Boğaç Ergene, Atabey Kaygun - Semantic Mapping of An Ottoman Fetva Collection (2021)

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 54

Journal of Islamic Studies 32:1 (2021) pp. 62–115 doi:10.

1093/jis/etaa032

SEMANTIC MAPPING OF AN OTTOMAN


FETVA COMPILATION: EBUSSUUD EFENDI’S
JURISPRUDENCE THROUGH A
COMPUTATIONAL LENS

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


B O Ğ AÇ E RG E N E
University of Vermont

ATA BEY KAYG UN*


Istanbul Technical University

INTRODUCTION

In this article, we propose computational methodologies to analyse the


fetva compilation by Şeyhülislam Ebussuud Efendi (d. 1574), a corpus
that contains about 6,000 rulings concerning all aspects of life. More
specifically, we aim to provide a semantic mapping of Fetava-yı
Ebussuud to reveal its substantive composition. The computational
approach that we introduce here captures the relative sizes of the text’s
semantic components and their textual relationships with others. Our
goal is to offer ways to conceptualize a previously uncharted textual
space, in much the same way as a map depicts a geographical one.
Fetva collections are important sources for Islamic legal history. In the
Ottoman context, they include the legal questions that renowned jurists
received, and the answers they provided based on their legal interpreta-
tions. In modern scholarship on Islamic law, it is common practice to
consult small groups of fetvas in the exploration of various topics. For
example, researchers who focus on marriage and gender issues, property
relations, and crimes and punishment habitually sift through selections
of relevant fetvas to understand the finer points of jurisprudential
interpretation on these issues. However, to our knowledge there are no
studies devoted to a single collection of opinions or the entirety of

* Authors’ note: The authors are indebted to Febe Armanios, Safa Saraçoğlu,
Malissa Taylor, and to this Journal’s anonymous referees for their comments
and suggestions on the earlier version of the article. They are also grateful to
Charlotte Weber for her editorial and stylistic assistance.

ß The Author(s) (2020). Published by Oxford University Press on behalf of the Oxford Centre for
Islamic Studies. All rights reserved. For permissions, please email: journals.permissions@oup.com
MAPPING A FETVA COMPILATION 63
opinions by a particular jurist. In other words, fetva compilations have
yet to be studied as ‘corpora’, in and of themselves.
Finding ways to address this gap could contribute to Islamic legal
scholarship in multiple ways, one of which is to provide enhanced in-
formation on important jurisprudential texts. Many fetva compilations,
including the one explored here, were highly regarded and frequently

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


cited sources for Ottoman jurists across generations, which is why they
remain important for modern scholarship on Ottoman–Islamic legal his-
tory. Yet there exists scant information on these compendia, in particular
their specific content and substantive coverage. Although scholars might
presume to have an impression of these works based on their knowledge
of premodern jurisprudential traditions or form an opinion on them
after a cursory perusal of their pages, a full understanding would require
a substantial investment of time and labour. The procedures that
we propose in this article make possible detailed and comprehensible
depictions of the manuscripts’ contents and render them accessible with
significantly less effort.
In a way, the analysis that we describe in the following pages is analo-
gous to indexing a corpus in that it identifies important concepts by
determining their relative prevalence in the text, and by doing so,
uncovers the corpus’ inner structure. Experienced book indexers often
describe their work as providing more than a list of words and page
numbers. Rather, their main contribution is to reveal the entrails of
the work so that the reader can quickly grasp its substance. While the
procedures that we propose here are significantly more capable of iden-
tifying the semantic structure of the Fetava compared to indexing, the
analogy is nevertheless a useful way to characterize how we render the
contents and inner structure of one important, previously unanalysed
jurisprudential source in its entirety, comprehensible to legal specialists
and lay readers.
In addition, the methodologies we propose could provide insights into
the jurisprudential legacies of important jurists. Indeed, a comprehensive
understanding of the various issues that a jurist considered during his
career would provide valuable information about his scholarly range and
interests. There certainly exists significant research on the contributions
of a handful of Ottoman jurists. Ebussuud’s opinions, in particular, have
received considerable attention in modern scholarship because of their
importance in shaping many aspects of the Ottoman legal system and
practices, specifically the land law and the institution of cash-waqf. Yet
based on the existing literature, one would be hard-pressed to quantify
how much individual jurists contributed to various areas of jurispru-
dence, or to assess how the volume of their interpretations related to
certain issues might compare to those on others. Specifically for
64 B O Ğ A Ç E R G E N E A N D A T A B E Y K A Y G U N

Ebussuud, one might wonder how his opinions on land- and


waqf-related matters match the bulk of his opinions on other aspects
of Islamic law, say family law, crime, and commerce. Relatedly, one
might wonder if there are hitherto underexplored aspects of his scholarly
output that are comparable in size or significance to the elements that
have already received significant attention.

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


The exercise might be particularly useful in Ebussuud’s case since,
despite his purported importance for Ottoman legal history, the jurist
never composed a comprehensive jurisprudential text and, according to
Düzenli,1 his jurisprudential contributions and import can only be
gleaned through his fetvas.2 While a full understanding of any jurist’s
contributions to Islamic law would require an in-depth engagement with
his articulations on specific issues (specifically, ‘what they say’), the ana-
lysis that we offer here can be regarded as a step towards a comprehen-
sive assessment of the legacy of an important jurist such as Ebussuud. It
is also one that could render Ebussuud’s opinions comparable, at least in
terms of their volume, to those of other prominent jurists, provided that
one could analyse the latter’s jurisprudential output in the same way.
Finally, the approaches we offer in this article could help us better
appreciate the socio-legal currents of the time as reflected in jurispru-
dential opinions. Since fetvas can be regarded as responses to socio-legal
stimuli, they can be analysed to extract information about the types of
issues that occupied both the populace and legal authorities. Our explor-
ation of a large and influential fetva corpus should thus permit a peek
into the societal tensions prevalent in its historical setting. To the extent
that our endeavour succeeds, the approaches and techniques introduced
in the following pages present a novel way to use a legal source in its
entirety for a more broadly defined social history.
Our analysis is based solely on Ebussuud’s jurisprudential opinions.
But the full potential of the analysis proposed here from a socio-legal
perspective would be more apparent if we applied it to multiple compi-
lations from different periods to discern how popular concerns and
apprehensions may have changed over time. We plan to pursue such a
comparative analysis in a subsequent publication. In the present article,
however, our purpose is primarily to outline the details of a methodology
that is capable of fulfilling this potential.

1
Pehlul Düzenli, ‘Osmanlı hukukçusu Şeyhülislâm Ebussu^ ud Efendi ve
Fetvâları’, (PhD diss., Konya Selçuk University, 2007), 38.
2
The jurist did compose shorter treatises with limited substantive focus on
specific legal issues, such as charitable endowments and various forms of land
tax. See Pehlul Düzenli, ‘Şeyhülislâm Ebussu^ ud Efendi: Bibliyografik Bir
Değerlendirme’, Türkiye Araştırmaları Literatür Dergisi, 3/5 (2005): 441–75.
MAPPING A FETVA COMPILATION 65
In principle, exploring a fetva collection in a comprehensive and
meticulous manner can be accomplished by reading each fetva and
assessing its thematic attributes—that is, what issues, concerns, notions
it raises—on a case-by-case basis. Such a method would also require the
assessor(s) to employ a categorization scheme in a consistent fashion
for thousands of fetvas. Thus, manual thematic labeling of fetvas in a

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


compilation is considerably labour-intensive, especially given the peri-
odic need to check and, often, revise the results in the process of cat-
egorization due to human error. It is no wonder, then, that scholars
have yet to produce works that contain fetva-based indexes for the
existing fetva corpora. In this preliminary study, it is our objective to
offer computational methodologies to engage one corpus to extract
information in a fashion that is not only reliable and historically ap-
propriate but also ‘affordable’ in terms of the labour it requires. We
also intend our procedures to be replicable in other corpora, which
would allow us and other researchers to easily perform comparative
cross-corpora analyses.

EBUSSUUD AND HIS FETAVA FOR


COMPUTATIONAL ANALYSIS

2.1 Contextual background on the era, jurist and the corpus


2.1.1 Historical context
Ebussuud Efendi, one of the well-known and distinguished premodern
Hanafi scholars, served as the şeyhülislam of the Ottoman Empire during
the reigns of two sixteenth-century sultans, Süleyman I (r. 1520–66) and
Selim II (r. 1566–74). Historians attribute to this era distinct religious,
political-administrative, and economic tendencies. By the middle of the
sixteenth century, the Ottoman Empire had reached its territorial peak,
occupying a landmass that extended from central Europe in the West to
the Caspian Sea in the east, from Crimea in the north to Yemen in the
south. It also controlled the Black Sea, eastern Mediterranean, Red Sea,
and competed with the Portuguese for the control of the Arabian Sea and
the Indian Ocean.3 The geographical reach of the empire helped connect
distinct economic zones and contributed to the proliferation of long-
distance and regional commerce in and outside of the empire’s lands.

3
See Giancarlo Cassale, The Ottoman Age of Exploration (New York: Oxford
University Press, 2010).
66 B O Ğ A Ç E R G E N E A N D A T A B E Y K A Y G U N

According to Pamuk,4 the Ottoman economy became increasingly


monetized and commercialized during this period, fueled by an increas-
ing availability of bullion on the global scale and the emergence of a
network of markets and fairs where peasants, landholders, and nomads
sold products to city dwellers. Pamuk also observed that credit networks
emerged in and around the urban centres and that credit-use was wide-

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


spread among all segments of the society.5 On the other hand, inflation-
ary trends remained relatively stable until the imperial government
debased the akçe in the mid-1580s.6
The sixteenth century is also a period of intense demographic expan-
sion.7 Although there is some debate on the issue, this tendency might
have generated population pressure and landlessness in the countryside,
and intensified migration to urban centres in the latter part of the cen-
tury. Alongside the climactic difficulties at the turn of the seventeenth
century, the demographic pressure would also contribute to the instiga-
tion of the Celali rebellions that devastated the central lands of the em-
pire. According to Genç,8 however, the government remained vigilant
about maintaining the socio-economic status quo, by ensuring the flow
of a steady supply from elsewhere of goods and services needed by the
urban populations at affordable, if at times artificially low, prices, and by
attempting to keep in check rural immigration to the cities.
In the realm of government and politics, the sixteenth century con-
stitutes a special era. It is during this period that Selim I and Süleyman I,
two of the most celebrated rulers of the dynasty, were in power; when
the Empire had the military upper-hand against its regional rivals in the
Mamluk, Safavid, and Habsburg polities; and when, according to most
historians, the polity’s legal, bureaucratic, and administrative institu-
tions fully matured. The latter development corresponds to the emer-
gence of highly bureaucratized imperial and provincial administrations,
the development of complex land-tenure and tax systems, and the

4
Şevket Pamuk, A Monetary History of the Ottoman Empire (Cambridge:
Cambridge University Press, 2000), 9–15 and id., ‘The price revolution in the
Ottoman Empire reconsidered’, International Journal of Middle East Studies, 33
(2001): 69–89, at 73.
5
Pamuk, ‘Price revolution’, 71.
6
Ibid, 78.
7
See Oktay Özel, ‘Population changes in Ottoman Anatolia during the 16th
and 17th centuries: the ‘‘demographic crisis’’ reconsidered’, International Journal
of Middle East Studies, 36 (2004): 183–205 and id., The Collapse of Rural Order in
Ottoman Anatolia: Amasya 1576–1643 (Leiden: Brill, 2016).
8
Mehmet Genç, ‘Economy and economic policy’ in Gabor Agisron and Bruce
Masters (eds.), Encyclopedia of the Ottoman Empire (New York: Facts on File,
2009), 192–5.
MAPPING A FETVA COMPILATION 67
articulation of a legal structure based on (initially disparate) Islamic
jurisprudential and dynastic legal traditions. What makes these trends
potentially relevant for the present study is the central role that
Ebussuud played in many of them. He was a leading figure in the
empire’s increasingly growing, systematized, and influential religio-
administrative (ilmiye) hierarchy. He is also recognized as the legal

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


mind behind many of the legislative initiatives pursued at the time,
including the formulation of the imperial land and tax codes and the
harmonization of the Ottoman kanun with inherited shari‘a-based
considerations.9
In terms of the prevalent social, religious, and cultural concerns of the
era, one should point out the issues associated with the religious and
ethnic plurality in the Ottoman lands and the consequent attempts by
those in power to manage this diversity. Included among these are ques-
tions pertaining to Muslim, Christian, and Jewish relations, boundaries
between religious orthodoxy and heterodoxy as related to specific Sufi
practices and/or folk beliefs, and the statuses of those communities (in
particular the Alevi in the Anatolian countryside) associated with
Shi6ism.10 The latter issue might be related to the Ottomans’ increasingly
hardening Sunni identity, which was a consequence of the Safavid chal-
lenge on the polity’s eastern borders. Burak has also claimed that from
the sixteenth century onwards, the Anatolian and Balkan jurists of the
empire, including Ebussuud, made efforts to formulate an ‘official’
Ottoman brand of Hanafi Sunnism, distinct from earlier traditions and
contemporary interpretations in the peripheries of the empire.11 Other
legal trends that received jurisprudential attention during the sixteenth
century include the status of cash-waqfs,12 permissibility of the growth
and consumption of coffee, and the legality of listening to music and
dancing as entertainment or as parts of Sufi practices, and the rights of
foreign residents in Ottoman lands.13

9
See Colin Imber, Ebu’s-su6ud: The Islamic Legal Tradition (Edinburgh:
Edinburgh University Press, 1997).
10
See Pehlul Düzenli, ‘Şeyhülislam Ebussu^ ud Efendi Fetvâları Işığında Osmanı
Sünniliği’, Marife, 5/3 (2005): 259–86 and Kaya Şahin, ‘The Ottoman Empire in
the long sixteenth century’, Renaissance Quarterly, 70 (2017): 220–34.
11
Guy Burak, The Second Formation of Islamic Law: The Eanafi School in the
Early Modern Ottoman Empire (New York: Cambridge University Press, 2015).
12
That is, whether religious-charitable foundations could be established by
money endowments and whether they could lend money to borrowers for specific
periods of time and accrue capital gains, i.e., interest, in return.
13
Düzenli, ‘Ebussu^ ud Efendi ve Fetvâları’, 10-7.
68 B O Ğ A Ç E R G E N E A N D A T A B E Y K A Y G U N

While the trends identified in the above discussion dominate the


literature on the period, it is difficult to appreciate how deeply they
were felt by various segments of the Ottoman society or how much
they occupied the minds of those who represented religion. A parsing
of the corpus’s contents might provide clues in these regards.
Alternatively, the analysis might also reveal the concerns that have hith-

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


erto been ignored in the scholarly literature.

2.1.2 The jurist and his work


Ebussuud Efendi14 was born between 1490 and 1492, possibly in
Istanbul, to a prominent family with notable religio-legal credentials.
After completing his education, he rose fast in the religious-
administrative establishment. Following a period of teaching in various
religious colleges, he served as the judge (kadı) of Bursa and Istanbul and
military-judge (kadıasker) of Rumelia before being appointed as the
chief müfti or şeyhülislam, the top position in the Ottoman religio-
legal hierarchy, in 1545. He would serve in this position until his death
in 1574.
Ebussuud Efendi was a prolific jurist. His writings include exegetical
(tefsir) studies of the Qur8:n, jurisprudential commentaries, dictionaries,
and treatises on mysticism, Arabic grammar, literature, and poetry. But
he is best known for the thousands of fetvas that he composed during his
three-decade-long career as the highest-ranking judicial official of the
most important Sunni polity in the early modern period. As is well
attested in the scholarship on the topic, his fetvas played a major role
in the systematization of Ottoman land and tax systems and aligning
them with Hanafi jurisprudential traditions. They were also heavily cited
and quoted by generations of legal scholars who came after him.
No one knows how many fetvas Ebussuud issued. Düzdağ claimed to
have seen more than 10,000 by the jurist.15 Düzenli identified 7,185 in
various manuscript libraries and archives of Turkey.16 There might be
more given the fact that Ebussuud served for decades as the chief müfti
and that there were days on which he issued hundreds of fetvas.17

14
Düzenli gives the full name of the scholar as ‘Muhammed Ahmed İmadüddin
İbnü’ş-Şeyh Muhyiddin b. Muhammed b. Mustafa el-İskilibi el-İmadi, el-Hanefi’;
id., ‘Bibliyografik’, 441. It is not clear how the designation ‘Ebussuud’ came to be
associated with the jurist.
15
Mehmet Ertuğrul Düzdağ, Şeyhülislâm Ebussuud Efendi Fetvaları Işığında
16. Asır Türk Hayatı (İstanbul: Enderun Kitabevi, 1972), 8.
16
Düzenli, ‘Ebussu^ud Efendi ve Fetvâları’, x.
17
According to one frequently-cited report, he issued more than 1,400 fetvas in
one particularly busy day; see Emine Arslan, ‘Nuk^ ullü Fetva Mecm^uaları ve
MAPPING A FETVA COMPILATION 69
He could not have been this prolific without assistance. In fact,
sources indicate that in the course of the sixteenth century the office of
the şeyhülislam gained importance, grew in size, and acquired a highly
bureaucratic character, a development in which Ebussuud himself played
a role.18 One of the most important responsibilities of the office was to
issue fetvas, a task that involved many staff members with distinct

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


responsibilities. During Ebussuud’s time questions were drafted by
scribes (sing. fetva katibi) with legal training and experience, according
to pre-determined technical and rhetorical standards.19 The clerks of
the şeyhülislam (sing. fetva emini), who were legal scholars sometimes
with high-level madrasa appointments, might have provided provisional
answers to these questions before they were presented to the chief müfti,
sometimes after conducting research in the jurisprudential literature. It
was the şeyhülislam’s responsibility to consider the question as drafted
by the scribes and the clerks’ response if available, and then provide an
opinion based on this material. This high level of bureaucratic division
of labour allowed the jurist to issue large numbers of fetvas in relatively
short periods of time.
Modern legal scholarship has generated a considerable body of
research based on Ebussuud’s jurisprudential writings, specifically his
fetvas. In particular, scholars have focused on his efforts in and contri-
butions to the systemization of Ottoman jurisprudence, his formulations
on land ownership and possession, taxation, cash-waqfs, status of non-
Muslims, family law, and crime and punishment.20 In these studies, and
Mehmed Fıkhı̂’nin el-Ecvibetü’l-Kâni‘a Adlı Eserinin Bunlar Arasındaki Yeri’,
(PhD diss., Marmara University, 2010), 15.
18
See Düzenli, ‘Ebussu^ ud Efendi ve Fetvâları’, 71–2; Ferhat Koca, art.
‘Fetvahâne’ in Türkiye Diyanet Vakfı İslam Ansiklopedisi (İstanbul: Diyanet
Vakfı, 1995): xii. 496–500; and Mehmet İpşirli, art. ‘Şeyhülislâm’ in Türkiye
Diyanet Vaqfı İslam Ansiklopedisi (İstanbul: Türkiye Diyanet Vakfı, 2010): xxxix.
91–6.
19
Ebussuud himself composed a manual, titled ‘Advice to Fetva Scribes’ (Fetva
Katiblerine Tenbih), about how to draft fetvas in ways that would allow queries to
be answered precisely, unequivocally, and in the fewest possible words: see Düzenli,
‘Ebussu^ ud Efendi ve Fetvâlari’, 72.
20
Overlapping lists of these studies, in multiple languages, could be found in
Düzenli, ‘Bibliyografik’ and Ahmet Akgündüz (ed.), Şeyhü’l-islaâm Ebüssu’^ ud
Efendi Fetvâları (Fetâvâ-yi Ebüssu’^ud Efendi) (İstanbul: Osmanlı Araştırmaları
Vakfı, 2018), 19–40. But also see, for example, Düzdağ, 16. Asır; Imber, Ebu’s-
su6ud; Tayyib M. Gökbilgin, ‘Ebussu^ ud Fetvalarında ve 16. Asır Şer’iyye
Sicillerinde İsbat ve Şehadet’, İslam Tetkikleri Enstitüsü Dergisi, 3/1–2 (1960):
117–32; Eugenia Kermeli, ‘Ebu’s Su‘^ ud’s definitions of church vakfs: theory and
practice in Ottoman law’ in Robert Gleave and Eugenia Kermeli (eds.), Islamic
Law, Theory and Practice (London: I.B. Tauris, 1997), 141–56; Richard Repp, The
70 B O Ğ A Ç E R G E N E A N D A T A B E Y K A Y G U N

for good reasons, researchers explored only specific groups of fetvas


directly relevant to their interests. To our knowledge, the present study
is the first to provide a comprehensive depiction of Ebussuud’s jurispru-
dential output as a müfti, one that can demonstrate, in some detail, how
or to what extent particular socio-legal themes occupied his attention.

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


2.1.3 Corpus
Ebussuud’s work contains thousands of fetvas covering almost all
aspects of law and legal interpretation during his time. Scholars refer
to various collections of the jurist’s opinions as Mecmuatü’l Fetava
(‘Collection of Fetvas’) by Ebussuud, Fetava Ebussuud, or Fetava-yı
Ebussuud. Multiple versions of various collections of Ebussuud’s fetvas
compiled by different individuals exist in Turkish archives and manu-
script libraries (see below). In this study, we examine the first complete
transliteration of the Fetava into the Latin alphabet by Ahmet Akgündüz
and his team, which is based on the original version compiled by Veli bin
Yusuf, a secretary of Ebussuud, in the late sixteenth century, shortly after
Ebussuud’s death.21 In addition to Ebussuud’s (both oral and written)
fetvas, this compilation also includes selected fetvas from previous and
contemporaneous Ottoman müftis such as Kemalpaşazade (also known
as Ibn Kemal), C¸ivizade, Sadi Efendi, and Zenbilli Ali Cemali Efendi,
presumably interpretations that Ebussuud leaned on or agreed with in
his own opinions. It thereby provides a significantly comprehensive de-
piction of the issues that occupied Ottoman legal minds in the fifteenth
and sixteenth centuries. In their effort to identify the most complete and
error-free version of Veli bin Yusuf’s compilation, Akgündüz and his
team compared the versions contained in multiple libraries and archives
in Turkey.22

Müfti of Istanbul: A Study in the Development of the Ottoman Learned Hierarchy


(London: Ithaca Press, 1986); and Ahmet Akgündüz, İslâm Hukukunda ve
Osmanlı Tatbikatında Vakıf Müessesesi (Ankara: Osmanlı Araştırmaları Vakfı
Yayını, 1988).
21
See Akgündüz, Şeyhü’l-islâm Ebüssu’^
ud Efendi Fetvâları, which is based on a
version of the compilation from the Süleymaniye Library in Istanbul (İsmihan
Sultan collection, ms. n. 223 with a copy date of 1575). Akgündüz suggests that
this version is complete: ibid, 4 and 41. Akgündüz and his team compared their
version to two others, one found in the Şehid Ali Paşa collection (ms. n. 1069) also
in the Süleymaniye Library and the other from the Türkiye Yazma Eserler Kurumu
collection (ms. n. 19 HK 1696): ibid, 4.
22
Ibid. Another well-known compilation of the Fetava by Bozanzade Mahmud
bin Kadı Bozan, a sixteenth-century scribe who also worked for Ebussuud, has
never been transliterated into modern Latin alphabet in full, which made it unsuit-
able for our analysis for the reasons we explain below. Unlike that of Veli bin Yusuf,
MAPPING A FETVA COMPILATION 71
Düzenli identified hundreds of manuscript collections comprised of
Ebussuud’s fetvas in Turkish libraries and archives.23 And there are sig-
nificant variations among these in terms of their contents, thematic
coverage, and internal organization.24 This is true even for different
versions of the same compilation. For example, the versions of the com-
pilation by Veli bin Yusuf that Düzenli and Akgündüz relied on differ in

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


terms of the numbers of fetvas that they contain and organization of the
material.25 The version of the compilation we analyse in this article
follows an organizational plan that provides clues about how Veli bin
Yusuf (and maybe others involved in the production of this version of
text) considered its overall contents, as is clear in the table of contents
presented in Akgündüz’s volume (based on Veli bin Yusuf’s plan of the
corpus), representing the primary components of the text. Accordingly,
the Fetava is divided into the following sections:26

Fetvas on cleanliness Fetvas on gifts (hibe)


Fetvas on prayer and its requirements Fetvas on lease/rent contracts
Fetvas on almsgiving Fetvas on slaves with contracts of
Fetvas on land reclamation (ihya-i manumission (mükateb köleler)
mevat) Fetvas on guardianship
(continues)

his compilation contains fetvas issued exclusively by Ebussuud. See Düzenli,


‘Bibliyografik’, 457; Akgündüz, Şeyhü’l-islâm Ebüssu’^ ud Efendi Fetvâları, 36
and 40; and id., art. ‘Fetâvâ-yı Ebüssu^
ud Efendi’ in Türkiye Diyanet Vakfı İslam
Ansiklopedisi (İstanbul: Türkiye Diyanet Vakfı, 1995): xii. 441–5, at 442.
23
Düzenli, ‘Ebussu^ ud Efendi ve Fetvâları’, 3.
24
Cf. ibid, 82.
25
The versions of the compilation that Düzenli examined in his ‘Ebussu^ ud Efendi
ve Fetvâları’ contain fewer fetvas compared to the version that Akgündüz and his team
worked on. There are also discrepancies among the primary section and subsections in
different versions. For example, one version of Veli bin Yusuf’s compilation that
Düzenli studied (ms. n. 178, vr. 2b–233b, in the Library of the Office of the Müfti
in Istanbul) contains sections, among others, on ‘Trees’, ‘Milk Nursing’, ‘Land Tax’
that either do not appear in Akgündüz’s version or they appear as subsections. On the
other hand, the version that Akgündüz and his team transliterated also contains
sections that either do not exist in versions that Düzenli examined or that appear in
them as subsections. These include sections on ‘the contracts between orchard owners
and tenants’, ‘water rights’, ‘blood-money’, and ‘homicide’. The latter two appear as
subcategories under ‘crimes against life and property’ in Düzenli’s versions.
26
A comparison of the table of contents of the corpus that Akgündüz placed at
the beginning of his volume and the text that he and his team transliterated con-
tained inconsistencies. The list of primary sections presented here is based on the
transliterated text.
72 B O Ğ A Ç E R G E N E A N D A T A B E Y K A Y G U N

Fetvas on fasting Fetvas on compulsion


Fetvas on obligatory and non- Fetvas on legal incapacity (hacr)
obligatory pilgrimage Fetvas on slaves and minors who are
Fetvas on marriage contract granted authority to possess
Fetvas on divorce property
Fetvas on manumission of slaves Fetvas on wrongful taking (gasb)

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


Fetvas on oaths and vows Fetvas on preemption (şufa) rights
Fetvas on hadd crimes and their Fetvas on division [of estates]
punishments Fetvas on agricultural contracts be-
Fetvas on theft and its punishments tween land owners and cultivators
Fetvas on holy war (cihad) (müzaraa)
Fetvas on lost property and found Fetvas on contracts between orchard
children (lakit ve lukata) owners and orchard tenants
Fetvas on escaped slaves involving the shares of the orchard’s
Fetvas on lost persons yield (müsakat)
Fetvas on partnership contracts Fetvas on animals to be slaughtered
Fetvas on the charitable foundations Fetvas on sacrifices [of animals]
Fetvas on purchases and sales Fetvas on things and practices con-
Fetvas on the expiration of debt (vade) sidered as discouraged (mekruh) in
Fetvas on suretyship Islamic law
Fetvas on assignments of debt (havale) Fetvas on land improvement [repeat]
Fetvas on judges and jurisconsults Fetvas on beverages
Fetvas on witnessing Fetvas on pledges (rehin)
Fetvas on agency representation Fetvas on crimes [against property and
(vekalet) life] (cinayat)
Fetvas on litigation Fetvas on water use rights (hakk-ı şirb)
Fetvas on acknowledgment (ikrar) Fetvas on blood-money (diyet)
Fetvas on amicable settlements Fetvas on homicide
Fetvas on investment partnerships Fetvas on wills
(müdarabe) Fetvas on hermaphrodites
Fetvas on safekeeping (vedia) Fetvas on various issues and on Islamic
contracts religion
Fetvas on loans for use (ariyet) Fetvas on acts of God (kaza ve kader)

The list above provides an impression of the overall thematic makeup


of the corpus as Veli bin Yusuf structured it. Many of the sections in the
list contain subdivisions, which help enhance the reader’s substantive
understanding of the text. For example, the section titled ‘fetvas
on prayer’ includes 18 subsections, all presumably related to prayer,
including ‘fetvas on the call to the prayer’, ‘direction of the prayer’,
‘times of the prayer’, ‘leading prayer services’ (or ‘prayer leadership’,
MAPPING A FETVA COMPILATION 73
i.e., imamet), ‘circumstances and actions that invalidate the prayer’,
‘Friday prayers’, ‘places where prayer may take place’, ‘prayers per-
formed while travelling’, ‘funeral prayers’, and ‘prayers for the ill’,
among others.27 Based on Akgündüz’s table of contents one can also
form a vague idea of the relative sizes of the text’s components. The
number of fetvas in each section, although not provided in Akgündüz’s

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


edition, could be determined by counting them individually, and presum-
ably corresponds roughly to the number of subsections in each section.28
Yet, although useful in certain ways,29 a ‘mapping’ of the corpus based
solely on the above information would have serious shortcomings. Some
of these are related to the organizational structure of the corpus. While
clearly situated within premodern jurisprudential terminology and
genre-specific organizational conventions (including the headings and
the general order in which they are presented in most fiqh works), the
corpus seems also to have been shaped partly by the compiler’s or later
copiers’ subjective decisions. This is evident in the fact that different
compilations in multiple versions contain significant variations.30 It is
also evident in a few idiosyncratic attributes of the text we examine. For
example, the compilation contains two separate sections on land reclam-
ation (ihya-i mevat) and its legal consequences (composed of different
but related fetvas), which must be due to an oversight. Another type of
redundancy is the presence of multiple fetva concentrations on ‘witness-
ing’ in different sections.31 Likewise, the reason for the existence of

27
The number of fetva sections with multiple subsections is 28. The number of
subsections in a single section ranges between 2 (for example, ‘fetvas on amicable
settlements’, ‘fetvas on holy war’, ‘fetvas on theft’) and 30 (‘fetvas on divorce’).
28
The largest sections are ‘fetvas on divorce’, ‘fetvas on charitable foundations’,
‘fetvas on homicide’, ‘fetvas on prayer’, ‘fetvas on judges and jurisconsults’, ‘fetvas
on witnessing’, and ‘fetvas on buying and selling’.
29
Düzenli attempted to determine the substantive range and concentration of
Ebussuud’s fetvas based on this type of approach. By counting the numbers of fetvas
listed in specific sections of four separate collections, Düzenli identified the follow-
ing thematic distribution in Ebussuud’s fetvas: amicable settlements: 106; obliga-
tory alms: 125; manumission of slaves: 135; guardianship: 138; tax on land or from
non-Muslims: 140; compensation for bodily harm or damages to property: 163;
gifts: 218; dowry: 227; judge-ship: 233; litigation: 247; lease or rental contracts:
271; divorce: 296; discretionary punishments: 302; witnessing: 526; sale or pur-
chase contracts: 560; prayer: 600; waqf: 655; issues related to belief and mysticism:
666. Unfortunately, these figures contain repeated fetvas in the surveyed collec-
tions. See Düzenli, ‘Ebussu^ ud Efendi ve Fetvâları’, 94–5.
30
See note 25.
31
A subsection on witnessing in ‘fetvas on homicides’ contains fetvas not
included in ‘fetvas on witnessing’.
74 B O Ğ A Ç E R G E N E A N D A T A B E Y K A Y G U N

multiple sections on slavery is not clear.32 Moreover, in some cases the


fetvas included in particular sections appear to be out of place. For ex-
ample, it is difficult to explain why the following fetva was included in
the section on cinayat, which refers to serious crimes against property
and life, excluding debt and loan disputes and financial disagreements:

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


Question: If Zeyd pledges (rehin koyma) his house to a charitable foundation
[in return for a loan] and [pledges it] again multiple times [to others?] during
his health, and then he dies and leaves behind no other property, can the
administrators of the [first?] foundation sell the house and claim the profits
after settling the debt?
Answer: If he pledged the house to the first foundation and surrendered it
according to the law and if he received loans [from others?], the administrator
of the [first] foundation may claim the entire yield [generated in the sale]. If
the yield is greater than the debt, the excess may be claimed by the others; if
[the property] was not originally surrendered, all parties share the yield.33

32
A preliminary consideration of various Ottoman fetva compilations by dif-
ferent jurists indicates a generally consistent organizational structure across pre-
modern corpora, comprised of parallel categorizations and rough orders of
presentation. It is, thus, common across the genre to begin with opinions on ritual
matters (including purification, prayer, fasting, and almsgiving), followed by those
on marriage and divorce, crime and punishment, slavery and manumission, buying
and selling, and so on. Yet, a comparison of Ebussud’s corpus that we examine with
two later corpora, Behcetü’l Fetâvâ by Yenişehirli Abdullah Efendi (d. 1743) and
Neticetü’l Fetava, composed of opinions formulated by a select group of
eighteenth-century şeyhülislams also indicated organizational variations, parallel
to those we identified across different campilations of Ebussuud’s fetava and their
versions. See Yenişehirli Abdullah Efendi, Behcetü’l Fetâvâ (eds. Süleyman Kaya,
Betül Algın, Zeynep Trabzonlu, Asuman Erkan; İstanbul: Klasik, 2011) and es-
Seyyid Ahmed Efendi and es-Seyyid Hafız Mehmed b. Ahmed el-Ged^ usı̂ (eds.),
Neticetü’l Fetâvâ: Şeyhülislam Fetvaları (eds. Süleyman Kaya, Betül Algın, Ayşe
Nagehan C¸eliçi, Emine Kaval; İstanbul: Klasik, 2014). Also see note 25.
33
Mesele: Zeyd, mülk evini sıhhatinde bir vakfa rehin koduktan sonra birbiri
akabince bir kaç defa dahi rehn koyup fevt oldukta Zeyd’in evden gayri ahar
terekesi olmayıcak şeran zikr olan vakfın mütevellileri evi bey edüb taksim-i gur-
ema etmeğe kadir olurlar mı? El-Cevab: Evvelki vakfın mütevellisine sıhhat üzerine
rehn edüb teslim etdikten sonra yine ariyet tariki ile aldı ise ol mütevelli tamam alur;
kalanı artar ise alurlar; eğer teslim-i sahih etmedi ise kısmet-i gurema olur;
Akgündüz, Ebüssu’^ ud Efendi Fetvâları, 673. According to one of our referees,
the fetva might have been included in the cinayat section because Zeyd appears
to have committed fraud, which could render him liable for punishment. We thank
our referee for this interpretation.
MAPPING A FETVA COMPILATION 75
Modern researchers would have additional concerns as well. Even if
one accepts the compiler’s subjective choices in the organization of the
material as it is presented, the thematic groupings we find in the corpus
are not wholly appropriate for most types of modern socio-legal re-
search. Indeed, the structure of the text as represented in Akgündüz’s
table of contents provides little direct information on the prevalence of

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


questions regarding certain aspects of life and society in Ebussuud’s
era––for example, non-Muslims, sexuality and gender roles and rela-
tions, and food––that have received scholarly attention in modern
Ottoman studies. For other issues, such as crime and property disputes,
researchers might require fewer categories compared to the ones used in
premodern times, which generate thematic pigeon-holes that are unhelp-
ful for modern concerns. For example, for modern scholars the utility of
treating ‘fetvas on contracts between orchard owners and tenants involv-
ing the shares of the orchard’s yield’ as a separate category is not
obvious.
Second, and more important for our concerns, the text’s organization
of its contents does not (and cannot) take into consideration the fact that
fetvas often contain multiple thematic attributes. Thus, a fetva’s inclu-
sion in one section should not lead the reader to assume that it can be
excluded from others. For this very reason, the compiler’s organizational
arrangement can be deceiving in representing the precise volume of spe-
cific issues in the corpus. For example, consider the following fetvas:
Question: If Zeyd has not cultivated a field (mezraa) under his possession for a
year and, consequently, grass has grown in this area and if the people of [his]
village desire to have their cattle graze on it, can he prevent them from doing
so and claim the grass for himself?

Answer: He can. He can prevent the people [of the village] with the help of a
judge’s approval.34

Question: Can Amr prevent Zeyd from habitually trespassing (yol edib) with
his cattle on a field for which Amr pays the taxes?

Answer: He can.35
Although these fetvas too are included in the section on cinayat, one
can immediately see that they also concern other distinct issues,

34
Mesele: Zeyd, mutasarrıf olduğu mezraayı bir yıl ekmeyüb ot bitse karye halkı
ota davarların salmak istedikte Zeyd men edüb kendi zabt eylemeğe kadir olur mu?
El Cevab: Olur, rey-i hakimle halkı men eder; ibid, 674.
35
Mesele: Zeyd, Amr’ın rüsumun eda etdüği yerin üzerine davarların uğradub
yol etmek dilese Amr mani olmağa kadir olur mu? El-Cevab: Olur; ibid.
76 B O Ğ A Ç E R G E N E A N D A T A B E Y K A Y G U N

including, the status of uncultivated agrarian lands, questions regarding


the ownership of vegetation that grows in such fields, the rights of culti-
vators who do not cultivate, communal rights and relations, and dispute
resolution through litigation. In other cases, we can find fetvas that
simultaneously concern, for example, issues related to charitable foun-
dations, inheritance, and family relations, or ones with references to

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


homicide, blood money, and guardianship (of orphaned minors).
Many more such examples exist. Thus, any representation of the corpus
that ignores the substantive plurality of individual fetvas cannot reliably
gauge the actual dimensions of the semantic masses that constitute the
corpus.
Before moving to the next section, we should mention a potential issue
about the corpus that might impact our interpretations. The scholarly
literature on Ottoman fetvas offers no indication about how comprehen-
sive various fetva compilations attributed to individual jurists might be or
the principles that compilers might have followed while preparing their
collections. More specifically, we cannot be sure if these works excluded
specific fetvas based on the compilers’ subjective considerations including,
for example, concerns about redundancy. In our case, it is precisely to
minimize the potential impact of this possibility that we chose to explore
the largest compilation of Ebussuud’s fetvas available to us. Moreover, we
possess no evidence that Veli bin Yusuf weeded out specific types of fetvas
on a large and, therefore, statistically significant scale. And even if he
perfomed an extensive culling of fetvas, one might reasonably assume
that this was done in a way that affected specific thematic groups rela-
tively evenly. A culling performed in this fashion should not impact our
interpretations pertaining to the socio-legal trends of the era.36

2.2 Data production


Before we delve into our analyses, we should explain how we converted
Ebussuud’s corpus into a data set that can be computationally explored.
In the preprocessing stage, we converted the corpus to a machine-
readable and searchable format through an optical character recognition

36
A large-scale culling performed in a fashion that was disproportionate to the
general representation of specific themes would make it difficult to use the corpus
to represent socio-legal currents in the sixteenth century. Yet it should not impact
our conclusions regarding the thematic structure of the corpus or, even, Ebussuud’s
overall legacy in Ottoman legal history, the two other focal points of this study. In
the latter case, this is because the impact of the jurist’s work on subsequent gen-
erations was shaped not by Ebussuud, the person, but indirectly by the corpora
attributed to him, such as the one that we study in this article.
MAPPING A FETVA COMPILATION 77
(OCR) system.37 Afterward, we cleaned the electronic text by manually
eliminating residual imperfections generated in the scanning process and
parts that are superfluous to our analysis. We also removed any Arabic
text accompanying individual fetvas since we used an off-the-shelf OCR
library specifically designed for contemporary Turkish. Next, we simpli-
fied the typographic characteristics of the electronic text by either elim-

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


inating or reducing diacritical marks, such as regular and reverse
apostrophes and circumflex accents often used in the transliteration of
Ottoman texts from Arabic script to the modern Turkish alphabet. The
latter procedure was necessary to reduce the character variations with
which our computations had to negotiate.
These procedures generated a raw textual base of approximately
6,000 fetvas containing about 250,000 words or word-like constructs,
of which more than 25,000 were unique. Next, we eliminated the ‘stop-
words’, which are words with insignificant substantive import. This list
includes words such as ve (and), ya da (or), bir (one), and etmek (to do).
Also, Ottoman and modern Turkish contain many noun and verb
phrases (such as meclis-i şer and arz-ı memalik). We treated them as
single words by removing the spaces between the words that constitute
the phrase. Finally, we stemmed each word and phrase in the corpus,
which reduced ‘inflectional forms and sometimes derivationally related
forms of a word to a common base form’.38 We used the Snowbol
Turkish stemmer available on most natural language processing (NLP)
platforms.39 The procedure is necessary because ‘for grammatical rea-
sons, documents are going to use different forms of a word, such as
organize, organizes, and organizing’.40 However, because the stemmer
we used is specifically designed for modern Turkish and the corpus con-
tained archaic and regional forms of many words, such as itdüm as
opposed to ettim, stemming left a large portion of the corpus untouched.
So, we had to resort to using an ad hoc normalization scheme specifically
designed for this corpus before we stemmed the whole compilation.

37
See Ray Smith, ‘An overview of the Tesseract OCR engine’, Proceedings of the
Ninth International Conference on Document Analysis and Recognition, 2 (2007):
629–33.
38
See Stemming and Lemmatization (n.d). Online: https://nlp.stanford.edu/IR-
book/html/htmledition/stemming-and-lemmatization-1.html. (Last accessed 11
November 2019.)
39
Gülşen Eryiğit and Eşref Adalı, ‘An affix stripping morphological analyzer for
Turkish’, Proceedings of the IAESTED (2004): 299–304. Online: https://admin.
turkofoni.org/files/an_affix_stripping_morphological_analyzer_for_turkish_g.y_
__t-e.adali-itu-2004.pdf. (Last accessed 2 May 2020.)
40
Stemming and Lemmatization.
78 B O Ğ A Ç E R G E N E A N D A T A B E Y K A Y G U N

Finally, after we stemmed and normalized the text, we also removed


words that appeared fewer than 10 times in the corpus, as these words
had little statistical significance. These procedures reduced the number
of ‘operational’ unique words in the corpus to approximately 1,500.
Consider the following fetva in its original Ottoman to see the result
of the process described above:

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


Mesele: Zeyd, yanında olan bıçağın sapında fildişinden ya balık dişinden bir
dirhem mikdarı belki ziyade nesne olsa salata kerahet olur mu? El-Cevab:
Olmaz. Cevab-ı ahar: Götürmek caizdir, müfsid-i salat değildür.41

After stemming and normalization the fetva transformed into:


bıçak sap fildiş balık diş dirhe salat kerahet müfsi42 salat

We should note that the use of the OCR systems and modern NLP
tools such as stemmers and lemmatizers in Ottoman corpora requires
extreme care, for several reasons:
1. Most modern OCR tools are designed for contemporary Turkish in
the modern Turkish alphabet, or for modern Arabic texts in print.
Although there have been attempts to develop OCR systems for
printed Ottoman texts, as of this writing there were no tools for
handwritten Ottoman texts. Thus, we relied on the modern Latin
transliteration of the Fetava rather than its original manuscript form.

41
Akgündüz, Ebüssu’^ud Efendi Fetvâları, 48. Question: If Zeyd had on him
during prayer (salat) a knife (bıçak) with a handle (sap) inlaid with small pieces
(bir dirhem mikdarı belki ziyade) of ivory (fildişi) or ‘fish tooth’ (balık dişi), would
his prayer be compromised (kerahet)? Answer: It would not. Another answer: It is
permissible to carry it [viz. the knife during prayer]. It would not spoil (müfsid) the
prayer.
42
Stemmers strip each word of its functional suffixes and prefixes. For example,
in English the word ‘miseducation’ can be stemmed as ‘educ’. However, for agglu-
tinative languages such as Turkish, and languages with hybrid vocabulary such as
Ottoman Turkish, stemming carries inherent difficulties and pitfalls as discussed in
the main text. One can see in the above results that the stemmer did not correctly
stem the word müfsid-i (as ‘müfsid’) since the stemmer we use is designed for
modern Turkish. Nevertheless, the stemmed results were consistent (all occur-
rences of müfsid-i and müfside were stemmed as ‘müfsi’ while each occurrence of
müfsid was stemmed as ‘müfsid’). And, since the resulting vocabulary after stem-
ming is used for calculating overlaps across different fetvas, consistency in stem-
ming was what we required in our analysis. When we encountered multiple
stemmings of related or identical words, we merged them to eliminate variations.
Finally, we dropped words with small frequencies, which helped us to eliminate
occasionally misspelled, mistranscribed, or mis-stemmed words.
MAPPING A FETVA COMPILATION 79
2. Ottoman jurisprudential corpora are multilingual: they often contain
Arabic words and phrases, which creates problems for Turkish
stemmers and lemmatizers, even when the text is transliterated in
Latin letters.
3. Incorrect and inconsistent transliterations of Ottoman words are fre-
quent within and across corpora, which affects the stemming and

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


lemmatizing processes.
4. Grammatical structures shift over time and tools designed for modern
Turkish grammar and vocabulary often require manual intervention
for proper functioning.

COMPUTATIONAL ANALYSIS OF FETAVA-YI


EBUSSUUD

3.1 The ‘supervised’ modeling of the corpus


The first analysis that we introduce assumes that researchers experienced
in Ottoman-Islamic law, knowledgeable about the genre of fetva collec-
tions in the Ottoman context, or familiar with a particular Ottoman
jurist’s work can make a reasonable guess about the types of issues found
in the Fetava-yı Ebussuud. Accordingly, the first step is to propose, in a
largely43 a priori fashion, the ‘thematic categories’ that constitute the
corpus. We tentatively call the procedures described in this section a
‘supervised’ analysis, because they are based on predetermined thematic
proxies, in turn based on specific words and phrases that we deemed
appropriate to ‘map’ the text. As we explain later, we opted to call the
types of analyses that do not rely on such predetermined conceptual
tools ‘unsupervised’ ones.
Here are the 29 thematic categories that we use to parse the fetva
compilation:
Some of these categories presented in Table 1 are self-explanatory. For
those that may not be, we have inserted brief explanations next to head-
ings to identify the intended coverage. The themes are not meant to be
exclusive. In fact, significant overlaps exist among fetvas that belong to
specific categories.

43
‘Largely’, and not ‘exclusively’, because the choices and decisions made in
delineating the thematic categories involved frequent and intensive interactions
with the corpus. At the same time, there is no objective way to be certain that
our list of categories is comprehensive of all thematic concentrations that constitute
the text, an issue we explicitly address later in the paper.
80 B O Ğ A Ç E R G E N E A N D A T A B E Y K A Y G U N

Table 1: List and selective definition of thematic categories

code thematic category selective descriptive notation for


fetvas cntaining

0 monetary references to monetary exchanges


1 property – real estate references to land and buildings that

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


may be privately owned
2 property – slavery
3 property – other references to property other than real
estate and slaves
4 property – terminology references to unidentified property
5 family
6 women
7 inheritance
8 waqf
9 gender issues and sexuality references to gender identity, sexual
acts, and their consequences
10 disputing and litigation references to litigation and legal pro-
cedures associated with dispute
resolution, such as amicable settle-
ments (sulh)
11 crime
12 punishment
13 taxation, prebendal rights, and references to taxation, other forms of
revenue extraction wealth extraction, and prebendal
rights and duties
14 land, not property references to state-owned land
15 contractual – transactions, references to buying and selling,
employment, and gifts property transfers including gifts,
commercial partnerships, and em-
ployment contracts
16 contractual – debts and loans
17 ritual and worship references to acts, aspects, or compo-
nents of religious rituals or worship
18 religious and doctrinal references to theological issues and
concerns, including, for example,
sins and virtuous behavior, mazheb-
based variations, permissible and
impermissible (helal-haram)
actions, bases of infidelity and
apostasy, heaven and hell.
(continues)
MAPPING A FETVA COMPILATION 81
Table 1: continued

code thematic category selective descriptive notation for


fetvas cntaining

19 religious, admin. functionaries and specific references to specific office

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


their actions holders and their functions
20 communal (including rights and references to communal life, relation-
obligations) ships, and responsibilities, includ-
ing, for example, collective rights to
way and water, neighbourhood
interactions, and crime committed
by unidentified parties in public
places.
21 non-Muslims
22 representations, suretyship, and
guardianship
23 agriculture and cultivation
24 food and beverages
25 animals
26 intoxicants
27 illness, death, and injury
28 military groups and matters references to the affairs and responsi-
bilities of military groups including
(but not exclusively) warfare,
policing, and interactions with the
populace.

In order to determine whether a particular fetva belonged to one or


more of these categories, we looked for specific words or phrases that we
associated with particular themes. For example, we took fetvas contain-
ing words or phrases such as dava (lawsuit), inkar (denial), kadı (shari‘a
judge), sulh (amicable settlement), meclis-i şer (court), and şahid (wit-
ness) as related to the thematic category ‘Disputing and litigation’. The
lists of words and phrases associated with specific categories are pro-
vided in the appendix. They are based on the authors’ familiarity with
Islamic legal terminology and with the ways various words and phrases
are used in the corpus based on hundreds of spot checks.
In what follows, we first provide the concentrations of each thematic
category based on the numbers of fetvas that contained at least one, at
82 B O Ğ A Ç E R G E N E A N D A T A B E Y K A Y G U N

least three, and at least five words or phrases associated with the specific
theme (Table 2). As is evident in the last row of the table, the categor-
ization based on one or more words or phrases per fetva (henceforth
‘1þW/P’) provides almost 100 percent coverage of the corpus. In other
words, almost all fetvas in the corpus (5,907 of them, to be precise)
contain at least one word or phrase identified in our lists for 29 thematic

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


categories. As we increase the threshold number of words or phrases
needed to link fetvas to specific themes, the overall coverage decreases.
The fetvas that contain three or more words or phrases associated
with specific themes (‘3þW/P’) provide an overall coverage of about
58 percent. Those containing five or more words or phrases (‘5þW/P’)
constitute a coverage of 23 percent.
According to our calculations, fetvas involving themes associated with
‘disputing and litigation’, ‘transactions, employment, gifts’, ‘family’, and
‘religious and doctrinal’ concerns, constitute the largest concentrations.
The fact that this ranking is relatively consistent across three types of
categorization is significant. Regardless of how we identify the thematic
content of the fetvas, we obtain a largely stable characterization of the
most sizeable concentrations. However, there appears to be some vari-
ation in the lower spots (see Table 3). Compared to their representations
in 1þW/P, fetvas that concern ‘waqf’, ‘slavery’, and ‘debts and loans’ are
better represented in both 3þW/P and 5þW/P. time, fetvas concerning
‘food and beverages’, ‘gender and sexuality’, ‘animals’, ‘intoxicants’, and
‘military groups and matters’ constitute the smallest groups in all three
categories. Variations in rankings based on 3þW/P and 5þW/P are con-
siderably less marked.
Aside from providing a comprehensive characterization of the themat-
ic rankings of the corpus, the ‘supervised’ analysis of the Fetava can also
yield observations pertaining to individual groups. Consider Graph 1
(below, p. 86) in this regard, where we find the percentile distributions
of individual thematic groups associated with specific categories in other
groups.
Here, it is not our intention to provide a comprehensive analysis of the
matrix since related issues will come up in what follows. But we should
note a few findings with interesting implications. For example, while we
can understand why only about 20 percent of the fetvas contain specific
references to women and 9 percent to non-Muslims (Table 2), given their
social and legal status in the Ottoman context, the finding that references
to these groups appear more often in the fetvas related to family and to
disputing and litigation (among others) is indicative of some of the pri-
mary socio-legal contexts in which these groups were most visible and,
implicitly, others where they were less so. References to non-Muslims are
particularly frequent in fetvas that concern religious and doctrinal
MAPPING A FETVA COMPILATION 83
Table 2: Fetva numbers for specific themes according to alternative coverage
estimations

1þW/P 3þW/P 5þW/P 1þW/P % 3þW/P % 5þW/P %

monetary 971 203 40 16.1 3.4 0.6


(p) real estate 1129 302 72 18.7 5.00 1.2

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


(p) slavery 617 215 64 10.2 3.6 1.1
(p) other 703 118 21 11.7 2.0 0.4
(p) terminology 479 42 12 8.0 0.7 0.2
family 1720 628 252 28.6 10.4 4.2
women 1316 370 41 21.8 6.1 0.7
inheritance 639 74 19 10.6 1.2 0.3
waqf 608 273 109 10.1 4.5 1.8
gender and 176 13 2 2.9 0.2 0.0
sexuality
disputing and 2114 861 385 35.1 14.3 6.4
litigation
crime 830 135 42 13.8 2.2 0.7
punishment 632 65 15 10.5 1.1 0.3
taxation/preb- 443 91 36 7.4 1.5 0.6
endal/
extraction
land, not 292 54 14 4.8 0.9 0.2
property
(c) transac- 1985 702 258 33.0 11.7 4.3
tions/ em-
ployment/
gifts
(c) debt/loan 477 108 29 7.9 1.8 0.5
ritual and 1078 325 106 17.9 5.4 1.8
worship
religious and 1555 349 115 25.8 5.8 1.9
doctrinal
relig., admin. 1268 267 86 21.1 4.4 1.4
functionaries
communal 879 179 67 14.6 3.0 1.1
non-muslims 523 98 17 8.7 1.6 0.3
representation/ 401 72 18 6.7 1.2 0.3
suretyship/
guardianship
agriculture and 359 85 17 6.0 1.4 0.3
cultivation
(continues)
84 B O Ğ A Ç E R G E N E A N D A T A B E Y K A Y G U N

Table 2: continued

1þW/P 3þW/P 5þW/P 1þW/P % 3þW/P % 5þW/P %

food and 343 51 12 5.7 0.9 0.2


beverage

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


animals 154 29 3 2.6 0.5 0.1
intoxicants 96 10 3 1.6 0.2 0.1
illness/ death / 1364 175 37 22.6 2.9 0.6
injury
military groups 296 27 5 4.9 0.4 0.1
and matters
all categories 5907 3512 1406 98.0 58.3 23.4

(Note: bold indicates relatively large numbers of fetvas and sizeable coverage in the corpus;
‘p’ indicates the ‘property’ category, ‘c’ the ‘contractual’.)

matters and communal concerns, which might be indicative of contem-


porary anxieties related to cross-confessional relations and shared spaces
involving members of different religious communities.
Also, our findings suggest that references to communal anxieties often
appear in fetvas that concern religious and doctrinal issues and in ones
related to ritual and worship, and that disputes and litigations frequently
concerned disagreements involving ‘transactions, employment and gifts’.
Among disputes involving various types of property, those that involved
real estate were twice as likely compared to other types.
More generally, the matrix also reveals the relatively central roles that
‘disputing and litigation’ and ‘transactions/employment/gifts’ play in the
corpus. Their significant representation in fetvas that also feature other
categories indicate that many themes co-appeared in fetvas with these
two, indicating the latter’s broadly-defined contextual influence in the
corpus.44

3.1.1 Composite clusterings of thematic categories


Next, we demonstrate the distances between specific themes and how the
latter might be clustered with others to constitute larger semantic
masses, which brings us closer to the objective of generating a semantic

44
The representation of ‘disputes and litigation’ exceeds 5 percent in 24 themat-
ic categories. The representation of ‘transactions/employment/gifts’ exceeds
5 percent in 23 categories.
MAPPING A FETVA COMPILATION 85
Table 3: Relative rankings of thematic categories

rank thematic ranking based thematic ranking based thematic ranking based
on 1þw/p on 3þw/p on 5þw/p

1 disputing and litigation disputing and litigation disputing and litigation


2 (c) transactions/ em- (c) transactions/ em- (c) transactions/ em-

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


ployment/ gifts ployment/ gifts ployment/ gifts
3 family family family
4 religious/ doctrinal women religious/ doctrinal
5 illness/ death/ injury religious / doctrinal waqf
6 women ritual/ worship ritual/ worship
7 religious/ administra- (p) real estate religious/ administra-
tive functionaries tive functionaries
8 (p) real estate waqf (p) real estate
9 ritual/ worship religious/ administra- communal
tive functionaries
10 monetary (p) slavery (p) slavery
11 communal monetary crime
12 crime communal women
13 (p) other illness/ death/ injury monetary
14 inheritance crime illness/ death/ injury
15 punishment (p) other taxation/ prebendal/
extraction
16 (p) slavery (c) debt/ loan (c) debt/ loan
17 waqf non-Muslims (p) other
18 non-Muslims taxation/ prebendal/ inheritance
extraction
19 (p) terminology agriculture/ cultivation representation/ surety-
ship/ guardianship
20 (c) debt/ loan inheritance non-Muslims
21 taxation/ prebendal/ representation/ surety- agriculture/cultivation
extraction ship/ guardianship
22 representation/ surety- punishment punishment
ship/ guardianship
23 agriculture/cultivation land, not property land, not property
24 food/ beverage food/ beverage food/ beverage
25 military groups/ (p) terminology (p) terminology
matters
26 land, not property animals military groups/
matters
27 gender/ sexuality military groups/ animals
matters
28 animals gender/ sexuality intoxicants
29 intoxicants intoxicants gender/ sexuality

(Note: bold and italic labels indicate, respectively, significant decreases and increases (four or
more spots) in ranking compared to 1þW/P classification; ‘(p)’ indicates property category,
‘(c)’ ‘contractual’.)
86 B O Ğ A Ç E R G E N E A N D A T A B E Y K A Y G U N

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


Graph 1: Cross-tabulation of thematic categories based on 1þW/P

cartography of the Fetava.45 In order to accomplish this, we prepared the


following ‘dendrogram’ by using an agglomerative hierarchical
clustering algorithm.46 The graph (2, next page) is based on the
categorization that takes one or more words/phrases per fetva as proxies
or predictors for thematic attributes.
The horizontal lines associated with particular thematic groups
indicate the relative propensities of the latter to remain isolated. The

45
The distance between any two fetva groups featuring specific thematic cate-
gories is the percentage of fetvas that belong to only one or the other category
among the totality of these two clusters. In other words,
distanceðC1 ; C2 Þ ¼ C1 nCC21þC2 nC1
[C2 ¼1C1 \C2
C1 [C2

where jCj indicates we count the number of fetvas in cluster C. Based on this
formula, let us consider two extreme cases: (i) If two groups of fetvas had no com-
mon fetvas among them, their distance would be 1, the maximum possible distance
value. (ii) If the groups were identical, that is if they contained the same exact fetvas,
then their distance would be 0, the minimum possible distance according to our
methodology.
46
See Trevor Hastie, Robert Tibshirani, and Jerome Friedman, The Elements of
Statistical Learning: Data Mining, Inference, and Prediction (New York: Springer,
2001), 520–8. Online: https://web.stanford.edu/hastie/ElemStatLearn/printings
/ESLII_print10.pdf. (Last accessed 2 May 2020.)
MAPPING A FETVA COMPILATION 87

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


Graph 2: Dendrogram based on 1þW/P
(Note: some thematic category labels have been shortened and some words
abbreviated for space considerations. Numbers on the x-axis are proportional.
They indicate relative distances among thematic groups.)

order in which these lines merge indicate the relative similarities among
different groups in terms of the words and phrases they shared. For
example, the relative brevity of distinct lines for ‘women’ and ‘family’
indicate that, compared to other themes, the numbers of fetvas in which
these themes appear isolated are small. And, the fact that the lines for
‘women’ and ‘family’ merge before they merge with others indicates that
fetvas with these thematic attributes were common. In other words,
these two themes often appeared in the fetvas simultaneously—that is,
fetvas that contained proxy words/phrases for ‘family’ also frequently
contained proxy words/phrases for ‘women’. The graph suggests that,
in addition to ‘family’ and ‘women’, the themes of ‘inheritance’ and
88 B O Ğ A Ç E R G E N E A N D A T A B E Y K A Y G U N

‘illness/ death/ injury’, ‘intoxicants’ and ‘food and beverage’, ‘animals’


and non-slave and non-real estate ‘property’, ‘military matters/ groups’
and ‘taxation/ prebendal/ extraction’, and ‘punishment’ and ‘crime’
(among others) were relatively likely to appear in the same clusters of
fetvas. In many other cases, some themes appeared in combination with
two or more other categories. For example, the theme of ‘gender and

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


sexuality’ often appeared in fetvas that combined two thematic compo-
sitions: ‘intoxicants’ and ‘food/ beverage’, as we also observed earlier.
Based on the information provided in the dendrogram, we can also
identify distinct thematic clusters composed of specific categories based
on their relative distances to one another:
The six thematic clusters (one with two distinct sub-clusters) in
Table 4 represent the semantic concentrations in the corpus based
on how individual themes might be linked. For example, issues related
to contractual arrangements (excluding debts and loans), waqf-related
matters, and real estate, and various references to property, frequently
appeared in close proximity with one other (cluster 2b), and collectively
constitute about 51 percent of the compilation. We also detected a close
relationship between this group with a smaller one (cluster 2a), composed
of ‘representation, guardianship and suretyship’ and ‘debts and loans’.
The cluster composed of the two sub-clusters makes up 54 percent of
Ebussuud’s work. Another large cluster comprises family- and women-
related issues on the one hand, and ‘illness/ death/ injury’ and ‘inherit-
ance’ on the other, constituting about 46 percent of the corpus.
The other, largest concentration (cluster 4) is composed of fetvas that
linked communal concerns, non-Muslims, doctrinal and ritual issues,
litigations, crime, punishment, and the actions and responsibilities
of public authorities. Collectively they provide a coverage of about
74 percent and, thus, these fetvas indicate a significant interest, whether
attributed to the müfti or those who sought his opinion, in topics related to
the socio-religious order and relationships and the legal means to enforce
and maintain relevant notions of propriety in the Ottoman context.
Among the three smaller clusters, the fifth and sixth are notable for
different reasons. The fact that the cluster that features the categories of
animals and property (excluding slaves and real estate) remained rela-
tively separate from those that concerned other forms of property sug-
gests that the Ottomans considered these topics to be distinct, a tendency
that might not be obvious to modern readers of the Fetava. On the other
hand, a modern reader of the Fetava would not be surprised to observe a
distinct group related to concerns about cultivation, land use and pos-
session, and the prebendal privileges of the military establishment, given
Ebussuud’s well-recognized role in shaping the Ottoman polity’s land
system and laws. Finally, we can identify in the table a very small cluster
Table 4: Thematic clusters and their coverages (1þW/P)

cluster 1 cluster 2 cluster 3 cluster 4 cluster 5 cluster 6

cluster 2a cluster 2b

intoxicants represent- transactions/ em- illness/ death/ non-Muslims animals military


food & beverage ation/ guardian- ployment/ gifts injury communal property: other taxation/ preb-
gender & sexuality ship/ suretyship monetary inheritance religious & endal/
debts & real estate women doctrinal extraction
loans waqf family ritual & worship land (not property)
property: religious, admin. agriculture &
terminology functionaries cultivation
13.3 51.3 disputing &
litigation
punishment
MAPPING A FETVA COMPILATION

crime
8.8 54.6 46.4 74.2 12.2 15.1

(Note: figures indicate coverage percentages for specific sub-clusters and clusters. Slavery (10.2) excluded based on the corresponding dendrogram.)
89

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


90 B O Ğ A Ç E R G E N E A N D A T A B E Y K A Y G U N

of fetvas (cluster 1) that linked food, beverages, intoxicants, and gender


and sexuality, implying their relative insignificance in premodern juris-
prudential considerations.
When we repeat the same analysis using categories identified by three
or more words/phrases per fetva, we obtain the dendrogram Graph 3
(below).

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


One major difference between the two graphs is that in the latter case,
many of the lines associated with specific themes appear to remain dis-
tinct (i.e., un-merged with others) longer in relative terms. This finding is
expected and is indicative of the hardening of the boundaries associated
with specific themes as we increase the number of words and phrases to
determine the fetvas’ thematic attributes. Compare, for example, the

Graph 3: Dendrogram based on 3þW/P


(Note: thematic labels have been shortened and some words abbreviated for space
considerations. The numbers on the x-axis are proportional. They indicate relative
distances among thematic groups.)
MAPPING A FETVA COMPILATION 91
lengths of lines for women and family before their merger in both
graphs.
The dendrogram based on 3þW/P also demonstrates some variations
regarding how specific groups of fetvas tend to get linked with others.
For example, crime-related fetvas merge first with those that demon-
strate communal attributes instead of those related to punishment, as

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


was the case in the previous dendrogram. And those associated with
‘transactions/ employment/ gifts’ first merge with the those related to
‘real estate’ rather than those with ‘monetary’ links.
Notwithstanding such differences, however, the graphs provide a
somewhat consistent mapping of the more prominent thematic clusters,
both in terms of their individual constituents and their relative size in the
corpus (see Table 5). In both graphs, one can identify prominent con-
centrations involving monetary, contractual, waqf- and (select) property-
related concerns, on the one hand, and concentrations involving com-
munal, confessional, and religious issues, and crimes, punishments, and
litigations, on the other. A close third in terms of its size is the group of
fetvas generally related to family, women, and inheritance, though one
could spot slight variations in the exact compositions of clusters in the
two analyses. Broadly speaking, these ‘areas’ of consolidation represent
the thematic foci of the corpus. And both analyses tend to represent less
prominent clusters in consistent fashions.

3.1.2 Contextual implications of the supervised analysis


Figure 1 (next page) provides a visualization of the thematic composition
of the corpus that juxtaposes the relative sizes of individual categories and
their relative proximities to one another, from a vantage point between
1þW/P- and 3þW/P-based analyses.47 Thus, it allows us to reflect more
intuitively on the contextual implications of our findings.
In Figure 1, the relative sizes of the bubbles display the relative preva-
lence of individual thematic categories. And the relative distances be-
tween specific bubbles indicate the relative likelihood of the associated
themes appearing within individual fetvas. Accordingly, a few themes
stand out in the corpus. The fact that ‘disputing and litigation’ appear in
a dominant fashion should be expected, given the fact that fetvas were
often composed to provide guidance to resolve conflicts in or outside of
the court. The centrality of this theme, established earlier, underscores
the very same point. In terms of representing the socio-legal concerns
of the sixteenth century, however, we find equally revealing the
prevalence of ‘transactions/ employment/ gifts’, ‘family’, ‘religious and

47
The figure is based on 15 percent thematic association threshold, which
roughly corresponds to 2.4 stemmed words/phrases per fetva in average.
Table 5: Thematic clusters and coverages (3þW/P)
92

cluster 1 cluster 2 cluster 3 cluster 4 cluster 5 cluster 6

cluster 3a cluster 3b

animals religious, admin. communal land (not property) women intoxicants


property: other transactions/ em- functionaries crime taxation/prebendal/ family food & beverage
ployment/ gifts ritual/worship punishment extraction slavery
disputing/ illness/ death/ military inheritance
real estate litigation injury property:
waqf non-Muslims terminology
monetary religious/ doctrinal agriculture/
debts & cultivation
loans 25.4 7.7
2.1 20.4 30.4 4.2 17.0 0.9

(Notes: figures indicate coverage percentages for specific sub-clusters and clusters. ‘Gender issues and sexuality’ (0.2) and ‘representations, suretyship, and
guardianship’ (1.2) are excluded based on the corresponding dendrogram.)
B O Ğ A Ç E R G E N E A N D A T A B E Y K A Y G U N

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


MAPPING A FETVA COMPILATION 93
doctrinal’, and ‘ritual and worship’. Issues related to these themes gener-
ated the most questions and forced the jurist to comment on them most
frequently.48

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


Figure 1: Semantic mapping of Fetava based on supervised analysis
(Note: the figure was prepared using the Python programming language libraries,49
Matplotlib,50 Scikit-learn,51 Gensim,52 and Pandas.53 The code we used for the
visualization is available at https://github.com/kaygun.)

48
Again, our interpretations are based on the assumptions either that the corpus
constitutes a comprehensive compilation of Ebussuud’s fetvas or that whatever
subjective culling of fetvas the compiler might have done was statistically insignifi-
cant, or that if he indeed weeded-out large numbers of fetvas he did this in a fashion
that was roughly proportionate to the relative sizes of specific themes.
49
Guido Van Rossum, Python Tutorial: Technical Report CS-R9526
(Amsterdam: Centrum voor Wiskunde en Informatica, 1995).
50
J.D. Hunter, ‘Matplotlib: a 2-D graphics environment’, Computing in Science
& Engineering, 9/3 (2007): 90–5.
51
Fabio Pedregosa et al., ‘Scikit-Learn: machine learning in Python’, Journal of
Machine Learning Research, 12 (2011): 2825–30.
52 
Radim Rehů rek and Petr Sojka, ‘Software framework for topic modelling with
large corpora’ in Proceedings of LREC 2010 workshop New Challenges for NLP
Frameworks (Valletta, Malta: University of Malta, 2010): 46–50.
53
Wes McKinney, ‘Data structures for statistical computing in Python’ in
Proceedings of the 9th Python in Science Conference (2010): 51–6. Online:
https://conference.scipy.org/proceedings/scipy2010/pdfs/mckinney.pdf. (Last
accessed 10 February 2020.)
94 B O Ğ A Ç E R G E N E A N D A T A B E Y K A Y G U N

At the same time our analyses also allow us to establish the connec-
tions among specific themes, which helps identify the broader socio-legal
trends that cannot be restricted to specific categories. Indeed, what we
observe in the figure, generally consistent with ones based on the dendro-
grams, leads us to identify two broadly-defined foci in Ebussuud’s fetvas.
The first one, concentrated in the lower-left side of the figure, concerns

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


economic, i.e., contractual and property-related, concerns. The second
one, concentrated in the right side, demarcates religious-doctrinal, com-
munal (including non-Muslims), and family matters. This finding is con-
sistent with how historical scholarship described the prevalent
characteristics of the sixteenth century. This was an increasingly mone-
tized and commercially-active era, and one during which the society and
those in positions of authority tried hard to manage diversity, establish
communal standards, define family relations, and, in the process, shape
(and re-shape) religious boundaries.
Our analyses also generated mild surprises for those familiar with the
period. Given how much Ebussuud’s contributions on the topics on
Ottoman land- and tax-systems have garnered interest in modern his-
toriography, it is noteworthy that his opinions on those issues occupied
a limited space in the corpus (concentrated in the upper left side of the
figure). One might as well make a similar claim regarding the jurist’s
waqf-related fetvas. Admittedly, a more complete assessment of the jur-
isprudential import of these opinions requires a dedicated engagement
of the ideas and formulations within them. Also, upon further research
on other fetva corpora, we may still determine that Ebussuud’s fetvas on
such matters constituted a significant bulk compared to those issued by
earlier and later jurists. Nevertheless, our findings call for more caution
in tying Ebussuud’s jurisprudential legacy to his fetvas on these specific
areas.54

3.2 ‘Unsupervised’ modeling of Fetava-yı Ebussuud


The ‘supervised’ analysis of Ebussuud’s fetvas is premised on a thematic
classification proposed by the authors. This choice can be justified based
on the assumption that the authors’ legally and contextually informed
understanding of Ottoman-Islamic legal system is sound and compatible

54
Our results by no means imply that the jurist’s land-, tax-, and waqf-related
opinions were less important compared to others. In fact, and as one of our referees
suggested, ‘[l]itigation is most likely in cases where the merits are unclear, and the
payoff from litigation is high. If the rules are clear, we would expect less litigation,
and accordingly, fewer occasions on which fatwas would be issued’. This is a
fascinating possibility that requires further research on the relationship between
fetva-use and litigation patterns in the Ottoman context.
MAPPING A FETVA COMPILATION 95
with the modus operandi of the Fetava and thus capable of accurately
representing the thematic structuring of the corpus. What if we dropped
this assumption?
In what follows, we propose a methodology in which predefined sets
of words and phrases indicative of specific themes are not employed in
the analysis of the Fetava, or at least not initially. Instead, we allow a

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


topic modeling algorithm, Latent Dirichlet Allocation,55 to scan the cor-
pus and identify the major topical concentrations that constitute the text.
Topic modeling is a type of computational methodology that analyzes
large texts, often composed of millions of words, to identify their latent
topics. Various modeling algorithms, including Latent Semantic
Indexing (LSI),56 Latent Dirichlet Allocation (LDA), and Gibbs
Sampling Algorithm for the Dirichlet Multinomial Mixture
(GSDMM),57 define these topics as composed of specific sets of words
or phrases that frequently appear in close proximity and thus may be
indicative of unique but not always easily identifiable semantic compo-
nents of the corpora that they collectively constitute. We experimented
with all of these various algorithms and decided to adopt a version of the
LDA as implemented in Lancichinetti et al. because this technique gen-
erated easy-to-interpret (contrary to LSA) and reproducible (contrary to
the standard LDA and the GSDMM) results.58
The LDA algorithm we used identified more than 100 ‘topics’ in
Fetava-yı Ebussuud, composed of 100 or fewer closely associated words
and phrases. Before we proceed to consider them, however, we should
make a terminological clarification. In the following analysis, we use the
term ‘topic’ to refer to the word collections generated by the LDA algo-
rithm, each composed of words and phrases identified from the entirety

55
See David M. Blei, Andrew Y. Ng, and Michael I. Jordan, ‘Latent Dirichlet
Allocation’, Journal of Machine Learning Research, 3/4–5 (2003): 993–1022; and
Andrea Lancichinetti et al., ‘High-reproducibility and high-accuracy method for
automated topic classification’, Physical Review, 10/5 (2015). Online: https://
www.researchgate.net/publication/271528914_High-Reproducibility_and_High-
Accuracy_Method_for_Automated_Topic_Classification. (Last accessed 27
December 2019.)
56
S. Deerwester et al., ‘Indexing by Latent Semantic Analysis’, Journal of the
American Society for Information Science, 41 (1990): 391–407.
57
Jianhua Yin and Jianyong Wang, ‘A Dirichlet Multinomial Mixture model-
based approach for short text clustering’ in Proceedings of the 20th ACM SIGKDD
International Conference on Knowledge Discovery and Data Mining (2014):
233–42. Online: https://dl.acm.org/citation.cfm?id=2623715. (Last accessed 27
December 2019.)
58
Lancichinetti, ‘High-reproducibility’.
96 B O Ğ A Ç E R G E N E A N D A T A B E Y K A Y G U N

of approximately 1,500 ‘operational’ unique words. Thus, the ‘topics’


are different from and should not be confused with the ‘themes’ or ‘the-
matic categories’ we used in the ‘supervised’ analysis based on our pre-
defined lists of words and phrases.

3.2.1 Findings of LDA topic modeling

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


The ‘topics’ identified by our LDA algorithm are composed of words and
phrases that still required an interpretive effort on the authors’ part. One
relatively straightforward method to interpret these is to look at their
content through the lens of the thematic categories that we relied on in
our ‘supervised’ analysis. The matrix presented in Graph 4 cross-tabulates
the fetvas associated with selected topics and with thematic categories and,
by doing so, provides an understanding of the overlaps between selected
LDA-generated topics and predefined thematic categories. The fetvas rep-
resentative of each thematic category in the matrix are those that contained
15 percent or more of the stemmed words/phrases linked to a specific
theme. On the other hand, and because the word/phrase lists are much
longer for LDA-generated topics, we identified the fetvas as representative
of specific topics when at least 22 percent or more of their stemmed con-
stituents overlapped with the specific topics’ word/phrase lists.59
The 100 or so topics that the LDA algorithm identified in the corpus
cover about 70 percent of the fetvas in the corpus, based on our 22-per-
cent-word/phrase-per-fetva standard. We included in the matrix 19 topics,
whose coverage rates are 1 percent or above (see examples of the word/
phrase lists for a few topics in Appendix 3). The following are some of the
more important observations based on the matrix:
1. In terms of specific thematic categories, the unsupervised analysis
generates results that are generally consistent with the ones based on
supervised modeling. Indeed, the largest thematic categories (‘trans-
actions/ employment/ gifts’, ‘disputing and litigation’, ‘family’, ‘rit-
ual and worship’, ‘religious and doctrinal’) in topics with the largest
fetva coverage (topics 0, 3, 7, 4, 1, 9) are among the major themes
that we identified previously. And ones that appear least conspicu-
ously in LDA-constituted topics (‘gender and sexuality’, ‘intoxi-
cants’, ‘animals’, ‘military matters’, and ‘land’) match the least
prominent ones based on the supervised analysis (not identified in
the matrix because of their negligible representation). The analysis
also captures the relative centrality of ‘disputing and litigation’ and

59
On average, 15 and 22 percent thresholds respectively correspond to 2.4 and
3.4 stemmed words/phrases per fetva. They help with reading sizes and distances
represented in figure proportionate to the overall scale.
MAPPING A FETVA COMPILATION 97

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


Graph 4: Cross-tabulation of fetvas associated with selected LDA topics and the-
matic categories. (Figures in parentheses indicate the members of associated fetras.)

‘transactions/ employment/ gifts’ in the sense that these themes are


well-represented and, thus, markedly impactful in multiple topics, a
finding that is consistent with what we observed in the supervised
analysis. Other centrally positioned categories include ‘family’, ‘re-
ligious and doctrinal’, and ‘ritual and worship’.60
2. While a few topics single out specific thematic categories (topic 3:
‘disputing and litigation’; topic 37: ‘agriculture and cultivation’), the
unsupervised analysis forces us to consider semantic concentrations
largely as composite thematic spaces. Indeed, topic 0, which pro-
vides the largest coverage in the corpus, clusters three categories:
‘transactions/ employment/ gifts’, ‘waqf’, and ‘disputing and litiga-
tion’. Among these, ‘transactions/ employment/ gifts’ constitutes
about 20 percent of those fetvas associated with the topic, and the
latter two each constitute about 6 to 7 percent. Other sizeable topics

60
Interestingly, there are variations among centrally positioned themes in terms
of how prominently they are represented. For example, while the coverage of ‘fam-
ily’ exceeds 37 percent in three of the six topics in which it is well represented, the
coverage of ‘transactions/ employment/ gifts’ exceeds 14 percent only for one. This
observation suggests that the relative impact of central themes in the topics in which
they appear could be quite significant or more diffuse.
98 B O Ğ A Ç E R G E N E A N D A T A B E Y K A Y G U N

tend to cluster two to five prominent themes each, including, for


example, ‘women’ and ‘family’ (topic 7); ‘real estate’, ‘waqf’, ‘tax-
ation/ prebendal/ extraction’, ‘transactions/ employment/ gifts’, ‘re-
ligious, administrative functionaries/ actions’, and ‘agriculture and
cultivation’ (topic 4); and ‘slavery’ and ‘transactions/ employment/
gifts’ (topic 20).

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


The matrix also allows us to recognize a few frequent thematic asso-
ciations across multiple topics (for example, ‘family’ and ‘women’;
‘crime’, ‘punishment’, and ‘disputing and litigation’; and ‘ritual and
worship’ and ‘religious/ doctrinal’), indicating the close links among
the constituent elements in numerous semantic contexts.61 In some
cases (for example, in topics 7 and 13, where both ‘family’ and
‘women’ constitute the only sizeable categories), it is difficult to dif-
ferentiate these topics at first glance. In other cases, the topics that
contain such clusters are different to the extent that they appear
associated with other themes or clusters. For example, topic 24 hints
at the links between the ‘family-’ and ‘women’-themed fetvas and
those related to ‘slavery’ and ‘transactions/ employment/ gifts’.
Likewise, while topics 5 and 8 highlight fetvas related to ‘crime’,
‘punishment’, and ‘disputing and litigation’, they diverge in terms
of how they link them to distinct sets of themes.
3. Thus, the matrix characterizes the most prominent semantic spaces
primarily as non-exclusive, overlapping textual spaces. In this
sense, the picture that it draws is dissimilar to the one offered by
the supervised analysis, which expresses difference across textual
space by compartmentalizing specific themes. Not only can ‘waqf’,
for example, be distinguished from, say, ‘transactions/ employ-
ment/ gifts’ (in Tables 2 and 3), but it also cannot be included in
any cluster or sub-cluster other than 2 in Table 5. The LDA-
analysis, however, implies that distinct textual spaces can be gen-
erated by slightly altering the exact composition of the thematic
ingredients common to many topics, which justifies representing
the latter as intersecting (though separate) entities and the corpus
as a space where meaning is shifting and fragile. Accordingly,

61
The fact that we identify such associations in many topics generates the im-
pression that the categories that constituted them might be regarded as fractured
components of more coherent thematic entities and, thus, they could/should have
been amalgamated. This is a possibility for ‘family’ and ‘women’ based on what we
observe in the matrix, which forces us to consider merging these into a single cat-
egory in future research. In other topics (topics 1 and 22, for example), however,
one or more constituents of the frequent associations also appear independently of
their usual partners.
MAPPING A FETVA COMPILATION 99
‘waqf’ is prominently visible in no context independent of ‘trans-
actions/ employment/ gifts’ (Topics 0 and 4), and the meanings that
the two themes co-generate vary based on their respective repre-
sentation and the strength of their links to other thematic elements
in distinct topics (Topics 0 versus 4).
4. If we were to highlight some of our primary findings with historical

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


significance from this perspective, we may suggest that the semantic
spaces that dominate the corpus include contractual disagreements,
often (but not exclusively) over waqf matters and/or family mem-
bers (topic 0), disputing and litigation, broadly defined (topic 3),
issues pertaining to family and (presumably, related) women (topic
7), concerns that linked wealth extraction, government authorities,
agrarian activities, and contractual and real estate related matters
(topic 4), and religious doctrine and, to a lesser extent, ritual,
including litigations involving these (topic 1). Again, the reader
should note here that a few themes are prominently represented in
multiple topical contexts.
5. The last point that we would like to make becomes clear in Figure 2,
which juxtaposes the relative coverages of the nineteen largest topics in
the corpus and their relative proximities to others, based on the pro-
portions of the fetvas in which they simultaneously appear in a manner,
as a geometric approximate representation, similar to Figure 1.
Figure 2 (next page) is useful not only because it nicely visualizes the
relative prevalence of topics 0, 3, and 7 in the corpus but it also reveals that,
based on how we might count them, the corpus comprises six or so ‘zones’
of fetvas, each with different degrees of topical diversity. A few of these
zones, individually associated with topics 0, 7, and 9, as found in the edges
of our textual space, appear to be thematically more homogeneous and,
thus, are easier to categorize. On the other hand, the algorithm also iden-
tifies other ‘zones’ in which topics as different as 1, 3, 4, 18, and 30; 8, 17,
20, 24, and 37; and 10, 16, and 22 (and perhaps 5, 27, 13) are clustered.
Thus, these three zones of fetvas display considerably more diversity.
Finally, the figure reveals that even relatively distant fetva zones could
have resembling thematic characteristics as is evident in topics 7 and 13,
which both feature ‘family’ and ‘women’ as the most prominent themes.62

62
Once again, a geographical analogy might help clarify these findings. Our
finding regarding restricted zones of fetvas with significant thematic diversity
might be akin to climactic and topographical variations that one could observe
in a relatively small region. And distant zones with shared thematic features resem-
bles the presence of comparable climactic and topographical circumstances in very
distant locations.
100 B O Ğ A Ç E R G E N E A N D A T A B E Y K A Y G U N

Legend:

Topic prevalent themes in descending order (dominant ones in bold)

0 transactions - dispute - waqf - family


3 dispute

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


7 family - women
4 taxation - functionaries - transactions - agriculture - real estate
1 doctrinal - ritual - dispute
9 ritual - doctrinal
20 slavery - transactions
16 non-Muslims - doctrinal - communal
8 crime - communal - illness - dispute - punishment
18 transactions - other property
22 punishment - family - dispute – ritual
27 doctrinal - ritual
17 debt - transactions - dispute - crime
10 real estate - transactions - ritual - communal
13 family - women
30 illness - ritual - doctrinal - family
5 crime - ritual - doctrinal - functionaries - punishment - dispute
24 family - women - slavery - transactions
37 agriculture

3.2.2 Categorical ‘others’ and topics omitted from the matrix


Given that one of the primary objectives of the unsupervised analysis is
to identify the conceptual blindspots of the supervised modeling, one
might question our choice of thematic parsing of the LDA-based topics.
Since thematic categories themselves are a priori constructs, does this not
allow our subjective dispositions towards the corpus to indirectly influ-
ence our findings?
Phenomenologically speaking, it might not be possible to interpret the
LDA-based topics in a fashion that is completely independent of our
preconceived dispositions towards the corpus, whether these seep into
the analysis through thematic categories or not. Yet the unsupervised
analysis also offers a practical means to assess the limitations of our
engagement with the text. One way to accomplish this task is to examine
those domains that our analysis fails to map out. In fact, and as we can
see in the matrix, our predefined thematic categories and the clusters
based on them cannot fully parse the semantic constituents of the topics,
which is why we identify relatively sizeable ‘others’ for every topic.
MAPPING A FETVA COMPILATION 101

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


Figure 2: Semantic mapping of Fetava based on unsupervised analysis
(Note: See the note for Figure 1.)

What should we make of this category and what is its relationship to the
topics’ other constituents?
Since the LDA algorithm recognized and coalesced those words/
phrases that frequently appeared close to each other, we hypothesized
that the fetvas captured by the ‘other’ category frequently contained
terms closely related to the dominant themes in a particular topic but
not included in our predetermined proxy lists. Indeed, when we tested
this hypothesis by picking a handful of topics, removing from each list of
words/phrases those that we identified as thematic proxies, and closely
reviewing the remainder, it turned out to be valid. Our examination
revealed that the words/phrases that constituted the categorical ‘others’
were excluded from thematic proxy lists often because they appear in-
frequently (fewer than 10 times) in the corpus. In other cases, they
included terms that we deliberately omitted from the proxy lists because,
while they too may be related to a particular thematic category, we could
not assume their associations to be adequately exclusive.63
63
For a more extensive discussion of the methodology and our findings, see
Appendix 4.
102 B O Ğ A Ç E R G E N E A N D A T A B E Y K A Y G U N

Finally, we also made an effort to examine the topics identified by the


LDA-algorithm but excluded from the matrix in Graph 4 because of
their insignificant coverages.64 Our impression based on the examples
we examined65 is that quite a few of the omitted topics are dominated by
terms that are related to the notions associated with our themes but not
included in our proxy lists because they appear infrequently in the cor-

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


pus. A few of these are mis-spelled or irregularly spelled versions of the
terms that could be found in the proxy lists. In other topics, however, we
could identify small thematic concentrations or potential appendages to
existing themes that we had not considered at the time that we designed
our thematic categories. Despite the marginal coverages of these topics,
the fact that the LDA algorithm could detect such coherent semantic
masses highlights its potential as a research tool, in particular for
researchers interested in identifying particular subject matters in a cor-
pus rather than its major components as we did in this article.

CONCLUSION

This article has described our multiple attempts to represent the seman-
tic constituents of Fetava-yı Ebussuud, an influential sixteenth-century
jurisprudential source that has yet to be explored in a comprehensive
fashion. We proposed supervised and unsupervised methodologies to
accomplish this task. The first was based on contextually informed
yet, ultimately, a priori ‘thematic categories’. The second used a model-
ing algorithm that identified ‘topics’ in the corpus based on the relative
proximities of words and phrases that constitute the text.
Both methodologies identified a handful of themes that prominently
appeared in the compilation, including ‘disputing and litigation’, ‘trans-
actions/ employment/ gifts’, ‘religious and doctrinal’ concerns, ‘ritual
and worship’, and ‘family’. They also determined that among these,
the first two were relatively central in that they often appeared simul-
taneously with other thematic concerns.
Our methodologies also provided clues to broader semantic concen-
trations in the corpus. The supervised analysis tended to prominently
cluster together fetvas that concerned: 1. ‘transactions/ employment/
gifts’, ‘debts and loans’, ‘real estate’, ‘waqf’, and ‘monetary’ issues;

64
The coverages of the omitted topics typically range between 0.0 to 0.5 percent
of the fetvas at our 22 percent word/phrase per fetva threshold. In fact, over half of
these topics generate coverages of 0.1 percent or less.
65
For a discussion of the methodology and our findings, see Appendix 5.
MAPPING A FETVA COMPILATION 103
2. ‘disputing and litigation’, ‘communal’ concerns, ‘religious and doctri-
nal’ matters, ‘ritual and worship’, ‘non-Muslims’, ‘crime’, and ‘punish-
ment’; and 3. ‘women’, ‘family’, and ‘inheritance’. Different iterations of
the supervised analysis also identified smaller clusters of fetvas related to
food, beverages, and intoxicants; animals and property (not real estate or
slaves); and land use/possession, taxation, and agriculture.

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


The unsupervised analysis, on the other hand, forced us to consider
the Fetava’s contextual associations in a significantly different manner.
Instead of linking a particular thematic concern to a single semantic
domain, the LDA algorithm identified multiple contexts in which specific
themes were prominently represented. While a few themes tended to
appear together in multiple topical contexts, they also demonstrated
conspicuous associations with other themes that cannot all be catego-
rized or clustered in a straightforward manner. The broader implication
of this finding is that the Fetava should be conceived less as a textual
space of distinct, compartmentalized components as we might imagine
based on the supervised analysis. Instead, we should think of it more as
an entity composed of non-exclusive, overlapping textual masses.
The supervised and unsupervised methodologies have different advan-
tages as analytical tools. While the former is easier to conceptualize and
implement and tends to generate results open to more straightforward
interpretation, the latter permits researchers to identify and remedy their
conceptual blind spots and also to detect a great range of semantic
concentrations in the corpus. For our purposes in this article, a comple-
mentary use of these methodologies seemed promising as it generated a
largely coherent yet multifaceted characterization of the Fetava, one that
is worthy of the corpus’ importance in Ottoman–Islamic legal history.
Regarding the historical significance of our findings, we should high-
light the prominence in the Fetava of material related to litigation and
dispute resolution processes, contractual and commercial concerns,
property ownership and transactions, and religious and doctrinal mat-
ters, which might be indicative of the prevalent socio-legal issues in
Ottoman lands during the sixteenth century. Based on a few reasonable
assumptions, the corpus may thus be seen as a gauge of the public opin-
ion, reflecting the most widespread anxieties among a cross-section of
Ottomans during this period. The fact that we also found material
related to waqfs, land use and possession, and taxation is expected given
the purported influence of Ebussuud’s jurisprudential formulations in the
relevant realms of the law. Yet the volume of fetvas on these topics
appeared modest compared to others. True, we can develop a more de-
finitive opinion on how well this material is represented in Ebussuud’s
work only by making comparisons with other jurists’ compilations.
Nevertheless, our findings do caution us about how the jurist’s scholarly
104 B O Ğ A Ç E R G E N E A N D A T A B E Y K A Y G U N

proclivities have been represented in the academic literature. To give one


example, the scholarship provides little indication of the possibility that
the numerous fetvas that combined property-related matters, contractual
arrangements, and family affairs (as detected in topic 0) constituted a
focal concern in Ebussuud’s work.
Aside from the historical import of our findings, however, let us em-

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


phasize the fact that the primary accomplishment of the present article is
methodological in nature. It proposed computational techniques and
approaches that successfully capture the substantive constituents of an
important Ottoman fetva collection. In future publications, we plan to
apply these to other compilations to generate cross-corpora comparisons
and thus identify the major substantive breaks and continuities in the
genre of Ottoman fetva compilations over time.

Appendix 1: Thematic categories

The words and phrases associated with specific thematic categories are as
follows. The reader should note that the lists are based on the ‘operational’
text, constituted by eliminating the infrequent words and after stemming
and lemmating processes. This is why many words or phrases that one
might have expected to appear in the lists are missing. Also, not every single
variation (including spelling) or derivative of a particular word or phrase is
indicated in the lists. For example, in the category of ‘Slavery’, while itak
(manumission) is listed below, mutak (manumitted slave), which we uti-
lized in our analysis, is not. Thus, the reader should take the lists below as
illustrative rather than comprehensive.
0. Monetary: akça, dirhem, filori, nakid, and their cognates.
1. Property – real estate: akar, arz-ı haraci, arz-ı havz-ı memleket, arz-ı
memlüke, arz-ı öşriyye, avlu, bağ, bahçe, beyt, bina, çardak, çiftlik, değir-
men, duvar, dükkan, emlak, ev, hamam, han, kapı, menzil, sükna, tavan,
and their cognates.
2. Property – slavery: abd, azad, cariye, esir, hür, ibak, itak, kul, mevla,
müdebber, mükateb, and cognates.
3. Property – other: altın, bez, davar, deve, emval, giysi, eşcar, gümüş,
harir, inek, kaftan, katır, koyun, küp, kuşak, libas, mal, meta, öküz, sığır,
yüzük, and cognates.
4. Property – terminology: malik, memlük, mülk, rakabe, temlik, and
their cognates.
5. Family: ağırlık, akraba, ana, asabe, baba, bain, benat, boşa, cihaz,
düğün, ebeveyn, ebna, evlat, hemşire, hul, hulle, idde, karındaş, kerime,
lian, mehr, nafaka, namzed, neseb, nesl, nikah, oğlu, talak, tezvic, valide,
veled, yetim, zevc, and their cognates.
MAPPING A FETVA COMPILATION 105
6. Women: ana, bint, avrat, baliğe, bikr, hatun, hemşire, hind, inas,
kadın, kerime, kız, mutallaka, nisa, valide, zevce, zeyneb, and cognates.
7. Inheritance: eytam, kısm, kısmet, miras, muhallefat, tereke, varis,
vasiyet, vefat, veraset, verese, yetim, and cognates.
8. Waqf: cabi, meremmet, mevkuf, mütevelli, rakabe, tevliyet, vakfiye,
vakıf, vakıfname, and cognates.

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


9. Gender issues and sexuality: bikr, namahrem, cima, emred, hamile,
hayz, livata, zina, and cognates.
10. Disputing and litigation: adil, ahkam, beyyine, dava, def, delil, fetva,
hakim, hakimülvakt, hasm, hükm, husümet, ikrar, inkar, isbat, işhad, istima,
itiraf, kadı, kaza, meclis-i şer, mesmua, müddei, müddea, müfti, münkir,
murafa, mütevatir, naib, niza, şahid, şehadet, sicil, sulh, and cognates.
11. Crime: alet-i cariha, alet-i harb, avaz, cebr, darb, feseka, fısk, gadr,
gasb, harami, iftira, kan, kasame, katil, katl, livata, maktul, mecruh, öldür,
riba, rüşvet, şetm, sirka, taarruz, teaddi, tekfir, tezvir, urub, zina, zındık,
zulm, and cognates.
12. Punishment: kasame, diyet, habs, had, kısas, nefy, tazir, tazmin,
tenbih, zindan, and cognates.
13. Taxation, prebendal rights, extraction: behr, harac, has, iltizam,
miri, mukataa, öşr, reaya, resm, rüsüm-i örfiyye, sahib-i arz, tahrir, tapu,
tekalif-i örfiyye, tımar, and cognates.
14. Land (not private property): arz-ı miri, çayır, ihya, mera, mezra,
orman, tarla, toprak, and cognates.
15. Contractual – transactions, employment, gifts: akd, bağış, baha,
bayi, bedel, bey, ecr, faide, feraiz, fesh, gabn, galle, hibe, hisse, icare, istib-
dal, istiğlal, istimal, iştira, ivaz, kabz, kesb, kira, kıymet, masraf, mecür,
muamele-i şeriye, mudarebe, müflis, muhasebe, müstecir, müşteri, riba,
sarf, satmak, semen, şira, şirket, şufa, tazmin, temessük, ticaret, ücret,
vade, and cognates.
16. Contractual – debts and loans: ariyet, borç, deyn, gurema, ibra,
istidane, istiğlal, karz, medyun, müflis, rehn, taksim-i gurema, temessük,
vade, and cognates.
17. Ritual and worship: abdest, bayram, besmele, cami, cenabet, cuma,
dua, ezan, fatiha, gusl, hac, hutbe, ibadet, iktida, imam, kabe, kefaret, kıble,
kuran, kurban, mescid, mesh, mihrab, minaret, müezzin, musalla, namaz,
nezr, nisab, rekat, sadaka, sala, salavat, savm, secde, sema, sevab, sufi, tekbir,
teravih, tesmiye, tevbe, tilavet, zaviye, zekat, zikrullah, and cognates.
18. Religious doctrinal issues: ahiret, ahkam-ı şer, alim, asım, ayet, batıl,
bidat, cehennem, cennet, din, eimme, enbiya, facir, farz, fısk, günah, hadis,
hanef, haram, helal, ilm, imameyn, iman, islam, istiğfar, itikad, kızılbaş,
küfr, lanet, mekruh, memnu, mezhep, mübah, murdar, mushaf, müte-
deyyin, mutekid, nafiz, napak, nass, peygamber, resülullah, şafi, şeytan,
sünnet, tahir, tanrı, vacib, zındık, and cognates.
106 B O Ğ A Ç E R G E N E A N D A T A B E Y K A Y G U N

19. Religious, administrative functionaries and functions: imam, müez-


zin, narh, azl, berat, beytülmal, buyuruldu, ferman, hatib, hilafet, hitabet,
kadı, kitabet, mansıb, mazul, müderris, mürtezika, muvazzaf, naib, nasb,
nazır, padişah, ref, sancak, sipahi, sultan, tapu, ulema, vaiz, vezir, vilayet,
and cognates.
20. Communal matters: amme-i nas, çeşme, hakk-ı şirb, halk, işidilür,

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


karye, kasaba, kasame, köy, kura, kuyu, mahalle, mera, müslüman, nas,
pınar, su, şirb-i kadim, tarik-i amm, and cognates.
21. Non-Muslims: kafir, küfür, cizye, kenise, kilise, nasrani, yahudi,
zımmi, and cognates.
22. Representation, suretyship, guardianship: kefil, müvekkil, tevkil,
vasi, vekil, velayet, and cognates.
23. Agriculture and cultivation: ağaç, arpa, buğday, darı, eşcar, fidan,
gars, habbe, harman, hasad, keyl, mahsul, meyve, mısır, tohum, ziraat, and
cognates.
24. Food and beverages: arak, bal, hamr, hıyar, içmek, kahve, meyve, su,
süt, şirb, taam, yağ, yemek, and cognates.
25. Animals: davar, deri, deve, hınzır, inek, katır, koyun, kurban,
mukari, öküz, sığır.
26. Intoxicants: arak, hamr, kahve, afyon, esrar.
27. Illness, death, and injury: musalla, cerahat, cerh, defn, ecel, fevt,
firaş, hasta, helak, kan, maktul, maraz, merhum, mevt, meyyit, mezar,
mübtela, müteveffa, ölmek, öldürmek, sıhhat, vefat, and cognates.
28. Military matters: asker, cihad, ehl-i örf, gazi, harb, hisar, piyade,
sancak, sipahi, and cognates.

Appendix 2: Predictive capacities of thematic proxy lists

For the purposes of our analysis, we take specific words and phrases asso-
ciated with particular thematic categories as proxies or predictors, indica-
tive of the thematic components of the fetvas in which they appear.
Obviously, while a single word or phrase in a fetva might be suggestive
of that fetva’s thematic attributes, identifying a greater number of words or
phrases in the same fetva linked to a specific theme would be a more certain
indication of the fetva’s association with the same theme. Thus, while the
grouping based on one or more words and phrases per fetva would provide
larger coverage within the corpus, groupings based on three or more words
and phrases per fetva would be more accurate indicators of thematic attrib-
utes. The following table shows how groupings based on one or more and
three or more words/phrases per fetva compare in terms of their predictive
strength, based on random spot-checks.
MAPPING A FETVA COMPILATION 107
Based on 600 random spot-checks we conducted in fetvas that contained
only one word/phrase and only three words/phrases per fetva associated
with specific themes for each grouping, we observed the corresponding
rates of accuracy as 76 and 93 percent. This means that while 76 percent
of the fetvas that contained a single word or phrase linked to a specific

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


Table A.2: Predictive capacities of thematic categories

1W/P spot checks 3W/P spot checks

monetary 10 of 10 10 of 10
(property) real estate 7 of 10 10 of 10
(property) slavery 8 of 10 10 of 10
(property) other 8 of 10 8 of 10
(property) terminology 10 of 10 10 of 10
family 12 of 15 15 of 15
women 9 of 10 10 of 10
inheritance 8 of 10 10 of 10
waqf 8 of 10 10 of 10
gender and sexuality 3 of 5 3 of 5
disputing and litigation 9 of 10 10 of 10
crime 7 of 10 10 of 10
punishment 9 of 10 10 of 10
taxation/prebendal/extraction 7 of 10 10 of 10
land, not property 8 of 10 9 of 10
transactions/employment/gifts 8 of 15 15 of 15
debt and loans 8 of 10 9 of 10
ritual and worship 7 of 10 10 of 10
religious and doctrinal 10 of 15 15 of 15
religious, administrative func- 6 of 10 8 of 10
tionaries/actions
communal 6 of 10 7 of 10
non-Muslims 6 of 10 9 of 10
representation/suretyship/ 8 of 10 10 of 10
guardianship
agriculture and cultivation 4 of 10 8 of 10
food and beverage 5 of 10 5 of 10
animals 9 of 10 10 of 10
intoxicants 10 of 10 5 of 5
illness/death/injury 11 of 15 15 of 15
military 7 of 10 8 of 10
overall rate 228 of 300 (76.0%) 279 of 300 (93.0%)
108 B O Ğ A Ç E R G E N E A N D A T A B E Y K A Y G U N

theme to be actually related to that theme, the likelihood of the fetvas that
contained three words or phrases linked to a theme to be actually associ-
ated with the same theme was 93 percent.66 Our findings support the as-
sumption we made in the text that increasing the numbers of words and
phrases associated with specific themes would increase the likelihood of
correctly identifying the fetvas’ thematic attributes.

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


Why did we observe such a variation between the two ways of categor-
ization? The answer is related to the multiple meanings of many words and
phrases included in our lists. For example, the word bina, which means
‘building’ in the word’s noun form in Turkish could also mean ‘to build’, ‘to
construct’, or even ‘to make’, when used as a verb. Thus, while the word
could be a predictor for the theme of ‘real estate’ in its noun form, it could
also appear in a fetva in ways unrelated to this theme. Such words with
multiple potential meanings, when they appear as the only indicators of a
specific thematic content in a particular fetva, as is possible in the 1þW/P
group, could lead to false predictions. On the other hand, the predictive
problems associated with such words significantly diminish when we seek
three or more words or phrases per fetva to presume a thematic content.

Appendix 3: Word/phrase lists for selected LDA topics

Here are the word/phrase lists in stemmed forms for select ‘topics’ as iden-
tified by the LDA algorithm.
Topic 0:

bey vakf fevt hükm mülk hakim hibe mütevel kız şart oğlu icare arz bina vakıf
misl varis dükkan kabz verese ecr bulmak sıhhat izn baha oğlan rüç bahçe
kabul meşru helak fetva emlak zabt vakfiye kadı nafiz beytülmal fesh fasid
padişah akd maraz intikal kısmet mücerred şira müşteri mevt tescil vefa mas-
raf meblağ sükna arazi tereke tazmin bedel tahsil mahsul miri şeri evvelki
harab tevliyet şerait ıtlak akar ibaret eşcar ivaz hariç değirmen rıza batıl
mukataa meşrüh sabık ibtida hukuk vakfiyyet fehm memlük satmak sadaka
mukarrer amme muvaza kemali imameyn ocak razi ahz itibar bayi tağyir azl
gayet vefat hıfz

66
Here we should also point out that the 76 percent accuracy rate is valid for
fetvas that contained only a single word or phrase related to a specific theme. Since
the group of fetvas composed of one or more words/phrases also contained fetvas
with greater numbers of words and phrases for each category, the more realistic
accuracy rate for this group should be higher, perhaps closer to 80 percent. For the
same reason the more realistic accuracy rate for the fetva groups that contained
three or more words and phrases should be significantly higher than the 93 percent.
MAPPING A FETVA COMPILATION 109
Topic 1:
kavl küfr amel süre hak halk adam kaza cebr terk cemaat itikad ilm kabil
kuran red iman hayat haber abd ihtiyar özr ayet resülullah naib kadı sevab
mutad minbad tazir muhtar acz mahz şeyh sarih tilavet marüf kelam kemal
medine fısk dünya tenbih ehadi düğün kitab salah delalet nehy raz kelime

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


mahalli mukteza izhar tehir kıyas akl evvela mezkürlar haddi taam nass esrar
cüzi şerfa nakl fasl alim şerife tercih azap taalluk işaret bilfiil sahip makbül
marifet baş mektüb zındık vacib yemek tahkik hata ısrar tevcih şedid tefsir
mutekid efali hayr hale irade takrir hass adet eyleme kütüp azl kadilık
Topic 3:

niza ibadet kerahet eimme madde katı tecv deveran şafi hurmet raks urub
zikrullah tafsi taif itimad fasık cevaz ahkamışer cenap şerh edna yazup süfi
hacr temessük evza kibar akil mecür sünen ebussuüd tevhid tavil diniyye
müslime şevk istidlal kasır selatin alayim denilür deva fizamanina ibram
mürted rahimehullahi icap ictiha mevali naks talil gazzali zebiha mukteda
meşahir mülahaz telif hep uzamai salik sedt tesir ibahatin tefahhus alem
mutasavvıfe tezaif şimdik tenzih mezheb tergib mübi fırsat rek hakayik ahkem
zebhiye sudür kudsiyye dürl bast adap mürsel nefyen emar değm sade hitap
uzam muhabbe nezayir mesabe eyleyicek cumhür badezzebh tefekkür hıll

Topic 4:
tasarruf bağ sipahi öşr eda dahl resm tapu reaya mutasarrıf haraç değirmen
tımar tarla tamir ziraat malik mera sahip harac mahsul berat adın alınan
mezra davar rakabe feth hadis kayd temlik defter ariyet taarruz han muattal
müşterek mukarrer ibka kaim toprak öküz çayır tefvik müstakil itibar marifet
memleket harc menafi istiğlal mukata cehl şehir hasad sehm yazmak adet
arzıöşr edegel heman vilayet behr oturmak tatil verildik muvazzaf açmak
orman ferağat edegeldik umümen aleyhirrahmetü eben mukasem dam tasar
kullanmak malikane alan tereke mazul kısmet hark makule anve merhüm

Table A.3: Coverage rates for selected topics

% of topical topic 0 topic 1 topic 3 topic 4 topic 7 topic 9 topic 16 topic 20


w/p in fetvas

5 67.4% 35.3% 43.3% 28.2% 30.1% 10.8% 10.5% 10.7%


10 53.0% 18.1% 29.8% 15,4% 21.6% 6.7% 6.2% 6.9%
15 37.3% 7.7% 19.4% 7.9% 14.9% 4.1% 3.5% 4.2%
20 28.2% 4.5% 14.5% 5.3% 11.5% 3.3% 2.4% 3.2%
25 19.3% 2.2% 10.0% 3.3% 8.2% 2.2% 1.5% 2.1%
30 11.1% 0.7% 5.4% 1.8% 5.1% 1.4% 0.7% 1.2%
110 B O Ğ A Ç E R G E N E A N D A T A B E Y K A Y G U N

edemez mezari harem velgufran matekaddem kullandık rüsümu çubuk mül-


kiyyet mukaseme denilir ede

Topic 7:
ehl karye diyet mahalle maktul mecrüh düş meclis katil avaz darb akreb karib
sadi savt hür bulunduk husüsı nef kasame habs ehliörf nida işidilür ahali

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


düşmez taş istifsar şari teftiş tazir şedit masa subaşı olunurken konuk mahallat
sahip odası yal zincir karip maktü meclisişer kande şeyhülislam hürrülasl ölen
esvak hamr diş meyyi harman hataen kasd cese fakr tahmil karbansaray mas-
lüb kaimen şevari ırz cami kelime kulak baid hüsni kahve baş sada cinayet
mağara evliyai konduk luns mecl tecavüz hizmetkar hane yara şurb itibar
görür muhkem vacib kın muhakkak durmak saz mazınne feseka taşra ölür
ihmal ibret makberelik şenlik muallimzade erer
Topic 8:

namaz salat sünnet farz iktida rekat hac zuhür mukaddem vakt kıraat iade
selam secde beyn cuma musalli öğl vacib cami teravih mihrap kıl günlük sehv
nihayet kalkub fatiha kura şurb seç sür bilmeyen musalla kılan kılınan mevaz
kade zuhri ihfa kılınmak durmak sahip tilavet meharip yun tekmil gir aksi
mahreç okuduk feraiz meyyi cehr taşra tesbih kaldırmak münferi südüs heman
muhalif adet teşehhüt kılar edas kılındık yats can tehir tazir hamr riaye okuma
teşehhüd vitir okuma kılma evvelki tip rekati gece iste suret ihtimali kıble cüz
hatır müstehab kılma işa ikinç sübhanek türl şafii kılıverdüği ikin tali nafi
oturub telaffuz
Topic 9:

ev zarar men havale mani def havlu ihdas fahişe duvar ref muttasıl pencere
tahli üst sakine çardak değirmen şuf yük yıkdırmak sedd yapub esir olıncak
budak yapmak ziya şer değildür tahliye taş tazmin abhane ifrat baç taşra bazar
civar çekmek tazir hav gök olımaz müteahhir kapusı konşu müşterek konub
çayır zahib kaldırub rahimehümallah üleşdik müraat örtü perde mimar şefi
tevakku ayb yem habs hurüc iste fitn iledüb müstakil toz namahrem şerik
yanub üçün filvaki düzm gelmeyüb yapdırmak dut açs hava murür işhad yıkub
memnü girmek geçmek istibdal alabilür değir cami rayiha tehir ihrac nur
müşahed içeri alçak gidermek itikaf görülmek duyub

Topic 20:
mal vasi müteveffa mudarep gaib emin baliğ faide eytam vasiyet gaip harc
tereke yem sulh kısmet mevt nasb sarf sığar hıyanet tasdik rızk kesb kadı kıble
yetime taam altun alıvirüp maraz tazmin zekat getdüği şirket mekan nabaliğa
mefküd ortak menkülat beynelver ıss mudarib rüşt gasb şerik yetim yazılub
subaş tutulur kullanub hesab bez metrüka gabni badezaman helake bakır
taksimigurema rabbülmal müşrif havf ikrar akraba helali vesayet beytülmalcı
MAPPING A FETVA COMPILATION 111
gidel itlaf halt oluncak ölür vilayet alabilür gönder katip kabul müşterek
eksik kazanç şüreyh kalmayıcak firar mefküdu beytül faidesı alıkomak katt
israf bils şerikeyn ifraz muflis itibar gidüp sahip meb marüf tedrici hüsran
Since none of the modeling algorithms available to us provide any direct
measures to estimate the portions of corpus that their topics cover, we

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


devised an indirect measure to serve this purpose, based on the numbers
of words/phrases associated with a specific topic in a given fetva. If the
numbers of such topical words/phrases exceed a certain percentage of the
stemmed words/phrases in the fetva, we associated that fetva with the topic
in question. The total numbers of fetvas (whether in absolute numbers of
percentage form) associated with individual topics in this fashion gave us
indirect measures of coverage for specific topics. The next table provides
the percentages of the fetvas linked to specific topics at different percen-
tages of word/phrase representation. As is clear in the table (see next page),
as the expectation of topical word/phrase representation increase, cover-
age rates decrease.

Appendix 4: Categorical ‘others’

Table A.4 presents the words/phrases for a few example topics that could
not be categorized with the help of our thematic proxy lists.
Our examination of the terms in remainder lists indicate that they are
closely related to the dominant theme(s) in a particular topic. For example,
for topic 3, in which ‘disputing and litigation’ appeared as the single most
dominant theme, the remainder list contained 19 words/ phrases (of 53
total) that could potentially be related to matters involving disputing and
litigation, including adalet (justice), hüccet (legal document), adavet (hos-
tility), and muvazaa (collusion). Similarly, for topic 7, which prominently
featured the themes of ‘family’ and ‘women’, the remainder list included 23
words and phrases (of 59 total) that had direct relevance to the themes,
including, for example ‘Zeynep’ and ‘Hint’, boşamak (to divorce), müeccel
(deferred), and muaccel (immediate or prompt).
Many terms are not included in the thematic proxy lists because they
appear fewer than ten times in the corpus. Often such terms are the mis-
spelled or differently-spelled versions of the ones included in the proxy lists.
So, while the LDA algorithm identifies them as connected to the dominant
theme(s) in a particular topic, the thematic parsing process fails to pick up
the fetvas that contained them, if the latter featured no other relevant proxy
term. For example, fetvas that spell ‘Hind’, a frequent female name, as
‘Hint’ are included in the lists of ‘women’-themed fetvas unless they con-
tain other related thematic proxies.
112 B O Ğ A Ç E R G E N E A N D A T A B E Y K A Y G U N

Table A.4: Remainder lists for selected topics

Topic dominant categorical remainder lists


themes

3 ‘disputing and litigation’ meclis adalet ayb sabit hüccet mahzar taleb
huzürun ihzar muvaza şahs aharden istihkak

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


teftiş tasdik adavet kizb ibraz tesamü
7 ‘women’ and ‘family’ zeynep hamlim müeccel boşamak mahram terbi
liüm nabaliğa hint sağire erin doğur muhala
tefrik yetime liebb emzir nafakas halvet
zımmiye badelbülüğ hatu muaccel
9 ‘ritual and worship’ kılarken sübhanek tesbih kıraat kılmamak işa
okunur cenaze mağrip kalkub kılma okuma
şafak kılm sehiv şafii kerih kılındık kılınmak
teşehhüd farziyyet fecr vudüu kılınan tanr
selam kılıverdüği ikindi kılar niyet kılan kırae
okum zeval zevale leyleikin kaldırmamak
kaldırmak rekati ihfa öğl kaldır okumayub
15 ‘religious and doctrinal’, muhterik konşuluk fukarai serhoş keşiş sadaka
‘communal’, and ‘non- tecdit avlu cimalafz semavi aleyhisselatü
Muslims’ siyase müslümanlık kafirlik ayin cemaa çalub
inayetillahi bihuzür orusp müslümi fakih
Dubrovnik
17 ‘slavery’ and ‘transac- satduğı satabilir kaçt hizmetlenmek kaçır satıl
tions/ employment/ ayartmak darulharbden hizme mükatep rıza
gifts’ meccanen alıver satılan gasıp kaçkun
alıvermeyüp memlükun gasıb azadlık kulluk
kulı ıtakname mudarip müdebbere ıtkname
hürr kulu hizmetlendik alelmal kaçub dutub
kaçku
35 ‘agriculture and erik büyüyüb emrüd büyütüb palamut meyves
cultivation’ budayub bahasi fida çekirdek çam budamak
kök eşle kesmemek aşılama asma zeytu dike
aşılayıverdüğin aşılayub kuruyan aşılaya
kavu diküp keser kestirme aşılamalr aşıladık
aşıl gölg dikt aşıladuğın aşılamak kesdirmek
aşılaş aşıla kestane diken
MAPPING A FETVA COMPILATION 113
Some other words/phrases that frequently appear in fetvas featuring
specific themes and are picked up by the topical analysis were deliberately
excluded from thematic lists based on our concerns about false identifica-
tion in the supervised analysis. For example, while the term adalet (justice)
has direct relevance to litigation, we chose to exclude it from our proxy list
for ‘disputing and litigation’ because the term’s link to the theme cannot be

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


assumed to be near-exclusive. Similarly, while the terms müeccel and muac-
cel often appear in the fetvas related to ‘family’ to differentiate various
types of dower, we elected to omit them from our predefined lists because
they possess other contextual uses.

Appendix 5: Omitted Topics

Table A.5 presents examples of topics with very limited coverages. Topics
41, 57, and 58 largely comprised infrequent terms, including mis-spelled or
alternatively-spelled versions of the ones included in the proxy lists.
Topics 43, 49, 54, 55, 70, 72, and 96, on the other hand, identify fringe
thematic concentrations or appendages including washing and nakedness
in a bathhouse, bodily impurity from associating with animals, accidental
fires and their consequences, permissibility of consuming sea animals, fes-
tivals, jewelry, and, interestingly, the renowned Sufi Mevlana Celaleddin-i
Rumi.

Table A.5: Word/phrase lists of omitted topics

topic impressionistic thematic related words/phrases in the list


content

41 soap, property, com- sabuncı florilik lütfun münzev avarız kasaba


merce, taxation ekinç hoş şerr tevekkül nalband münacat
sabuncunı kalmayub bazargah lorilik ran
vafir mutazazarrır sabun avar mania pare
tazarru yaramaz istiaze mercüdur varduk
43 washing and nakedness üryan hama mekşüfülavret onat ebussüd ferş
in public space / uryan maşraba müstekreh teye eders vacib
bathhouse ağzı mezbe cenabe kubulün müstahab hamam
maza yakmaz sayim vudüa tebevvüli gözet-
mek kertme guslü sils bulucak sevp yakar cuk
ekalde yumak teberrü temizleyegel mermer
ıslanub niyets gasli gusle fasil odun kunduk
esvab ehaf teyvemmüm yunduğı müstamel
(continues)
114 B O Ğ A Ç E R G E N E A N D A T A B E Y K A Y G U N

Table A.5: continued

topic impressionistic thematic related words/phrases in the list


content

bedih lasr futa azas debbag saim benzer dübür

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


istinşak dirsyeter katran terleyüb nesnedür
olunucak buhar mazmaz gözet yunma gargar
kuşanıb yuduk teannü mübalak vudüda yuma
ovmak yumayub yuns bırakıb ruyan iğtisal
nicedür istinca bağdad cenabet çıkmak beden
temir mubalağa değişme bihasep mutahher
murder şükür damlayan değmek gusl gezüb
mai kepek meks
49 religious impurity based yuyup değs eşribe çanak dub encase yunan kur-
on contamination or uyub sıkdık meşkük sabu kesilen mutahhir
association with edviye boyayıcak yıkanıcak bulaş işledüb
animals damlamaz reng deri bağat tahir sabbağ olun-
cayadek kurutmak yıkayıcak sultant çuka
boğa tıraş neci boya hımar perdahtlatmak
çıkardık ısl kelb asılmak pişen debagat esvab
katub talaş sahtiyan tetahhuri debağat göbek
sıkmak
54 fire, burning, accident pişirib premeç preme tokuşub orman ola-
mayıcak muhakkak meşe iskele yakarken
savul yakduğı dol tuşub yanmayub çüğneyüb
devalık depüb yakıcak yans kebab yakdık
evlik durmak boğaz yakub yükü üreyüb yonul
ihtimali esüb rüzgar yel dinlendirmek dep
vulmayub yaks ulaşmak vurub uğradık ateş
dağıtub tutuşub
55 sea animals, sea food, öldürüb midye kırs kuşçu kerahe alelaks arul
slaughter of sea istihzaen istirye veciz aru tüts alunur ıstakos
animals yedür zekat istakos maktülu aşüra kerevid
kova kokar zahiriyye kavm yengeç kamet
mürtef arıs sebilittedai buhlden kerevit
menetmek müştebih nahiye irtifa tahrim
öldür sath istiridye debt debüb mekrüh kıçı
57 taxation, differences buyurulan rir fakirulhalden nafe gani satulhal-
among classes dir sulah ihlali faik evsat vasatulhalin vasatul
dirhe yukarusı vasatulhal fakirulhalin cizye
(continues)
MAPPING A FETVA COMPILATION 115
Table A.5: continued

topic impressionistic thematic related words/phrases in the list


content

58 agriculture and dabbe sürs suladuğı tohum üleşür bitme kay-

Downloaded from https://academic.oup.com/jis/article/32/1/62/6048526 by Sabanci University user on 11 October 2021


cultivation nadub saçub ekse müzaraai tüdüb kışr bitdik
zabtedüb yemen baran içmeyüb çel yetişdük
kıy kanünı hasıli ekdik herk eküb deştbani
saçm saçar kablelhasad müzara doküb ekdir
nadasu ketan emirs lazımdüur amfr yürüdüb
kabuk istimaolunur eks
70 festivals, işlemeyüb temaş seyr arifane isabet riz muba-
new year celebrations eret uzatmamak adat nevruz mecüsi tazimen
tapars işlemezs iyd hırfet tazim guslün nevrüz
iyt yağr sohbet ekmek aztm zuhre ilyas
72 combination of medical yağlayub ısırub hatem zebercet mübeddel cer-
and jewelry rah kuşak namaaz dost zümrüt kaş yüzük
hasiyyet güder kabız makat mevzünca akik
yemeni nasur yakütdan anın torp parmak
halka yaşl hizmetlenüb dizg pirüze kazdırmak
basurı
96 Mevlana Celaleddin-i celalet hünkar mesnevi errümi söy kasem türk
Rumi dil ışık zemm hanis

You might also like