the infinitival
-un (see also Haspelmath 1993: 177):

(13) a. lekelamiun stain <A. lkl-

terbijalamiun educate <A. trbiyl-

b.
lekelami qun stain oneself <A. lkl-
terbijalami qun educate oneself <A. trbiyl-
(Talibov and Gadzhiev 1966: 187, 208)

Similar verbs borrowed from Azerbaijani are also attested in Udi, another
Lezgic language of the Northeastern Caucasus. While transitives are most often
expressed with besun do/make, passives and unaccusatives are formed with
baksun be:

Brought to you by | University of Sydney Library

Authenticated
Download Date | 3/30/17 10:41 PM
 Borrowing verbs from Oghuz Turkic: two linguistic areas 121

(14) bagilami besun forgive <A. bala-

begmi besun approve <A. byn-
agulami baksun get poisoned <A. ala-
begmi baksun be approved of <A. byn-
(Gukasyan 1974: 35, 66, 76)

Tsakhur, a Western Lezgic language, also possesses many forms of root-mI with
the light verb (i)es be or haas do which reflect difference in valency:

(15) tsekmmi es doubt <A. kin-

telesmi es hasten <A. tls-
attami haas fold <A. qatla-
arzulami haas desire sth <A. arzula-
(Kazenin et al. 1999)

As mentioned earlier, verbs with -mI- were attested as early as in 13th century
Persian texts. Bulut (2006: 108) states that a big number of these borrowings
were from Chagatay to Persian, in which the -mI suffix was then still productive.
Doerfer (1993) asserts that they are mostly from Old Uzbek, as they are also attes-
ted in Northern Tajik dialects. The suffix in Old Uzbek was present between the
11th and the 15th centuries, whereas in Modern Uzbek, the suffix is rather rarely
in use. In either case, however, the root+mI contrary to languages around Azer-
baijan and in Asia Minor always surfaces with the light verb kardan do. Yet,
in the languages around Azerbaijan, valency plays a systematic role in the selec-
tion of the light verb. Kurmanji (6a), Zaza (6b), Tati (12a), Talysh (12b), Lezgian
(13), Udi (14) and Tsakhur (15) employ different light verbs for accommodating
the root+mI borrowings as intranstive or transitive verbs. In our opinion, this
indicates a lingustic subarea within the overall linguistic area of the mI-type
languages. In these languages, the borrowing is mainly from Azerbaijani. This
is plausible particularly in light of the fact that Azerbaijani served as a lingua
franca in the Transcaucasia, Eastern Asia Minor excluding Pontus and
Northern Iran from the 16th to the 20th century when Russian took its place:

The Turkic Azerbaijani language, in the southeastern part of the Caucasus area, has been
a very important lingua franca [] amongst speakers of most of the Lezgian languages []
and also some of the Avar languages. [] The same is the case with the Iranian Tat and
Talysh languages whose speakers are culturally and linguistically incorporated into the
Azerbaijan world [reference omitted].
In the recent decades, the influence of the Azerbaijani language has been very much weake-
ned by mounting competition from Russian [].
 (Wurm 1996: 956)

Brought to you by | University of Sydney Library

Authenticated
Download Date | 3/30/17 10:41 PM
122 Metin Barak, Angela Ralli and Dimitra Melissaropoulou

There are a number of international languages used by the various peoples to communicate
with each other. [] Azerbaijani prevails in much of Transcaucasia (except for the Black Sea
Coast) and, in addition, in Turkish Armenia, Kurdistan and Northern Persia.
(Trubetzkoy [1925]1999: 478, our translation)

One piece of evidence for this is provided by Doerfer (1993: 93): in Tati, verbs
which involve the -l- cluster in Turkish, Uzbek, and Western Azerbaijani, are bor-
rowed as -d-, e.g.:

(16) Tati
batami stn start <A. bala-

The l > d/t alteration is only observed in Eastern Azerbaijani, in the region that
was possibly the source language of the borrowing.
Ultimately, we hypothesize that the already adopted template in Northern
Tajik and perhaps Persian, root+mI+do, served as a prefigured frame for the
adaptation of Azerbaijani verbs in the neighboring languages with further diffe-
rentiation between transitives and intransitives. The case is possibly extended to
Kurmanji and Zaza in Asia Minor, as well, where it is frequently and productively
used. However, contrary to the case of Tati, Lezgian, and others, for Kurmanji and
Zaza it cannot be clearly stated whether the source language is Azerbaijani or
Turkish. This is partly because of the lack of written sources17 but mostly because
Azerbaijani and Turkish can be viewed as a dialect continuum (see Schnig 1998:
248) and the local varieties of Eastern Asia Minor from which the verbs are borro-
wed stand halfway in this continuum.
Despite the fact that we have given an account of how root+mI(+light verb)
diffused from Uzbek/Chagatay and further from Azerbaijani to neighboring lan-
guages of the area, the main question why the suffix -mI and not another suffix
is employed in the accommodation of verbs remains unanswered. To provide a
tentative answer, however, we have to give an account of -mI in Uzbek, which is
a member of the Karluk branch of Turkic, and, crucially, in Azerbaijani. In the dis-
cussion revolving around (5), we have stated that mI in Turkish expresses both
perfectivity and evidentiality. Johanson (2000) further elaborates on the semantic
functions of mI in Turkic languages. More specifically, he argues that mI in
Turkic expresses both indirectivity and post-terminality. Indirectivity is a cover
term and, more importantly, a cognitive category that entails various meanings,
such as hearsay, inferential, admirative, and so on, also entailed in asserted

17The oldest records of root+mI+light verb in Kurmanji date back to some 150 years ago
(cf. Lerch 18578).

Brought to you by | University of Sydney Library

Authenticated
Download Date | 3/30/17 10:41 PM
 Borrowing verbs from Oghuz Turkic: two linguistic areas 123

sentences, i.e. sentences with contradictible content. Indirectivity does not occur
in embedded contexts (Johanson 2000: 61). It should also be noted that indirec-
tive meaning is entailed only by the finite mI and that this meaning does not
derive from the non-finite, i.e. participial, one.
Post-terminality, however, is a different notion: both finite and non-finite
mI suffixes share the aspectual quality of post-terminality. The aspect of post-
terminality, which is expressed by -mI in Turkish, envisages an event at a point
where its relevant limit is transgressed, having done (Johanson 2000: 62).
Johanson (1971: 280292) states that indirectivity and post-terminality are highly
related. Post-terminality the view of a concluded anterior event from the vie-
wpoint of the present easily entails the reading of indirectivity. At this point,
whether the inference of indirectivity is established by way of inference from per-
ceptional traces, or through hearsay, becomes only ancillary.
Azerbaijani and Turkish show considerable differences in the semantic func-
tions of the finite mI (Johanson 1971: chapter 8). While in Turkish the semantic
interrelatedness between post-terminality and indirectivity is rather complex and
post-terminal meaning entails the indirective one, in Azerbaijani the use of mI
tends purely to its post-terminal meaning (Johanson 2000: 74, 2006: 79) (for an
elaborate analysis of mI in Turkish, see, among others, Johanson 1971; Aksu-Ko
2000; Csat 2000):18

(17) yar-m- itir-mi-m

beloved.one-poss1sg-acc lose-post-1sg
I have lost my beloved one.
 (Johanson 2000: 74)

What is crucial here is that this type of post-terminality is similar to the Persian
perfective construction, which has a vague indirective meaning roughly corres-
ponding to Azerbaijani post-terminal -mI:

(18) a. dideh st
seen 3sg
s/he has seen

b. rfteh m
gone 1sg
I have gone

18In some East Anatolian [Turkish] dialects that are close to Azeri, the finite inflectional mar-
ker -mi, which conveys indirectivity in Standard Turkish, tends to express pure postterminality
in the sense of a perfect [] (Johanson 2006: 79).

Brought to you by | University of Sydney Library

Authenticated
Download Date | 3/30/17 10:41 PM
124 Metin Barak, Angela Ralli and Dimitra Melissaropoulou

Johanson attests this roughly perfective use of mI in Azerbaijani to a possible

Persian influence on the language. Conversely, in Northern Tajik, a tense-neutral
indirective category is grammaticalized due to Uzbek interference, as Windfuhr
(2005: 99) states:

[The] appearance [of examples of inference] in early texts, as well as their reappearance in
contemporary standard Persian of Iran, can again be explained by interference from Turkic
where inference is marked by emi. Unlike Turkic, inference is not tense-neutral in Persian,
but confined to the past. In Tajiki, however, mi-raft-e ast has already become tense-neutral.

Similarly, although not grammaticalized as a category, indirectivity became

salient in Kurmanji in Asia Minor mostly because of Turkish influence (Bulut
2000: 176).
We think that the contact-induced aspect of perfectivity/indirectivity in the
surrounding Iranian languages served as a basis for the borrowing of verbs with
mI from Uzbek and later from Azerbaijani. For us, this entails that verbs from
Turkic are not borrowed as participles into Iranian languages.
At this point, the employment of a complex predicate with a light verb
must be discussed in more detail. It is well-known that Iranian (and Lezgic)
languages often use a light verb, rather than an affixal verbalizer, to verbalize
a nominal item (cf. Windfuhr and Perry 2009: 496). This possibly fostered the
accommodation of Persian and Arabic verbs in (Ottoman) Turkish by means of
the same strategy. Compare the Persian example (19a) with the Turkish coun-
terpart in (19b):

(19) a. Persian
sohbet krdn chat <Ar. suhba friendship

b. Turkish
sohbet etmek chat <Ar. suhba friendship
(possibly via Persian)

In the Iranian languages under examination, this light verb strategy is highly
productive also with native nominals:

(20) rng krdn paint <P. rng color

ba:z krdeh st <P. ba:z open
open done 3sg
s/he has opened

Thus, root+mI concatenations, whose borrowing is promoted by the new aspect

of perfectivity/indirectivity, were simply accommodated with the most salient

Brought to you by | University of Sydney Library

Authenticated
Download Date | 3/30/17 10:41 PM
 Borrowing verbs from Oghuz Turkic: two linguistic areas 125

verb creation/loanverb accommodation strategy, i.e. the light verb strategy.

Middle Persian went even further here and employed the abstract nominal suffix
i to nominalize the loan verbal root+mI concatenation. This newly created
root+mI+light verb complex possibly served as a preconfigured template for the
other non-Iranian languages of the area, i.e. Armenian dialects, Circassian langu-
ages, and so on, which employ the light verb strategy to accommodate borrowed
items, verbs or nouns.
Table 1 below lists the languages treated in this paper, according to the two
loanverb accommodation strategies, including the respective loanverb suffix.

2.3A note on Pontic and Laz

It has been noted in Section 2.1 that, contrary to other Greek dialects, Pontic Greek
does not have verbs borrowed with the marker D(I). The Turkish verbal roots are
accommodated by the native verbalizer ev-:

(21) Pontic Greek

kiralavo rent <kirala-
palavo start <bala-

Ralli (2013: 441) states that the exceptionality of Pontic among the Greek dia-
lects is largely caused by the high productivity of the verbalizing suffix ev. She
further shows that the lack of phonological similarity between the Turkish past
in D(I) and the Pontic past form in eps <ev+s was one of the causes for the
absence of loanverbs with D(I) in Pontic. However, Pontic does not employ the
mI suffix in the integration of verbs either. This seems to be because Pontic
does not possess (a counterpart of) post-terminality, contrary to the Iranian lan-
guages of the area. In addition, the area of Pontus had always been out of the
linguistic area of Azerbaijani. As a consequence, it did not acquire the already
available root+mI template that is employed largely in Transcaucasia and
Eastern Asia Minor.

Tab. 1:Languages with -D(I) and -mI loanverb suffix

-D(I) suffix, no light verb -mI suffix+ light verb

Albanian, Bulgarian, Bulgarian Romani, Greek Armenian dialects (Aslanbeg, Bolsahayeren,
vernaculars (Aivaliot, Cappadocian, Cypriot, Homshetsi), Kabardian, Kurmanji, Lezgian,
Lesbian, peninsular Greek, Pharasiot), Northern Tajik, Talysh, Tati, Tsakhur, Udi,
Pomak, Serbo-Croatian, Slavo-Macedonian Zaza

Brought to you by | University of Sydney Library

Authenticated
Download Date | 3/30/17 10:41 PM
126 Metin Barak, Angela Ralli and Dimitra Melissaropoulou

In Pontus, Laz also (Kartvelian: Zan) misses Turkish verbs displaying D(I) or
-mI. The fact that only few borrowed verbs occur in Laz is peculiar in compari-
son with the number of other borrowed lexical items and considering the high
degree of structural influence from both Turkish and Greek on it (Haig 2001;
Brendemoen 1990). Crucially, these verbs are never marked by -mI:19

(22) a. Laz, Mutafi dialect

i.duun.ai <dn-
think.prs.3sg
he thinks
(Wodarg 1995: 119)

b. Laz, Arhavi dialect

b-i-alis-am-ti-a <al-
i1-val2-work-ths-pfv-rep
I worked
(Lacroix 2009: 824)

Laz is a head-marking language with complex verbal suffixation. Languages of

this type are generally more resistant to borrowing verbs, as Haig (2001: 214)
states. However, considering the abundant root+mI verbs in Kabardian, which
is also a head-marking language with complex verbal suffixation (8), the total
absence of root+mI verbs in Laz is probably explained through the fact that
Pontus where Laz was spoken as well was exempted from the Azerbaijani
linguistic area.
At the end of our discussion, we can tentatively define two linguistic areas in
terms of verbal borrowing from Oghuz Turkic. The first the area of the D(I)-type
languages extends from the western frontiers of the old Ottoman boundaries in
the Balkan peninsula to Western Asia Minor; the second area, which includes the
-mI-type languages, starts from Transoxiana and covers Chorasan and Cauca-
sus, as well as Eastern Asia Minor. Strikingly, the second area was first shaped by
Uzbek, a Karluk Turkic language, and was further spread by Azerbaijani. Between
these two areas, Pontus constitutes an exception. As already mentioned, we
suggest that this exception is caused by the structure of the two languages in the
area, but also the fact that Pontus was excluded from the radius of influence of
Azerbaijani, which served as a lingua franca (see Figure 1).

19Georgian is similar to Laz in that sense. We thank Peter Arkadiev for pointing this to us.

Brought to you by | University of Sydney Library

Authenticated
Download Date | 3/30/17 10:41 PM
 

6 1 5
2 6
4 2
5
3 1 1 7
2 3
7 8 4 8 13
9 9 10
10
11

11 12

D(I)-type languages mI-type languages

1 Bulgarian 1 Bolsahayeren 8 Kurmanji
2 Bulgarian Romani 2 Aslanbeg 9 Zaza
3 Pomak 3 Homshetsi 10 Talysh
4 Slavo-Macedonian 4 Kabardian 11 Tati
5 Albanian 5 Lezgian 12 Persian
6 Serbo-Croatian 6 Tsakhur 13 Tajik
7 Peninsular Greek 7 Udi
8 Aivaliot/Lesbian
9 Cappadocian
10 Pharasiot no DI or mi verbs
11 Cypriot 1 Pontic Greek 2 Laz
0 250 500 1.000
km

Fig. 1:Two linguistic areas of borrowing verbs from Oghuz Turkic.

Borrowing verbs from Oghuz Turkic: two linguistic areas

Authenticated
Download Date | 3/30/17 10:41 PM
Brought to you by | University of Sydney Library
127
128 Metin Barak, Angela Ralli and Dimitra Melissaropoulou

3Conclusion

This paper has defined two linguistic areas in which Oghuz Turkic verbs are bor-
rowed into a number of languages along with two distinct Turkic suffixes: the
past D(I) from Turkish and the post-terminal/indirective suffix -mI from Azer-
baijani (though Uzbek, a Karluk Turkic language, is the source of loanverbs for
a number of languages). We have stated that the first area overlaps with the
western borders of the Ottoman Empire and the verbs are borrowed into the lan-
guages of the area from Turkish. The choice of the -D(I) suffix is due to two inter-
related factors: (a) Greek deverbal word formation is based on the past stem, and
(b) the partial phonological similarity between the Turkish past and the Greek
past triggered an analogy process for the molding of the verbal loans in the
present tense. The other Balkan languages, which also borrow the verbal forms in
-D(I), have been subject to this property of Greek to assign a prominent role to the
past stem in loanverb formation.
With respect to the second area, in which verbs are borrowed with the post-
terminal/indirective suffix -mI, we have shown that its borders go beyond Asia
Minor and reach Transoxiana. The root+mI template with the use of a light verb
possibly emerged in Transoxiana, more specifically in Northern Tajik dialects,
under the influence of Uzbek, a Karluk Turkic language, and further diffused into
the languages of the area surrounding Azerbaijan, where transitives and intransi-
tives are differentiated according to the type of the light verb employed. We have
suggested that the emergence of -mI as a loanverb marker is similar to the case
of D(I) not coincidential and largely hinges on the contact-induced emergence
of post-terminality in Northern Tajik and other Iranian languages. The employ-
ment of the light verb strategy is only auxiliary to the discussion, as it is the most
salient native strategy of denominal verb-formation in Iranian languages. We
have stated that the template root+mI is created inside the Iranian language
family and is adopted by the surrounding languages of different typologies with
a further differentiation based on valency.
Pontic Greek constitutes an exception among the Greek dialects as it borro-
wed no verbs from Turkish with the -D(I) suffix. Aligning with Ralli (2012), we
have shown that this is partly because the Turkish past tense suffix and the Pontic
Greek past tense suffix do not exhibit the phonological compatibility that other
Greek dialects show. Concerning the lack of verbs borrowed according to the tem-
plate root+mI, we have proposed that this is caused by the isolation of Pontus
from the linguistic dominion of Azerbaijani as a lingua franca. This also exp-
lains why Laz similarly lacks verbs marked via the suffix -mI.
The discussion in Sections 2.1 and 2.2 has clearly demonstrated that exter-
nal factors triggered by high exposure to another language may lead to verb

Brought to you by | University of Sydney Library

Authenticated
Download Date | 3/30/17 10:41 PM
 Borrowing verbs from Oghuz Turkic: two linguistic areas 129

borrowing, even in extreme cases such as Laz, but the decisive factors for the
shape of these loanverbs are language-internal and structural. In the case of the
D(I) type, the decisive factor is the type of the base (i.e. the particular stem allo-
morph) that is operative in the recipient language for word-formation purposes,
and in the case of -mI-type languages, it is the existence of post-terminality com-
bined with the extensive use of light verbs for the creation of denominal verbs.
The current paper has left uncovered a few issues related to verbal borro-
wing from Oghuz Turkic. Firstly, the two templates are not exclusive to the lan-
guages exemplified. For example, in the Armenian dialects, verbs from Turkic
are either borrowed with -mI or the Turkish verbal root is directly inserted into
the native paradigm; in Pharasiot, verbs come either with the -D(I) suffix or bare
roots are affixed with native verbalizers. However, no language makes use of
both the root+mI and root+-D(I) template. Because we think that factors that
define the choice of bare verb-roots and/or their accommodation with either -mI
or -D(I) should be studied separately for each language, we have left this topic
for future research. Secondly, although we have shown that the choice of -mI is
largely related to the emergence of post-terminality in Northern Tajik and Iranian
languages, we have not discussed it extensively for all the languages employing
the same template. Beside the Iranian languages, languages which are within
the mI-linguistic area also developed in various shapes morphologically
expressed post-terminality and/or indirectivity (see, e.g., Donabedian 1996 for
Western Armenian). A final topic that has not been dealt with here is the reanaly-
sis of the Turkic denominal verbalizer suffix -lA in the languages exemplified in
the paper. Robbeets (this volume) provides an extensive survey of the correspon-
dences of the suffix in the Transeurasian languages (Japonic, Koreanic, Tungusic,
Mongolic and Turkic; cf. Johanson and Robbeets 2010: 12) and argues that these
correspondences are not caused by borrowing but inheritance from the common
Proto-Transeurasian. Although in most of the languages analyzed in this paper
we observed that the suffix can even combine with native nominals or non-Turkic
nominals to render them verbs, we see the necessity of a systematic analysis and
also leave this issue for further research.

Acknowledgments

We wish to thank Peter Arkadiev, Francesco Gardani, Asl Gksel, Yaron Matras
and one anonymous reviewer for their most constructive remarks; Sarven Akel-
yan, Walter Breu, Ridvan Kiose, Andreas Konstantinidis, Sophia Konstantinidi,
Aivaz Osmantsa, Ergin pengin and Soren Seifi for their help with the data; and
Marieke Krijnen for the map. Needless to say, none of these people can be held

Brought to you by | University of Sydney Library

Authenticated
Download Date | 3/30/17 10:41 PM
130 Metin Barak, Angela Ralli and Dimitra Melissaropoulou

responsible for the way we have used their suggestions and the data they have
provided.

Abbreviations

A. Azerbaijani
acc accusative
aff affirmative
aor aorist
Ar. Arabic
def definite
erg ergative
fut future
ipfv imperfective
infl inflection
i1 1st Series-1st person
lvm loanverb marker
M. Mongolian
npg Greek non-progressive
P. Persian
pst past
pstt Turkish past tense marker
pl plural
poss possessive
post post-terminal
prs present
rep reported
sg singular
T. Turkish
ths thematic suffix
U. Uzbek
val2 valency operator 2
vrbz verbalizer
W.M. Written Mongolian

Brought to you by | University of Sydney Library

Authenticated
Download Date | 3/30/17 10:41 PM
 Borrowing verbs from Oghuz Turkic: two linguistic areas 131

References
Aaryan, Hrachya. 1898. Knnutyun Aslanbegi Barba [The study of Aslanbeg dialect].
Bazmavep 56(1). 3135; 56(2). 8388.
Adamou, Evangelia. 2012. Verb morphologies in contact: Evidence from the Balkan area. In
Martine Vanhove, Thomas Stolz, Aina Urdze & Hitami Otsuka (eds.), Morphologies in
contact, 143162. Berlin: Akademie Verlag.
Aikhenvald, Alexandra Y. 2006. Grammars in contact: A cross-linguistic perspective. In
Alexandra Y. Aikhenvald & Robert M. W. Dixon (eds.), Grammars in contact: A cross-
linguistic typology, 166. Oxford: Oxford University Press.
Aikhenvald, Alexandra Y. & Robert M. W. Dixon (eds.). 2006. Grammars in contact: A cross-
linguistic typology. Oxford: Oxford University Press.
Aksu-Ko, Ayhan. 2000. Some aspects of the acquisition of evidentials in Turkish. In Lars
Johanson & Bo Utas (eds.), Evidentials. Turkic, Iranian and neighbouring languages,
1528. Berlin & New York: Mouton de Gruyter.
Alagozlu, Nuray. 2002. Code-switching patterns and underlying socio-psychological configu-
rations in Kabardian speech community: The probable effects of urbanization. Ankara:
Hacettepe University, Institute of Social Sciences dissertation.
Alagozlu, Nuray. 2007. Code-switching between Kabardian and Turkish. Paper presented at the
6th International Symposium on Bilingualism (ISB6) May 30 June 2, 2007. University of
Hamburg, Germany. Available at: http://www.isb6.org/static/proceedings/alagozlu.pdf.
Bernstein, Samuil B. 1953. Bolgarsko-russkij slovar [Bulgarian-Russian dictionary]. Moskva:
Gosudarstvennoje Izdatelstvo Inostrannyx i nacionalnyx slovarej.
Brendemoen, Bernt. 1990. Laz influence on Black Sea Turkish? In Bernt Brendemoen (ed.),
Altaica Osloensia. Proceedings from the 32nd Meeting of the Permanent International
Altaistic Conference, Oslo, June 1216, 1989, 4964. Oslo: Universitetsforlaget.
Breu, Walter. 1991a. Abweichungen vom phonetischen Prinzip bei der Integration von
Lehnwrtern. In Klaus Hartenstein & Helmut Jachnow (eds.), Slavistische Linguistik 1990,
3669. Mnchen: Otto Sagner.
Breu, Walter. 1991b. System und Analogie bei der Integration von Lehnwrtern. Versuch einer
Klassifikation. Incontri Linguistici 14. 1328.
Bulut, Christiane. 2000. Indirectivity in Kurmanji. In Lars Johanson & Bo Utas (eds.),
Evidentials. Turkic, Iranian and neighbouring languages, 147184. Berlin & New York:
Mouton de Gruyter.
Bulut, Christiane. 2006. Turkish elements in spoken Kurmanji. In Hendrik Boeschoten & Lars
Johanson (eds.), Turkic languages in contact, 95121. Wiesbaden: Harrassowitz.
Csat, va . 2000. m- and m-items. Dimensions of a functional analysis. In Lars Johanson &
Bo Utas (eds.), Evidentials. Turkic, Iranian and neighbouring languages, 2944. Berlin &
New York: Mouton de Gruyter.
Dawkins, Richard M. 1916. Modern Greek in Asia Minor. A study of the dialect of Slli,
Cappadocia and Phrasa with grammar, texts, translations and glossary. Cambridge:
Cambridge University Press.
Donabedian, Anad. 1996. Perfect and mediative in Modern Western Armenian. In Dora Sakayan
(ed.), Proceedings of the Fifth International Conference on Armenian Linguistics, 149166.
Delmar & New York: Caravan.
Dixon, Robert M. W. 1997. The rise and fall of languages. Cambridge: Cambridge University
Press.

Brought to you by | University of Sydney Library

Authenticated
Download Date | 3/30/17 10:41 PM
132 Metin Barak, Angela Ralli and Dimitra Melissaropoulou

Doerfer, Gerhard. 1963. Trkische und mongolische Elemente im Neupersischen: unter

besonderer Bercksichtigung lterer neupersischer Geschichtsquellen, vor allem
der Mongolen- und Timuridenzeit, vol. 1: mongolische Elemente im Neupersischen.
Wiesbaden: Steiner.
Doerfer, Gerhard. 1967. Trkische Lehnwrter im Tadschikischen. Wiesbaden: Steiner.
Doerfer, Gerhard. 1993. Das trkische Suffix mI als Lehnelement. In Jens P. Laut & Klaus
Rhrborn (Hrsg.), Sprach-und Kulturkontakte der trkischen Vlker (Verffentlichungen
der Societas Uralo-Altaica 37), 8793. Wiesbaden: Harrassowitz.
Dorleijn, Margreet. 1996. The decay of ergativity in Kurmanci. Language internal or contact-
induced? Tilburg: Tilburg University Press.
Field, Fredric. 2002. Linguistic borrowing in bilingual contexts. Amsterdam & Philadelphia:
John Benjamins.
Friedman, Viktor. 1989. Toward defining the turkisms in Romani. In Milan ipka (red.), Jezik i
kultura Roma [Romani language and culture], 251267. Sarajevo: Institut za Prouavanje
Nacionalnih Odnosa.
Friedman, Viktor. 1996. The Turkish lexical element in the languages of the Republic of
Macedonia from the Ottoman period to independence. Zeitschrift fr Balkanologie 32(2).
133150.
Gardani, Francesco. 2008. Borrowing of inflectional morphemes in language contact. Frankfurt
am Main: Peter Lang.
Gksel, Asl & Celia Kerslake. 2005. Turkish. A comprehensive grammar. New York: Routledge.
Gukasyan, Voroshil. 1974. Udinsko-azerbajdzhansko-russkij slovar Udinc-azrbaycanca-rusca
lt [Udi-Azerbaijani-Russian dictionary]. Baku: ELM.
Haig, Geoffrey. 2001. Linguistic diffusion in present-day East Anatolia: From top to bottom.
In Alexandra Y. Aikhenvald & Robert M. W. Dixon (eds.), Areal diffusion and genetic
inheritance. Problems in comparative linguistics, 195224. Oxford: Oxford University
Press.
Haspelmath, Martin. 1993. A grammar of Lezgian. Berlin & New York: Mouton de Gruyter.
Hatzidakis, Georgios N. 1905. Meseonika ke Nea Ellinika [Medieval and Modern Greek]. Athens:
Sakellariou.
Heine, Bernard & Tania Kuteva. 2005. Language contact and grammatical change. New York:
Cambridge University Press.
Husi, Geoff. 2002. Albanian verb dictionary and manual. Lawrence, KS: University of Kansas
Libraries.
Igla, Birgit. 1996. Das Romani von Ajia Varvara. Wiesbaden: Harrassowitz.
Jakobson, Roman. 1938. Sur la thorie des affinits phonologiques entre les langues. In Actes
du IV Congrs International des Linguistes, Copenhague, 1936, 4858. Copenhagen: Einar
Munksgaard.
Jger, Andreas. 2004. The cross-linguistic function of obligatory do-periphrasis. In
Proceedings of the 2004 Conference of the Australian Linguistic Society. Available at:
http://ses.library.usyd.edu.au/bitstream/2123/ 111/1/ ALS-20050630-AJ.pdf.
Janse, Mark. 2001. Morphological borrowing in Asia Minor. In Yoryia Aggouraki, Amalia
Arvaniti, J. I. M. Davy, Dionysis Goutsos, Marilena Karyolaimou, Anna Panagiotou, Andreas
Papapavlou, Pavlos Pavlou & Anna Roussou (eds.), Proceedings of the 4th International
Conference on Greek Linguistics (Nicosia, 1719 September 1999), 473479. Thessaloniki:
University Studio Press.

Brought to you by | University of Sydney Library

Authenticated
Download Date | 3/30/17 10:41 PM
 Borrowing verbs from Oghuz Turkic: two linguistic areas 133

Johanson, Lars. 1971. Aspekt im Trkischen. Vorstudien zu einer Beschreibung des trkeitr-
kischen Aspektsystems. Uppsala: Almqvist and Wiksell.
Johanson, Lars. 1999. The dynamics of code-copying in language encounters. In Bernt
Brendemoen, Elizabeth Lanza & Else Ryen (eds.), Language encounters across time and
space. Studies in language contact, 3762. Oslo: Novus forlag.
Johanson, Lars. 2000. Turkic indirectives. In Lars Johanson & Bo Utas (eds.), Evidentials. Turkic,
Iranian and neighbouring languages, 6188. Berlin & New York: Mouton de Gruyter.
Johanson, Lars. 2002. Contact-induced change in a code-copying framework. In Mari C. Jones &
Edith Esch (eds.), Language change. The internal, external and extra-linguistic factors,
285313. Berlin & New York: Mouton de Gruyter.
Johanson, Lars. 2006. Indirective sentence types. Turkic Languages 10. 7287.
Johanson, Lars & Martine Robbeets. 2010. Introduction. In Lars Johanson & Martine Robbeets
(eds.), Transeurasian verbal morphology in a comparative perspective: Genealogy,
contact, chance (Turcologica 78), 15. Wiesbaden: Harrassowitz.
Kazenin, Konstantin, Sandro Kodzasov, Grigorij Strokin & Jakov Testelets. 1999. Caxursko-
russkij slovar [Tsakhur-Russian dictionary]. In Alexander E. Kibrik (ed.), lementy
caxurskogo jazyka v tipologieskom osveenii [Studies in Tsakhur: A typological
perspective], 865890. Moskva: Nasnedie.
Kornfilt, Jaklin. 1997. Turkish. London & New York: Routledge.
Kyuchukov, Hristo. 1998. Turkish and Armenian loanwords in Bulgarian Romani. Grazer
linguistische Studien 50. 103108.
Lacroix, Ren. 2009. Description du dialecte laze dArhavi (caucasique du sud, Turquie)
Grammaire et textes. Lyon: Universit Lyon 2 PhD dissertation.
Lerch, Peter. 18578. Forschungen ber die Kurden und die iranischen Nordchalder, I-II.
St. Petersburg: Eggers et Comp.
Matras, Yaron. 2007. The borrowability of structural categories. In Yaron Matras & Jeanette
Sakel (eds.), Grammatical borrowing in cross-linguistic perspective, 3173. Berlin & New
York: Mouton de Gruyter.
Matras, Yaron. 2009. Language contact. Cambridge: Cambridge University Press.
Matras, Yaron & Jeanette Sakel. 2007. Investigating the mechanisms of pattern replication in
language convergence. Studies in Language 31(4). 829865.
Meillet, Antoine. 1921. Linguistique historique et linguistique generale. Paris: Champion.
Melissaropoulou, Dimitra. 2010. Loan verb adaptation: Evidence from Greek dialectal
variation. In Mark Janse, Brian Joseph, Angela Ralli & Athanasios Karasimos (eds.), Online
proceedings of the 4th International Conference of Modern Greek Dialects and Linguistic
Theory, Chios, Greece, 1114 June 2009, 243258. Available at: http://lmgd.philology.
upatras.gr/el/research/downloads/MGDLT4_Proceedings.pdf.
Melissaropoulou, Dimitra. 2011. n loan verb accommodation: Evidence from Greek dialectal
variation. In Emanuele Miola & Paolo Ramat (eds.), Language contact and language decay:
Socio-political and linguistic perspectives, 165189. Pavia: IUSS Press.
Menges, Karl. 1956. Das ayatajische in der persischen Darstellung von Mirza Mahdi Xan.
Wiesbaden: Steiner.
Mifsud, Manwel. 1995. Loan verbs in Maltese. A descriptive and comparative study.
Leiden: Brill.
Moravcsik, Edith. 1978. Language contact. In Joseph Greenberg, Charles A. Ferguson & Edith A.
Moravcsik (eds.), Universals of human language, vol. 1: Method & theory, 93. Stanford:
Stanford University Press.

Brought to you by | University of Sydney Library

Authenticated
Download Date | 3/30/17 10:41 PM
134 Metin Barak, Angela Ralli and Dimitra Melissaropoulou

Paul, Ludwig. 1998. Zazaki. Grammatik und Versuch einer Dialektologie. Wiesbaden: Reichert.
Ralli, Angela. 2005. Morfologia [Morphology]. Athens: Patakis.
Ralli, Angela. 2006. n the role of allomorphy in inflectional morphology: Evidence from
dialectal variation. In Giandomenico Sica (ed.), Open problems in linguistics and
lexicography, 3554. Milano: Polimetrica.
Ralli, Angela. 2012. Morphologies in language contact: Verbal loanblend formation in Asia
Minor Greek (Aivaliot). In Martine Vanhove, Thomas Stolz, Aina Urdze & Hitomi Otsuka
(eds.), Morphologies in contact, 177194. Berlin: Akademie Verlag.
Ralli, Angela. 2013. Romance verbal loans in Modern Greek dialects. In Mark Janse, Brian
Joseph, Angela Ralli & Metin Bagriacik (eds.), Online Proceedings of the 5th International
Conference on Modern Greek Dialects and Linguistic Theory, Ghent, Belgium, 2022
September, 2012, 430444. Available at: http://lmgd.philology.upatras.gr/el/research/
downloads/MGDLT4_Proceedings.pdf.
Sandfeld, Kristian. 1930 [1921]. Linguistique balkanique. Problmes et rsultats. Paris: Librairie
Ancienne Honor Champion. (First published as Balkanfilologien: en oversight over dens
resultater og problemer, 1921, Copenhagen.)
Sapir, Edward. 1921. Language: An introduction to the study of speech. New York: Harcourt
Brace.
Schnig, Claus. 1998. Azerbaijanian. In Lars Johanson & va . Csat (eds.), The Turkic
languages, 248260. London: Routledge.
Schuchardt, Hugo E.M. 1884. Slawo-deutsches und Slawo-italienisches. Graz: Leuschner &
Lubensky.
Schuchardt, Hugo E. M. 1922. Ein Vademecum der allgemeinen Sprachwissenschaft. Als
Festgabe zum 80. Geburtstag des Meisters zusammengestellt und eingeleitet von Leo
Spitzer. Halle: Niemeyer.
Sechidou, Irene. 2011. Balkan Romani. The dialect of Ajios Athanasios/Greece (Languages of
the World/Materials 481). Mnchen: LINCOM Europa.
kalji, Abdulah. 1966. Turcizmi u srpskohrvatskom jeziku [Turkish loanwords in Serbo-
Croatian]. Sarajevo: Svjetlost Izdavako Preduzee.
Talibov, Bukar B. & Magomed M. Gadzhiev. 1966. Lezgi tsialanni urus tsialan gafalag
Lezginsko-russkoj slovar [Lezgian-Russian dictionary]. Moskva: Izdatelstvo Akademii
Nauk SSSR.
Thomason, Sarah. 2001. Language contact: An introduction. Edinburgh: Edinburgh University
Press.
Trubetzkoy, Nikolai. 1999 [1925]. Nasledie ingisxana [The legacy of Genghis Khan]. Moskva:
Agraf.
Uygun, Dilek. 2009. A split model for category specification: Lexical categories in Turkish.
Istanbul: Boazii University PhD dissertation.
Vaux, Bert. 2001a. Hemshinli: The forgotten Black Sea Armenians. Journal of Armenian Studies
6(2). 4771. Available at: https://dl.dropboxusercontent.com/u/476232/my%20papers/
vaux%20jas%202001.pdf.
Vaux, Bert. 2001b. The Armenian dialect of Aslanbeg. Annual of Armenian Linguistics 21.
3164.
Vaux, Bert. 2007. Homshetsma: The language of the Armenians of Hamshen. In Hovann
Simonian (ed.), The Hemshin. History, society and identity in the highlands of Northeast
Turkey, 257278. London & New York: Routledge.

Brought to you by | University of Sydney Library

Authenticated
Download Date | 3/30/17 10:41 PM
 Borrowing verbs from Oghuz Turkic: two linguistic areas 135

Wackernagel, Jacob. 19268. Vorlesungen ber Syntax: Mit besonderer Bercksichtigung von
Griechisch, Lateinisch und Deutsch. Basel: Birkhuser.
Weinreich, Uriel. 1968. Language in contact: Findings and problems. 2nd edn. The Hague:
Mouton.
Wichmann, Sren & Jan Wohlgemuth. 2008. Loan verbs in a typological perspective. In
Thomas Stolz, Dik Bakker & Rosa Salas Palomo (eds.), Aspects of language contact. New
theoretical, methodological and empirical findings with special focus on romancisation
processes, 89121. Berlin & New York: Mouton de Gruyter.
Windfuhr, Gernot. 2005. Persian. In Bernard Comrie (ed.), The major languages of South Asia,
the Middle East and Africa, 81100. London: Routledge.
Windfuhr, Gernot & John R. Perry. 2009. Persian and Tajik. In Gernot Windfuhr (ed.), The Iranian
languages, 416544. London & New York: Routledge.
Wodarg, Anke. 1995. Textsammlung. In Silvia Kutscher, Johanna Mattissen & Anke Wodarg
(eds.), Das Mutafi-Lazische, 113132. Kln: Universitt Kln.
Wohlgemuth, Jan. 2009. A typology of verbal borrowings. Berlin & New York: Mouton de
Gruyter.
Wurm, Stephen. 1996. Language situation, language contacts and contact languages in
the Caucasus area. In Stephen Wurm, Peter Mhlhuser & Darrell Tryon (eds.), Atlas of
languages of intercultural communication in the Pacific, Asia and the Americas, 955958.
Berlin & New York: Mouton de Gruyter.

Brought to you by | University of Sydney Library

Authenticated
Download Date | 3/30/17 10:41 PM
Brought to you by | University of Sydney Library
Authenticated
Download Date | 3/30/17 10:41 PM