Information on borrowed affixes was compiled from a variety of sources, as explicitly indicated for each case. Most information comes from published sources, especially descriptive grammars and other descriptive studies on, e.g., language contact or morphology. Different sources on the same language were consulted wherever possible. In many cases important additional information (and in some cases, all information) comes from personal communications from experts on the languages in question.
The sample of languages includes in principle all cases of affix borrowing that have come to our attention between 2007 and 2026, i.e. no attempt has been made to make the sample genealogically or areally balanced. If two or more pairs of languages (or dialects) are very similar in the aspects relevant here, only one pair has been included in the database, namely the language pair with the higher number of borrowed affixes. For instance, Chuvash (Turkic) affixes in Mari (Uralic) are included in the database, but excluded is the similar set of Chuvash affixes that the Mordvinian languages borrowed, which are closely related to Mari.
There is a clear bias in the language sample towards those language families and areas that are linguistically best described, especially European/Western Eurasian and Oriental languages. This is because detecting affix borrowing requires relatively detailed information not only on the recipient and donor language but crucially also comparative evidence from both of these languages for the proof of borrowing and determining the direction of borrowing.
Included in the sample are two languages that are often considered as "mixed languages", Gurindji Kriol from Northern Australia and Copper Island Aleut from the Commander Islands in the Bering Strait. Unlike other mixed languages, it is possible for these two to clearly identify one language as the matrix language (Myers-Scotton 2002; Myers-Scotton 2007), which contributes the morphosyntactic framework as well as a substantial portion of the vocabulary. This language is identified as the recipient language (English for Gurindji Kriol and Aleut for Copper Island Aleut) and the other contributing languages as the donor language (Gurindji for Gurindji Kriol and Russian for Copper Island Aleut), following Meakins' (2011) analysis of (English-based) Gurindji Kriol as having borrowed Gurindji case markers. Note that in AfBo, Gurindji Kriol and Copper Island Aleut do not appear as borrowing exceptionally many affixes: Gurindji Kriol is the sixth most heavily affix-borrowing language in the sample, Copper Island Aleut the eight (sharing that rank with Khanty, Assamese, and Vietnamese).
Any morphologically bound form from a closed class that fulfills a derivational or inflectional function counts as an affix in AfBo. By this definition, clitics are included as instances of affix borrowing, as long as they fulfill a derivational or inflectional function, e.g. tense, evidentiality, or topic marking. Arguments for considering a form is bound vs. free are explicitly given in the descriptions where possible.
Some forms are included here that might be considered morphologically conditioned allomorphs because they fulfill the same function in different environments. For instance, a plural marker used with animate nouns and a plural marker used with inanimate nouns are counted as two borrowed affixes.
An affix is considered as effectively borrowed only if it is attested in at least some hybrid formations, i.e. combinations of the borrowed affix with native stems. An affix is not considered as borrowed if it is only attested in complex loanwords, i.e. in combination with stems that are borrowed from the same language. In addition, a complete proof that a given affix is borrowed would ideally include (i) evidence that the borrowed affix was not present in the recipient language before contact, (ii) evidence that the source form was present in the donor language at the time of contact, and (iii) evidence that the similarity between source form and borrowed form is not coincidental. Even though the sources consulted for AfBo rarely if ever explicitly provide such complete information, the authors of these sources are often authorities in the language families concerned, which gives credibility to their judgments that a given form is borrowed.
Data are coded for a number of properties for comparative analyses (note that parts of this information are included in the web interface, while other parts are included in the downloadable database). Information on recipient languages involved in affix borrowing include:
For each language pair, additionally the total number of borrowed affixes is given. Note that the previous version of AfBo (but not AfBo 2.0) also included information on the total number of interrelated borrowed affixes in the sense of Seifart (2012) and on the reliability of borrowed status/affixhood. The latter was omitted in AfBo 2.0 since all cases with a low reliability were removed.
The information provided for each borrowed affix consists of the following:
Data are given throughout using the transcription conventions provided by the original sources, i.e. no attempt at standardization through transliteration was made.
If you refer to information exclusively stemming from sources that are cited in AfBo, those original sources should be consulted and cited, optionally in addition to AfBo. If you refer to analyses provided by AfBo or results obtained from AfBo, such as the frequency of borrowing affixes with a specific function, you should cite AfBo as
Seifart, Frank, and Gardani, Francesco. 2026. AfBo 2.0: A world-wide survey of affix borrowing. Leipzig: Max Planck Institute for Evolutionary Anthropology. DOI: 10.5281/zenodo.3610154
Many people have contributed to building AfBo, as explicitly noted in the descriptions of individual cases of affix borrowing. Tsan Tsai Chan and Nikita L. Beklemishev provided substantial research assistance for AfBo 2.0. Tsan Tsai Chan contributed to 10 language pairs (Coos - Alsea, English - Japanese, English - Mandarin, English - Spanish, English - Tagalog, French - Turkish, Middle Chinese - Japanese, Middle Chinese - Vietnamese, Middle Low German - Danish, and Spanish - Tagalog). Nikita L. Beklemishev contributed to 20 language pairs (Azeri - Udi, Belarusian - Lithuanian, Central Asian Turkic - Russian, Chamacoco - Kadiwéu, English - Welsh, French - Russian, German - Estonian, German - Polish, Hebrew - Yiddish, Italian - Arbëreshë Albanian, Kurdish - Chaldean Neo-Aramaic, Kurdish - Suleimaniya Jewish Neo-Aramaic, Lombardic - Tuscan Italian, Middle Low German - Russian, Molisano Neapolitan - Slavomolisano, Nivaclé - Zamucoan, Russian - Estonian, Spanish - Nahuatl, Tasawaq - Tamajeq Tuareg, and Upper Chinook - Bela Coola). In addition, we are grateful to Robert Forkel for converting AfBo to CLLD, and to Lena Sell, Lisa Steinbach, Evgeniya Zhivotova, and Senta Zeugin for extensive proofreading.