Linguistic Evidence for the Borrowing and Origin of “Matchmaker” in Liangshan Yi

Abstract

This study uses linguistic evidence to examine the cultural dimensions of the term “matchmaker” within Liangshan Yi society, Sichuan, China. The present linguistic analysis reveals that two distinct Chinese words related to marriage were borrowed into the Liangshan Yi language, based on morphological analyzability, cognate comparison within Sino-Tibetan languages, and phonological and semantic adaptation. Specifically, (i) the Chinese word 伐 (fá), meaning ‘to hew, chop, or matchmake’ was borrowed as a standalone verb, phonologically adapted as ho³³/fu³³ depending on dialect into Liangshan Yi with the meaning ‘to marry (generally)’; and (ii) the term 伐柯 (fákē), originally meaning ‘to cut wood for an axe handle’ and later to ‘matchmaking, matchmaker,’ was borrowed into Liangshan Yi as a noun meaning ‘matchmaker’ (hɔ⁴⁴ka³³/fu⁴⁴ka³³, depending on dialect). Phonological evidence suggests these borrowings occurred during the Late Middle Chinese period, prior to the completion of labiodentalization, but after the shift of labial consonants to labiodental sounds, between the late Tang and Song dynasties (ca. 9th to 13th centuries). The lower limit is 1324 CE, when Zhōngyuán Yīnyùn was compiled. The dating of these borrowings is corroborated by our further corpus analysis of the usage of 伐柯 fákē and 伐 fá in ancient Chinese texts. This analysis reveals that while the two Chinese words originated in Shījīng (Book of Odes, 11th–7th centuries BCE), their usage frequency was extremely low in Early Middle Chinese; a resurgence subsequently occurred in Late Middle Chinese, with the highest frequency attested in the Song Dynasty (960–1279 CE). Moreover, a survey of additional Loloish communities shows that 伐柯 fákē serves as the common source of the borrowing for the term “matchmaker” in many Northern Loloish and some Central Loloish varieties, reflecting the influence of Chinese marital customs across a broad range of Yi communities. Although the borrowing derives from a common source, given the similar phonetic forms, the terms for “matchmaker” in these Loloish languages lack sound correspondences. This indicates that the Loloish languages had already diverged from one another by the time the borrowing took place. Finally, the linguistic evidence from the present study illuminates the historical processes that shaped marital customs among Yi ethnic societies, demonstrating that the concept of matchmaker is a recent innovation within Yi cultural practice, linguistically and culturally borrowed from Han Chinese—probably during 9th century to early 14th century—rather than being an indigenous tradition.

1. Introduction

As a Loloish (Tibeto-Burman) population of over two million, the modern Liangshan Yi people, mainly living in Liangshan, Sichuan, and adjacent mountainous regions of Yunnan, Southwest China, practice ethnic endogamy, clan exogamy, and rank-based endogamy as their marriage conventions.

Firstly, ethnic endogamy of the Liangshan Yi speakers requires individuals to marry within the Yi ethnic group. While most marriages occur within the Liangshan Yi community, marriages with ethnic Yi from Yunnan and Guizhou are also acceptable, although Yi speakers in Yunnan and Guizhou speak different Loloish languages. In traditional Yi society, marrying individuals of other ethnic groups, who often constituted the slave caste, was nearly impossible, but in modern society, such marriages are possible, though still generally not preferred (Z. Wang & Zhang, 2018). Next, as a patrilineal society, the Liangshan Yi community practices clan exogamy to prohibit marriage within the same patrilineal clan. This practice serves to prevent consanguineous marriages and ensure lineage continuity, and functions as a defensive measure to expand and strengthen the clan’s influence and power. This practice is still strictly observed in Liangshan as an important moral law.

Furthermore, the above two marriage rules are reinforced by strict rank-based endogamy. The traditional Yi society is broadly divided into four hereditary castes, namely ndzɨ³³xo²¹ (lord:caste) ‘the ruling class’, nɔ³³xo²¹ (black:caste) ‘the noble class’, tɕ^hu³³xo⁴⁴ (white:caste) ‘the commoner’, and dʑi³³ ‘slave’ (also meaning ‘enemy’). Rank-based endogamy mandates marriage within one’s social stratum, and cross-caste marriage is thus prohibited. Nowadays, some exceptions to this rule are occasionally observed in more developed and open Yi communities.

Due to the social conventions and sensitive nature of information involved, such as familial background and caste, the matchmaker in the Liangshan Yi society is entrusted, as a crucial intermediary, with collecting and verifying these essential details for satisfactory marriage arrangements. Consequently, the matchmaker’s involvement is integral throughout the Yi marriage process, which typically unfolds across four interrelated stages: the proposal stage, the bridewealth negotiation, the engagement phase, and the final wedding ceremony (see Luobian & Mose, 2014; Qubi, 2017).

Scholars, as well as ethnic Yi people, consistently describe the matchmaker as a longstanding marital component in the Liangshan Yi society (e.g., Feng & Yuan, 2019; J. Zhang, 2019). This view is further reinforced by the presence of “matchmaker” in the Yi genesis epic hnɯ³³ɣo³³t^hɯ²¹ʑi³³ (see Zhu, 2017, p. 305; Hu, 2020, p. 390).

However, hnɯ³³ɣo³³t^hɯ²¹ʑi³³ should not be considered as a reliable source for inferring the presence or absence of the concept of “matchmaker”. As the Liangshan Yi society has been primarily oral, with the ethnic language transmitted and used orally among its speakers, the epic, although believed to have a long history, was only written down recently. The first Chinese translation was produced in the 1960s, and the Yi-language version in the 1970s after the initial standardization of the Nuosu scripts. In the epic, lexicons of different historical layers, as well as different dialects, co-exist, showing that the extant epic is not homogeneous and not entirely archaic. For example, va³³, as an earlier form of Proto-Liangshan Yi (PLS) *ɣa¹ ‘chicken’, co-occurs with ʑɛ³³, a later form, in the epic; another case is the co-occurrence of two functionally identical negators a²¹ and ma²¹ (PLS *ma³). Additionally, recent Chinese loanwords appear in the epic, such as hɛ³³ŋga⁵⁵ ‘ethnic Han Chinese or 汉家’ and ma³³pɛ³³ ‘Mabian as a place name or 马边’. The latter, as a place, was only established around 1764 during the Qing Dynasty as a sub-prefecture in Sichuan. The native Liangshan Yi word for Mabian is mo²¹xo³³la³³da³³. According to Zhu (2017, p. i), the content of hnɯ³³ɣo³³t^hɯ²¹ʑi³³ consists of both ancient events and modern events, suggesting continuous revisions and additions made by later generations during its long process of oral transmission and eventual codification.

Instead, we use Manshu (also historically known as Book of Southern Barbarians, from a Han-centric perspective) as a historical source to evaluate the presence or absence of “matchmaker” among the ancestral populations of what is now Liangshan Yi society. Manshu, written in 863 by the Han Chinese official Fanchuo, is a comprehensive Tang Dynasty text detailing the social structures and customs of early southwestern ethnic groups. It classified many Loloish tribal groups in southwestern China as Wuman (literally “Black Barbarians”), and others as Baiman (literally “White Barbarians”, meaning more Siniticized). The Liangshan Yi ancestors belonged to the group of “Black Barbarians”, known for their preference for the color black and resistance to Han assimilation, contrasted with the more Siniticized Baiman. Chapter 8 of Manshu, titled Customs and Manners of the Barbarian Peoples, describes the marriage customs of both the Wuman and the Baiman. However, it contains no mention of “matchmaker” or any descriptive expressions related to a matchmaking role, likely suggesting the absence of the concept by that time1.

Furthermore, linguistically, the Liangshan Yi word ‘matchmaker’ hɔ⁴⁴ka³³ (Niesu dialect) or fu⁴⁴ka³³ (Nuosu dialect) is not transparently analyzable (see Section 4), and the combined meaning of their constituents fails to yield a relevant semantic match. The morphological opacity suggests a non-native etymology as it aligns with the cross-linguistic criterion for identifying potential loanwords through analyzability (Haspelmath & Tadmor, 2009). Namely, if a word is analyzable in one language but not in another, it was likely borrowed from the former (see Section 4.1).

Consequently, the present study aims to prove that “matchmaker” is a Chinese loanword into the Liangshan Yi language: hɔ⁴⁴ka³³ (Niesu dialect) or fu⁴⁴ka³³ (Nuosu dialect). At the same time, ho³³ (Niesu dialect) or fu³³ (Nuosu dialect) as a standalone verb meaning ‘to marry (generally)’ was also borrowed from Chinese, which makes the first syllable of “matchmaker” identifiable in Liangshan Yi. The source words are 伐柯 (fákē or fa³⁵kʰɤ⁵⁵ in modern Standard Mandarin) ‘to cut wood to make an axe handle; matchmaker, matchmaking’ and 伐 (fá or fa³⁵ in modern Standard Mandarin) ‘to hew, to chop, to matchmake, or arrange the marriage for someone’, from Middle Chinese as the donor language. Accordingly, the first research question of the present study is:

(1)

Is “matchmaker” in the Liangshan Yi language a loanword of 伐柯 (fákē) from Chinese? How did the Liangshan Yi language adapt the word phonologically and semantically?

Next, lexical borrowing often occurs under conditions of language contact and cultural diffusion, particularly when new social institutions or practices from the donor culture, which are previously unfamiliar or substantially different from endogenous traditions, are introduced into the recipient community (e.g., Sayahi, 2014, p. 124). Thus, if “matchmaker” was borrowed, the practice itself may not be as “traditional” as was commonly believed by the ethnic Yi, which challenges the assumption of its indigenous origin and leads to the second research question:

(2)

When did the term for “matchmaker” enter the Liangshan Yi lexicon, and hence when did the associated cultural practice likely emerge?

2. The Language of the Liangshan Yi People

The Liangshan Yi language is spoken by approximately 2.5 million people, which are officially recognized in China as the ethnic Yi group (彝族). According to Bradley (1997, 2007), the Liangshan Yi language belongs to the Northern Loloish (or Northern Ngwi) branch. It distinguishes itself from other Loloish languages in Yunnan and Guizhou by two features: (1) retention of the proto-feature of prenasalization; and (2) post-nasal voicing of the following oral obstruents (F. Wang, 2013). In contrast, other major Loloish languages, including Weishan Yi (aka Lalo, Central Loloish in Yunnan), Nanhua Yi (aka Lolopho, Central Loloish in Yunnan), and Phula (Southeastern Loloish in Yunnan), have largely lost the prenasalization. Meanwhile, Northern Loloish varieties, such as Luquan Yi (aka Nasu, Yunnan) and Bijie Yi (aka Nyisu, Guizhou), retain prenasalization but do not consistently exhibit post-nasal voicing, with the following oral obstruents being either completely or partially voiceless (see Zhai, 2011; Pu, 2016).

Nuosu proper is a term used by Lama (2012) to denote the Liangshan Yi language. Two principal dialects are identified, Nuosu and Niesu, as illustrated in Figure 1 below. Both Nuosu and Niesu serve as autonyms for the Yi people in Sichuan, meaning ‘black(-worshipping) people’. Liangshan Yi is a verb-final, analytic tonal language with limited inflectional morphology. Grammatical functions are expressed primarily through clitics and postpositions, often employing a topic–comment structure. According to the 2010 census of Liangshan Prefecture and the research conducted by S. Chen et al. (1985) and Gerner (2013), approximately 1.5 million Yi in Sichuan speak the Nuosu dialect, while around 1 million use the Niesu dialect.

Moreover, the mutual intelligibility of the Nuosu and Niesu dialects is not high (Bradley, 2001), with notable differences in lexicon, phonology, and structure (Z. Pan, 2001; Matisoff, 2006; Hajek, 2006; Lama, 2012; Mahai, 2015; Ding & Lama, 2021). However, their phylogenetic relationship is well-established. Based on shared innovations, the Liangshan Yi language can be subgrouped into the Northwestern branch (Xide and Mabian) and the Southeastern branch (Huili, Puge, Butuo, and Jinyang); see Figure 2. Notably, speakers of the Northwestern branch dialects consistently employ the autonym Nuosu, while most Southeastern branch dialects use Niesu, except for Jinyang where the term Nuosu is used. Jinyang, a region situated along the border between Sichuan and Yunnan immediately adjacent to the Jinsha River, is likely a key dispersal center for Liangshan Yi speakers, as indicated by the presence of mixed patterns of shared innovations in the area.

From a historical perspective, dialects from the Northwestern branch tend to exhibit palatalization of the consonant clusters in the Proto-Liangshan Yi consonantal inventory, while those from the Southeastern branch tend to show retroflexion of these clusters. For example, the initial of ‘star’ is reconstructed as *kr- in Proto-Liangshan Yi (also see Lama, 2012). In modern Northwestern branch dialects, the main reflex is mu³³tɕi³³ (sky:star) ‘star’. In contrast, Butuo Jizhi uses mu³³kʂɨ³³ (sky:star) ‘star’, and Butuo Tuojue and Puge use mu³³tʂɨ³³. The Huili Niesu dialect likely underwent further development by shifting the place of articulation from retroflex to alveolar, resulting in mu³³tsɨ³³ (sky:star) ‘star’. The consonantal development of ‘star’ is summarized as *kr > kʂ > tʂ > ts in the Southeastern branch and *kr (>k) > tɕ in the Northwestern branch with a possible intermediate stage of *k before it became palatalized.

Moreover, dialects in the Northwestern branch are characterized by vowel raising, whereas those in the Southeastern branch display vowel fronting. For instance, the two autonyms Nuosu and Niesu are transcribed as nɔ³³su³³ and nɛ³³su³³, both deriving from the Proto-Liangshan Yi *nɑ¹, meaning ‘black’. Other similar examples include dɔ³³ vs. dɛ³³, meaning ‘the first half of the month’, and tɕ^hɔ³³ vs. tɕ^hɛ³³, meaning ‘to take apart’. In all these cases, the Northwestern branch underwent the vowel raising *ɑ > ɔ, whereas the Southeastern branch experienced vowel fronting *ɑ > ɛ.

3. Methodology

This section outlines the methods for identifying inherited words versus loanwords in Liangshan Yi.

3.1. Identification of Words of Inheritance and Borrowing

Every loanword is a linguistic time capsule of cultural exchange, entering the recipient language at a specific historical moment with an inherent temporal signature. This is particularly feasible to study Chinese loanwords in the Liangshan Yi language, as well as other minority languages with intensive contact with Chinese, due to the well-documented history of Chinese phonology, preserved through a series of dated rhyme books (e.g., Qièyùn, ca. 600 CE; Yùnjìng, ca. 1000 CE).

However, almost all existing studies focus on modern borrowings from Chinese into Liangshan Yi after the Qing Dynasty, such as terms referring to ‘bullet’ tsɨ³³ta²¹ and ‘automobiles’ tʂ^hɯ⁴⁴tsɨ³³ (Zhu & Zhang, 2005; Jiaoluomu, 2015; Azimo, 2020). Chinese loanwords of greater time depth in the Liangshan Yi language remain largely underexplored.

The existing studies mainly rely on the lexical phonetic and semantic similarity between the recipient language and donor language, namely that loanwords are recognized when their form and meaning closely match those of a potential source language, provided a plausible contact scenario exists and other explanations—such as a common ancestor—can be ruled out (Haspelmath & Tadmor, 2009, pp. 43–44).

Yet this similarity-based approach is more effective for recent borrowings, as older loanwords often undergo significant sound modification over time (J. Wang, 1962). Consequently, many ancient Chinese loanwords in Liangshan Yi have been phonologically adapted, making them dissimilar from their original Chinese forms and difficult to identify.

Zhu’s (1997, 1999) studies are the only works attempting to explore the Old and Middle Chinese loanwords in Liangshan Yi. However, his identification was still based on similarity, rather than sound correspondences. As a result, many of his so-called Chinese loanwords are in fact Liangshan Yi native words which share a common ancestry with Chinese, such as bu³³ ‘to float’ (cf. Middle Chinese bǐəu).

Therefore, a series of criteria are applied in the present study to distinguish between inherent Liangshan Yi words and Chinese loanwords, besides the above-mentioned similarity approach. To address words of greater historical depth, morphological analyzability and the Explicability Principle are used. The former states that if a word can be decomposed into meaningful morphemes according to the language’s morphological rules, it is unlikely to be a borrowing (Haspelmath & Tadmor, 2009). This principle can be illustrated with examples from both European and Asian languages. The German grenze ‘border’ is unanalyzable, whereas its Polish source granica is transparent, composed of the stem gran- and the suffix -ica. Similarly, the Cantonese word马蹄 (maa⁵tai⁴, meaning ‘water chestnut’) is an unanalyzable term borrowed from the areal Kra-Dai languages. Its Zhuang cognate mak daez is made of two morphemes: fruit + earth, literally ‘ground fruit’.

B. Chen (1996, p. 81) further pointed out that the fundamental unit of borrowing is a word or a larger linguistic unit, rather than bound morphemes/roots. For example, while the Chinese morpheme ‘three’ does not appear independently in Liangshan Yi, the entire word ‘third’ is borrowed, namely ti²¹sa⁵⁵, to express the ordinal concept.

Moreover, the Explicability Principle is derived from F. Wang (2006) and based on historical comparative analysis, namely whether a word can be explained through the phonological development of the donor language, and whether the word is cognate with morphemes in other languages/dialects of closer phylogenetic branch. Specifically, if a word in a language cannot be accounted for by regular sound changes from its proto-language but aligns with forms in another language (e.g., Middle Chinese or Old Chinese), it is likely a loanword. On the contrary, if the word is not phonologically explicable by a donor language but is cognate with related words in its sister languages/dialects, it is considered a native word, which is known as the Inexplicability Principle in F. Wang (2006) (see

Table 1. Historical comparative criteria to distinguish words of inheritance and borrowing.

Whether a Word Can Be Explained Through the Phonological Development of the Donor Language	Whether the Word Is Cognate with Morphemes in Other Varieties from Closer Phylogenetic Branch	Result	Notes
Yes	No	Borrowing	Explicability Principle
No	Yes	Native	Inexplicability Principle
Yes	Yes	Undecided	Older shared ancestry or uncertain origin
No	No	Undecided	Lexical replacement/innovation or another donor language

).

Additionally, if both historical comparative criteria yield a ‘yes’, there are two sub-scenarios: (a) if the suspected donor language belongs to the same language family as the recipient language, the target word is a native word originating from an earlier shared historical layer; (b) if the suspected donor language is not phylogenetically related to the target variety, the situation becomes complex and cannot be definitively decided—though this scenario does not exist in the present study. However, if both criteria yield a ‘no’, the word may either be a native word that has undergone lexical replacement/innovation or a loanword from a different donor language.

3.2. Data of the Present Study

This study uses six Liangshan Yi dialects (Northern Loloish) for the comparative analysis. Furthermore, four Loloish languages from Yunnan and Guizhou, namely Luquan Yi (aka Nasu, Northern Loloish), Weishan Yi (aka Lalo, Central Loloish), and Bijie Yi (aka Nyisu, Northern Loloish), Azha (Southeastern Loloish), are included as sister branches for higher-node comparison. These languages represent three major branches of the Loloish languages—the last one is Southern Loloish—according to Bradley’s (1997, 2007) classification. These languages were sampled because their speakers are collectively recognized as ethnic Yi in China. The purpose is to take care of the Chinese emic perspective upon the ethnic Yi (S. Chen et al., 1985; K. Chen, 2010). A major population of Bradley’s (1997, 2007) Southern Loloish branch is recognized in China as ethnic Hani, not ethnic Yi. Moreover, we have not yet identified well-resourced Southern Loloish languages spoken by the ethnic Yi for comparison (see

Table 2. The data of the present study.

Data Point	Address	Sources
Butuo (Northern Loloish)	Tuojue, Butuo, Liangshan, Sichuan	Own data
Xide (Northern Loloish)	Heboluo, Xide, Liangshan, Sichuan	Own data
Jinyang (Northern Loloish)	Luojue, Jinyang, Liangshan, Sichuan	Own data
Huili (Northern Loloish)	Xiaoheiqing, Huili, Liangshan, Sichuan	(G. Chen, 2022), https://zhongguoyuyan.cn/point/60828, accessed on 2 January 2025
Puge (Northern Loloish)	Luojishan, Puge, Liangshan, Sichuan	(Munai, 2022), https://zhongguoyuyan.cn/point/60699, accessed on 2 January 2025
Mabian (Northern Loloish)	Minjian, Mabian, Leshan, Sichuan	(Qumu, 2022), https://zhongguoyuyan.cn/point/60M10, accessed on 2 January 2025
Luquan (Northern Loloish)	Sayingpan, Luquan, Kunming, Yunnan	(Pu, 2016, 2022), https://zhongguoyuyan.cn/point/60M38, accessed on 2 January 2025
Weishan (Central Loloish)	Ma’anshan, Weishan, Dali, Yunnan	(T. S. Zhou, 2022), https://zhongguoyuyan.cn/point/60858, accessed on 2 January 2025
Bijie (Northern Loloish)	Datun, Qixingguan, Bijie, Guizhou	(Zhai, 2011, 2022) https://zhongguoyuyan.cn/point/60821, accessed on 2 January 2025
Azha (Southeastern Loloish)	Dongshan, Wenshan, Yunnan	(C. L. Zhou, 2022) https://zhongguoyuyan.cn/point/60714, accessed on 2 January 2025

).

Then each sample language is supported by a word list of about 2500 items, namely ca. 2500 ∗ 10 = 25,000 words altogether, to establish sound correspondence. Among the six Liangshan Yi dialects, word lists for three dialects—Butuo, Jinyang, and Xide—were ready through my previous fieldwork. The remaining three Liangshan Yi dialects, as well as the four Loloish languages from Yunnan and Guizhou, were sourced from The Data Collection, Recording, and Display Platform for the Chinese Language Resources Protection Project (https://zhongguoyuyan.cn/index, accessed on 8 January 2025) and available descriptive grammars and dictionaries. Given that the database provided audio recordings for each word, phonetic transcriptions were verified by listening to the recordings and corrected whenever transcription errors were identified.

4. “Matchmaker” in Liangshan Yi as a Chinese Loanword

This section examines the historical origin of the Liangshan Yi term for “matchmaker” through three lines of evidence: morphological analyzability (Section 4.1), cognate comparison within Sino-Tibetan languages (Section 4.2), and phonological adaptation (Section 4.3). The semantic shift between donor and recipient languages is discussed in Section 4.4. The two lexical forms under discussion are presented in

Table 3. Forms of “matchmaker” and “to marry” in six Liangshan Yi dialects.

	Southeastern Branch				Northwestern Branch
	Butuo	Puge	Huili	Jinyang	Mabian	Xide
‘matchmaker’	hɔ33(44)ka33	hɔ33(44)ka33	hɔ33(44)ka33	fu33ka33	fu33ka33	fu33(44)ka33
‘to marry’	ho33	ho33	ho33	fu33	fu33	fu33

Note. Numbers in parentheses indicate sandhi tones.

.

4.1. The Morphological Analyzability of “Matchmaker” in Liangshan Yi

The disyllabic term hɔ⁴⁴ka³³ (Niesu) or fu⁴⁴ka³³ (Nuosu) is morphologically opaque in Liangshan Yi. The first syllable hɔ³³- or fu³³- is identifiable, not because it is a productive formative in Liangshan Yi, but because another semantically related verb, ho³³ or fu³³ ‘to marry (generally)’, was also borrowed (see the discussion below).

To start with, the first syllable of “matchmaker” hɔ³³- or fu³³- does not participate in any derivational processes to form semantically related words in Liangshan Yi. Similarly, -ka³³ cannot be analyzed as a productive morpheme either. In Liangshan Yi, the productive agentive derivation (‘one who does something’) is realized by the nominalizer -su³³, not by -ka³³, such as nɔ³³su³³ (black-nominalizer) ‘autonym, Nuosu Yi ethnic group’ and mo²¹su³³ (old-nominalizer) ‘old person’. Moreover, attested meanings for native homophones of -ka³³ include the verb ka³³ ‘to hit’ (e.g., ʥi⁴⁴ka³³, mutually + hit, ‘to fight’) and ‘to exhaust’ (e.g., ɣɯ³³ka³³, strength + exhaust, ‘to make great efforts’), causative verb, and nouns (e.g., ‘staple foods’, ‘steel’, ‘firepit’), none of which yields a plausible compositional interpretation for “matchmaker”. These facts indicate that neither constituent has a recognizable morphological role, and the meaning of the whole cannot be derived from its parts under Liangshan Yi morphological rules. Such internal opacity indicates that the term was not a native Liangshan Yi word (see also B. Chen, 1996).

Since “matchmaker” is borrowed as a whole word into Liangshan Yi, the word itself is productive in forming new expressions in the language, such as the verb phrase hɔ⁴⁴ka³³ mu³³ (Niesu) or fu⁴⁴ka³³ mu³³ (Nuosu) (matchmaker + do) ‘to do matchmaking’ and the compound noun hɔ⁴⁴ka³³ dʐɯ³³ (Niesu) or fu⁴⁴ka³³ dʐɯ³³ (Nuosu) (matchmaker + money) ‘the remuneration for a matchmaker’.

The source of “matchmaker” (hɔ⁴⁴ka³³ or fu⁴⁴ka³³) in Liangshan Yi is the Chinese verb 伐柯 (fákē or fa³⁵kʰɤ⁵⁵ in modern Standard Mandarin) ‘to cut wood to make an axe handle; matchmaker, matchmaking’, which is fully morphologically analyzable in Chinese (伐 or fá: ‘to cut’; 柯 or kē: ‘axe handle’). The term first appeared in the poem fákē (伐柯) from the Odes of Bin in Shījīng, see sentence (1), the oldest existing collection of Chinese poetry, comprising 305 works dating from the 11th to 7th century BC.

(1)	伐柯	如何,	匪	斧	不	克
	cut: axe handle	how	not	axe	not	done
	取妻	如何,	匪	媒	不	得
	marry: wife	how	not	matchmaker	not	able
	‘In hewing the wood for an axe-handle, how do you proceed? Without an axe it cannot be done. In taking a wife, how do you proceed? Without a go-between it cannot be done.’

伐柯 (fákē) in Chinese is a verb–object compound, with fá as the action and kē as the patient. See below for the productivity of 伐 fá as a Chinese verb. The second morpheme 柯 kē can also occur productively as a formative of different compound verbs in Chinese (e.g., 执柯 zhíkē ‘to hold the axe handle’). The literal act of cutting wood to make an axe handle first functioned as a metaphor for doing matchmaking, then underwent semantic extension and became lexicalized in Chinese.

As was mentioned above, an additional lexeme meaning ‘to marry (generally)’ exists in Liangshan Yi: ho³³ (Niesu) and fu³³ (Nuosu). Its existence does not indicate any morphological transparency of the word “matchmaker”; rather, it suggests that besides “matchmaker”, another Chinese word 伐 fá ‘to hew, to chop, to matchmake’ was borrowed as a standalone verb ho³³/fu³³ ‘to marry’ in Liangshan Yi (see Section 4.3 for phonological adaptation). It can be observed that following this borrowing, the original Chinese meaning ‘to matchmake’ shifted to denote ‘to marry’ in Liangshan Yi (see Section 4.4). Moreover, this verb is productive in Liangshan Yi and occurs in both derivatives, such as fu³³dʑu⁴⁴ (marry-nmlz, ‘the status of marriage’), and compounds, such as fu³³tɕ^ho⁴⁴ (marry:friend, ‘marital affinity’) and fu⁴⁴p^hu³³ (marry:price, ‘the value of the bride’).

In sum, this morphological analysis, namely transparent analyzability in the donor language but opacity in the recipient language, supports a non-native origin of “matchmaker” in Liangshan Yi.

4.2. The Cognacy of “Matchmaker” in Sino-Tibetan Languages

To verify whether the Liangshan Yi “matchmaker” is a native inheritance or a loanword, the Explicability Principle is adopted. There are two steps of employing the Explicability Principle: to conduct cognate comparison with sister languages (the present section), and to examine the phonological adaptation between the donor language and the recipient language (see Section 4.3).

Cognates are compared according to phylogenetic closeness, namely first within the Loloish languages, then with other Lolo-Burmese languages, and finally with available Tibeto-Burman languages. Under this framework, if neither cognate morphemes nor comparable morphological patterns are attested in the nearer branches, this absence is taken as evidence against inheritance and in favor of borrowing (also see Table 1).

4.2.1. Phonetic Resemblance Without Regular Correspondence

First, *fu³ʔgra² is considered as the reconstructed Proto-Loloish compound for “matchmaker”2, where *fu³ means ‘to send for marriage’ and *ʔgra² ‘between’.

Regarding the first syllable, although hɔ⁴⁴ka³³/fu⁴⁴ka³³ was borrowed from Chinese 伐柯 fákē into Liangshan Yi as a whole and thus should not be analyzable (see analyzability in Section 4.1), a related element ho³³ or fu³³ can be used as a free verb in Liangshan Yi, meaning ‘to marry (generally)’ (i.e., 结婚). As was stated previously in the paper, the analyzability of ho³³/fu³³ does not arise from it being a free morpheme within hɔ⁴⁴ka³³/fu⁴⁴ka³³, but rather because ho³³/fu³³ was also borrowed independently, from Chinese 伐 fá.

Although many similar forms with related meanings, such as Luquan Yi fv̩³³(t^hv̩³³) ‘to marry’, Weishan Yi fu⁵⁵(tsɿ⁵⁵) ‘to marry’, and Azha xʌ³³ ‘to marry a female’, are found in Loloish languages, they are not cognates. Initially, Azha xʌ³³ has a different origin, meaning ‘(man) to marry a female’, similar to Chinese 娶. Sound correspondence regarding the initials can be found among Azha, Luquan, and Bijie; see

Table 4. “(Man) to marry a female” in Loloish languages.

	Luquan	Bijie	Wenshan (Azha)
‘(man) to marry a female’	χə21	khɯ21	xʌ33
‘be correct’	(kɔ33)χo̱55	(dʑiɛ21dʑiɛ21khɯ21)khɯ21	xu33

.

Next, after removing Azha xʌ³³ from the comparison due to its different origin, if *fu³ ‘to send for marriage’ is considered as the Proto-Loloish form, the above-mentioned words, i.e., Nuosu fu³³ ‘to marry’, Luquan Yi fv̩³³(t^hv̩³³) ‘to marry’, Weishan Yi fu⁵⁵(tsɿ⁵⁵) ‘to marry’, should be native and more importantly, sound correspondence should be established. However, this is not the case. Just by examining the correspondence pattern of the initials, let alone the finals and tones, the expected f:f:f correspondence is not found. Instead, the only attested pattern involving an f-sound is v:f:v; see

Table 5. Correspondence of initials v:f:v:ɣ in Loloish languages.

	Nuosu	Luquan	Weishan (Lalo)	Lisu
‘cliff’	va55	fa̱55(ne21)	va31(tsɿ31khɛ31)	ɣɑ55(phi31)
‘left’	vv̩33	(la̱55)fǫ55	(li̱31)vɛ31	(lɛ31)ɣɯ55

. Being attested means that the correspondence is supported by at least two examples, so as to exclude accidental correspondence. If the “pattern” is only found with one morpheme, it may not be qualified to be called a pattern. If we also include Lisu fu³³ ‘to marry off’, which is another suspected cognate, into the comparison, the expected f:f:f:f is not found; instead, v:f:v:ɣ is the only pattern attested, as seen in Table 5. The Lisu data is from ECTBPL (1991).

Therefore, regarding the forms in question, namely Liangshan Yi ho³³/fu³³ ‘to marry’, Luquan Yi fv̩³³(t^hv̩³³) ‘to marry’, Bijie Yi fʊ³³(m³³) ‘to marry’, Weishan Yi fu⁵⁵(tsɿ⁵⁵) ‘to marry’, Azha xʌ³³ ‘to marry a female’, as well as Lisu fu³³ ‘to marry off’, their phonetic resemblance to one another does not suggest that they are cognate, given the lack of regular sound correspondences and the different origin of Azha xʌ³³. J. Wang (1962) observes that old loanwords, borrowed before the divergence of the compared varieties, exhibit sound correspondences, though their phonetic forms may differ significantly. In contrast, words borrowed after the divergence typically show phonetic similarities but lack consistent phonological correspondence. Their phonetic similarity indicates that, apart from Azha, these words likely share the same Chinese origin (i.e., Chinese伐 fá), but their lack of sound correspondence suggests that when these words entered the Loloish languages, the Loloish languages had already split. Therefore, for all related words to be called cognates, they must be able to pass the test of sound correspondence, but not merely based on surface phonetic similarity.

Regarding the second syllable, the reflex of *ʔgra² (also Matisoff’s Proto-Lolo-Burmese: ʔ-gla² ‘between/have a space between/interval’) in Liangshan Yi is ka⁵⁵, not ka³³ in the word “matchmaker”, such as ka⁵⁵tsɨ³³ ‘middle’, ka⁵⁵tsɨ³³ts^ho³³ ‘mediator, witness’ (literally: middle man), na³³ka⁵⁵ ‘forehead’ (literally: the middle areas above the nose), and a³³ka⁵⁵ ‘the second younger brother of the father’ (literally: middle uncle). This reflex differs in tone from the syllable ka³³ in hɔ⁴⁴ka³³/fu⁴⁴ka³³ (depending on dialect).

Since Weishan Yi (dzɿ³¹mu³¹) and Azha (mo³³lɛ⁵⁵nɔ⁴⁴kə³³po³³) use different etyma for “matchmaker” and thus cannot provide comparable segments, if ka³³ in fu⁴⁴ka³³ (Nuosu)/hɔ³³⁽⁴⁴⁾ka³³ (Niesu), and tɕo⁵⁵ in fu³³tɕo⁵⁵ (Luquan) are true reflexes of Proto-Loloish *ʔgra², the correspondence of the initials (i.e., k:tɕ) should be able to be verifiable across a broader range of Loloish languages or dialects, like the above-mentioned pattern v:f:v being able to be extended to a genuine Loloish pattern (v:f:v:ɣ). Nevertheless, when a Central Loloish language with a comparable segment—namely Limi (tsu⁵⁵ in fu³³tsu⁵⁵p^ho²¹ ‘matchmaker’, p^ho²¹: ‘man’)—is considered, the correspondence k:tɕ:ts is not found. Instead, the likely Loloish pattern is k:tɕ:tʂ, as supported by the following two morphemes in

Table 6. Correspondence of initials k:tɕ:tʂ in Loloish languages.

	Nuosu	Luquan	Limi
‘colander’	(ba21)ka33	(ʂʅ̱55)tɕo33	(lu21)tʂu33(khu21)
‘plum’	(sɿ21)ka33	(sɔ21)tɕo33	(sɛ21)tʂu55(sɛ21)

.

Therefore, this suggests that ka³³ in fu⁴⁴ka³³ (Nuosu)/hɔ³³⁽⁴⁴⁾ka³³ (Niesu), and tɕo⁵⁵ in fu³³tɕo⁵⁵ (Luquan) are not reflexes of Proto-Loloish *ʔgra². The impression that Liangshan Yi ka³³ (as in fu⁴⁴ka³³) and Luquan tɕo⁵⁵ (as in fu³³tɕo⁵⁵) are cognate is probably due to their phonetic similarity to the Northern Loloish words meaning “middle”, such as Nuosu ka⁵⁵ ‘middle’ vs. ka³³; in many Northern Loloish languages, the two words are coincidentally homophones, such as Luquan Nasu fu³³tɕo⁵⁵ ‘matchmaker’ vs. tɕo⁵⁵ ‘middle’, and Bijie Nyisu fʊ³³tɕu⁵⁵ ‘matchmaker’ vs. tɕu⁵⁵ ‘middle’.

4.2.2. Different Etyma in Other Loloish and Tibeto-Burman Languages

Lisu, a Central Loloish (or Ngwi) language (Bradley, 1997), uses dzi²¹mu²¹ (also dzi²¹mu³³) for ‘matchmaker, middleman’ (Bradley, 1994, also see STEDT), which is transparently analyzable: dzi²¹ means ‘to mediate, to matchmake’, while mu²¹⁽³³⁾ is the augmentative morpheme. The morphological makeup differs from Liangshan Yi hɔ⁴⁴ka³³/fu⁴⁴ka³³. Moreover, neither of the Lisu morphemes in “matchmaker” is cognate with those in Liangshan Yi. Instead, the Weishan Yi or Lalo, another Central Loloish language, uses dzɿ³¹mu³¹ (matchmake:augmentative) for ‘matchmaker’, reflecting a two-syllable etymon with clear cognates in Lisu.

Lahu (another Central Loloish language) and Hani (Southern Loloish) also use etymologically distinct words for “matchmaker/matchmaking”. Examples include tsɿ³³ka³¹pa³¹ in Lahu (Nanduan), and ʑɛ⁵⁵tɕ^ho³¹ in Hani (Baihong). Etymologically, none of these forms can be linked to Chinese fákē.

Luquan Yi and Bijie Yi, two Northern Loloish languages, exhibit forms that phonetically resemble the Liangshan Yi word “matchmaker”, fu³³tɕo⁵⁵ and fʊ³³tɕu⁵⁵, respectively (see

Table 7. Loloish forms for “matchmaker” in the sample varieties.

	Butuo	Xide	Luquan	Bijie	Weishan	Azha
‘matchmaker’	hɔ44ka33	fu44ka33	fu33tɕo55	fʊ33tɕu55	dzɿ31mu31	mo33lɛ55nɔ44kə33po33

), where the initial syllable fu³³/fʊ³³ resembles Nuosu fu- in fu⁴⁴ka³³, and tɕo⁵⁵/tɕu⁵⁵ is likely the palatalized form of -ka³³. Despite the phonetic similarity, systematic sound correspondence cannot be established among Liangshan Yi, Luquan Yi and Bijie Yi (see Section 4.2.1).

However, the sound correspondences, regarding not just initials, but also finals and tones, among Liangshan Yi dialects can be well established, showing their close phylogenetic relationship (see

Table 8. Sound correspondence of Liangshan Yi dialects.

	Sound Correspondence	Butuo	Puge	Huili	Mabian	Jinyang	Xide
媒人 ‘matchmaker’		hɔ33/44ka33	hɔ33/44ka33	hɔ33/44ka33	fu33ka33	fu33ka33	fu33/44ka33
Syllable 1 日出 ‘sunrise’	final, tone	dʐɨ44dɔ33	dzɨ44dɔ33	dzɨ44dɔ33	bu33du33	dʐɨ33du33	gɯ44du33
Syllable 1 八 ‘eight’	initial	hi55	hɛ55	hɛ55	hɛ35	hi55	hi55
Syllable 2 火塘 ‘firepit’	initial, final, tone	ka33khu33	ka33khu33	ka33khu33	ka33bɛ35	ka33khu33	ka33khu33
结婚 ‘to marry’		ho33	ho33	ho33	fu33	fu33	fu33
八 ‘eight’	initial	hi55	hɛ55	hɛ55	hɛ35	hi55	hi55
年 ‘year, age’	final	kho55	kho55	kho55	khu33	khu33	khu33
鱼 ‘fish’	tone	ŋɯ33	hɯ33	ŋɯ33	hɯ33	hɯ33	hɯ33

). Note that in Liangshan Yi, h- changes to f- if followed by u or u, namely f > h/_ u or u, which explains the correspondence in Mabian, Jinyang and Xide (also see Section 4.3).

Azha, a Southeastern Loloish language (Pelkey, 2011a, 2011b), presents an even more divergent form (mo³³lɛ⁵⁵nɔ⁴⁴kə³³po³³), whose complex structure and etymology do not align with either the Central Loloish forms or the Northern Loloish forms.

Taken together, the Loloish forms for “matchmaker” show diverse origins and morphological patterns, and no coherent cognate set can be identified or reconstructed. Therefore, the hypothesis of “matchmaker” as a native word of Loloish languages is not supported.

Next, we extend the search for cognates beyond Loloish to the more distant branches of Tibeto-Burman (

Table 9. Comparison of “matchmaker” with other Tibeto-Burman languages.

	Lolo-Burmese		Tibetic		Qiangic	Western Tani		Tujia
	Burmese	Achang	Amdo Tibetan (Xiahe)	Lhasa Tibetan	Qiang (Puxi)	Bengni	Bokar	Tujia
‘matchmaker’	aʊ̀̃ θwè	mɑŋ31	war wa	phar11ma55	bʐi-pɑ-mə	ñi-mɯ	gin-to	thu21kha21
morphological pattern	matchmaker + connect		middle + person	middle + person	matchmaker + act + person	not segmentable in source		? + person

The references of the data are: Burmese (Myanmar Language Commission, 1993)3, Achang (Shi, 2009), Amdo Tibetan and Lhasa Tibetan (http://www.kaom.net/, accessed on 24 November 2025), Qiang (Puxi) (C. L. Huang, 2004), Bengni and Bokar (Sun, 1993), Tujia (K. Chen, 2006).

).

In Burmese, which belongs to the Lolo-Burmese branch—a higher-level node than Loloish—the term for ‘matchmaker’ is aʊ̀̃ θwè. This term is a transparent compound formed by native Burmese morphemes, with the first syllable (aʊ̀̃) meaning ‘go-between’ or ‘matchmaker,’ and θwè meaning ‘connect’ or ‘join’. In Lianghe Achang, another Burmish language, ‘matchmaker’ is referred to as mɑŋ³¹ (Shi, 2009). Neither of the Burmish forms show any connection with the Loloish forms.

When compared with other Tibeto-Burman languages beyond Lolo-Burmese, none of them exhibit any signs of cognacy or morphological similarity to hɔ⁴⁴ka³³ or fu⁴⁴ka³³ in Liangshan Yi (as summarized in Table 9).

Furthermore, no possible cognates are found in other Tibeto-Burman languages regarding ‘to marry’ (see

Table 10. Comparison of “to marry” with other Tibeto-Burman languages.

	Northern Loloish		Central Loloish		Southeastern Loloish	Lolo-Burmese		Tibetic		Qiangic	Western Tani		Tujia
	Luquan	Bijie	Weishan	Lisu	Azha	Burmese	Achang (Longchuan)	Amdo Tibetan (Xiahe)	Lhasa Tibetan	Qiang (Puxi)	Bengni	Bokar	Tujia
‘to marry’	fv̩33thv̩33	fʊ33m33	ʔmɪ̱21ʐo33 ʔlo21	fu33za21	i44tshu31xʌ33	lɛʔ4thɑʔ4	tsaŋ31i31ʂa55	za wa jə	tɕhaŋ55sa55ca132	tɕhe su	laː-mu-su	me taŋ moː	tso55ɣɑ21

The references of the data are: Luquan (Pu, 2022), Weishan (T. S. Zhou, 2022), Bijie (Zhai, 2022), Burmese, Achang (Longchuan), Amdo Tibetan, Lhasa Tibetan and Tujia (B. Huang, 1992); Lisu (Bradley, 1994); Qiang (Puxi) (C. L. Huang, 2004); Bengni and Bokar (Sun, 1993).

). They also display different morphological patterns from the Liangshan Yi forms. For example, in Tujia, tso⁵⁵ɣɑ²¹ consists of two morphemes meaning ‘relative’ and ‘to marry’, respectively (K. Chen, 2006); similarly, in Amdo Tibetan (Xiahe), the expression literally means ‘to be a couple’: za wa ‘couple’ and jə ‘do’ (B. Huang, 1992).

Therefore, the additional Tibeto-Burman data do not support any cognate relation with “matchmaker” in Liangshan Yi, as well as the verb “to marry (generally)”, suggesting that the Liangshan Yi terms are loanwords, rather than inherited Tibeto-Burman etyma.

4.3. The Phonological Adaptation Between the Donor and Recipient Languages

The second step of the Explicable Principle is to examine the phonological adaptation between the donor language and the recipient language.

The analysis begins with the first syllable 伐 (fá). The initials of hɔ³³- (Niesu) and fu³³ (Nuosu) provide information on the process of labiodentalization from Middle Chinese to Early Mandarin. By the time when the borrowing occurred, the initial of the Chinese form can be reconstructed as *fɦ-, with reference to the Late Middle Chinese 伐 (LMC: *fɦ-, in contrast to *b- of Early Middle Chinese) in Pulleyblank (1983). This reconstruction suggests that the borrowing took place close to the completion of labiodentalization.

Moreover, the initial is reconstructed as *fɦ- with voiced aspiration to reflect the distinction between h- and x- in modern Liangshan Yi. Specifically, h- corresponds to voiced Chinese fricatives, while x- aligns with voiceless ones at the time of borrowing. For instance, the MC form *hwɰa for 華 ‘China’, with its voiceless initial, is borrowed into Liangshan Yi as xɔ³³ ‘Han Chinese’. In contrast, the MC form *ɦ^ʷɯa for 滑 ‘slippery’, with its voiced initial, is borrowed as hɔ³³ ‘slippery’. The Middle Chinese forms are drawn from W. Pan (2000).

Labiodentalization is a major phonological change in Chinese, which began during the Tang Dynasty, ca. 8th century (C. B. Huang, 2010), when bilabial consonants (p-, pʰ-, b-, m-) started changing to labiodental fricatives (f-, v-, ɱ-) in Middle Chinese before the palatal glide -j-. Early evidence appears in Huì Lín’s (736–820) Yīqiè Jīng Yīnyì, a collection of glosses of the Buddhist canon. By the Northern Song Dynasty (960 to 1127 CE), labiodentalization was more pronounced and largely completed, at least in many Mandarin-like Sinitic varieties, as evidenced in the Shēngyīn Chànghè Tú (between 1011 and 1077 CE), which categorizes labiodental fricatives as distinct initials in Middle Chinese (Shen, 2020, p. 173). By the time of Zhōngyuán Yīnyùn (1324 CE), labiodentalization was completed (see Shen, 2011).

Regarding the process of vowel adaptation, the Middle Chinese final is reconstructed as *-uɑC, featuring a back vowel as the nucleus and a stop coda, with the full proto-form reconstructed as *fɦuɑC. Since we have not yet obtained systematic evidence to determine the exact value of the coda of the donor Chinese, C is used to represent the final stop.

Consequently, the Liangshan Yi dialects initially employed *h- to correspond with the *fɦ- onset of *fɦuɑC, resulting in an intermediate form *huɑC. Since no Loloish languages keep any stop coda, including the reconstructed Proto-Lalo (Yang, 2010), our reconstruction suggests that Proto-Liangshan Yi used *o to adapt the Chinese final *-uɑC. This inference is based on differing correspondence patterns between Nuosu and Niesu: Nuosu u corresponds to Niesu o in some cases, and to Niesu ɔ in others. These two patterns reflect two distinct Proto-Liangshan Yi finals, which are reconstructed as *o (for Nuosu u ≡ Niesu o) and *ɔ (for Nuosu u ≡ Niesu ɔ). The two finals have merged in Nuosu, whereas in Niesu they remain distinct, though decayed, as illustrated by the words ‘to patch’ and ‘be full’: Nuosu du⁵⁵ and mbu³³, respectively, vs. Niesu do⁵⁵ and mbɔ³³.

Therefore, *ho⁴ is the reconstructed localized form4. The assignment of Tone 4 to *ho⁴ is supported by parallel examples such as (Proto-Loloish *s-nök^H >) PLS *nɔ⁴ ‘bean’ > nu³³ (Nuosu) and nɔ³³ (Niesu), and (Proto-Loloish *(sə)-grok^H >) PLS *kɔ⁴ ‘to frighten’ > ku³³ (Nuosu) and kɔ³³ (Niesu). Proto-Liangshan Yi Tone 4 represents an inheritance of Proto-Loloish checked Tone H. The Proto-Loloish forms are taken from Bradley (1979).

Subsequently, in the Northwestern branch of Liangshan Yi, the vowel underwent raising5, transforming the syllable into an intermediate form *hu⁴. However, in Liangshan Yi, h- changes to f- if followed by u or u, which eventually results in fu³³ in the Northwestern branch. Moreover, the tense phonation was lost in Southeastern branch as ho³³. The phonological adaptation of the Chinese word 伐 fá is summarized below in Figure 3.

Moreover, besides the adaptation shown in Figure 3, the phonological adaptation of 伐柯 fákē involves two additional processes. First, vowel lowering occurs under the influence of an adjacent non-high vowel, specifically the low vowel (i.e., a) in the second syllable. This changes ho³³ka³³ (the underlying form in the Southeastern branch dialects) into hɔ⁴⁴ka³³ (the surface form). This phenomenon is commonly observed within both the Southeastern branch and the Northwestern branch, such as mu³³ ‘horse’ vs mu³³ve³³ ‘steed’, mu³³ɣa⁵⁵ ‘horse saddle’, and ŋɡɯ³³ ‘buckwheat’ vs ŋɡa³³va⁵⁵ ‘buckwheat pie’ (see more examples in S. Chen et al., 1985). Since the vowel in fu³³ is already the lower counterpart of u, no such change applies to fu⁴⁴- in fu⁴⁴ka³³ in the Northwestern branch dialects.

Secondly, in Liangshan Yi, tone 33 will change to 44 when followed by another 33 tone, namely 33 > 44/_33; other examples include su⁴⁴ʐɨ³³ (Niesu) and su⁴⁴ʑi³³ (Nuosu) ‘elder, influential figure’ and bu⁴⁴dzɯ³³ (Niesu) and bo⁴⁴dzɯ³³ (Nuosu) ‘hero, brave man’. It should be noted that the sandhi rule 33 > 44/_33 is more pervasive in the Northwestern branch; in the Southeastern branch, the tone sandhi is more frequently observed during isolated word reading and occurs less commonly in naturalistic speech. However, according to my consultants from Butuo, a dialect of the Southeastern branch, the pronunciation of the words remains natural and acceptable in everyday communication regardless of whether the sandhi is applied.

Moreover, the second syllable 柯 kē is consistent across the Liangshan Yi dialects, preserving the low vowel as in the Middle Chinese form *kɑ, but adapted as PLS *ka¹. By the Yuan Dynasty, Middle Chinese *kɑ had undergone vowel raising to *ko, as evidenced in Zhōngyuán Yīnyùn (1324 CE) (Shen, 2020). This suggests that the second syllable was borrowed during the same period as the first syllable, entering the language together as a word (see borrowability of the linguistic unit in B. Chen, 1996).

Overall, the surface form is derived as shown below in Figure 4:

4.4. The Semantic Adaptation Between the Donor and Recipient Languages

While the semantic match of “matchmaker” between Chinese 伐柯 fákē and Liangshan Yi hɔ⁴⁴ka³³/fu⁴⁴ka³³ is straightforward, a semantic skew exists between Chinese 伐 fá ‘to matchmake’ and Liangshan Yi ho³³/fu³³ ‘to marry (generally)’, as well as ‘to marry off a female’ in other Loloish languages (highlighted cells in

Table 11. “Matchmaker”, “to marry (generally)”, “to marry a female off”, and “(man) to marry a wife” in Loloish languages.

Datapoint	Loloish Branch	Matchmaker	Marry (Generally)	Marry a Female Off [女子] 出嫁	Marry a Wife [男子] 娶妻
Limi	Central	fu33tsu55pho21	tɕi33fu33	fu33do33	ma21ȵe33tsi̱33
Lalo (Weishan)		dzɿ31mu31	fu55tsɿ55	phɯ31thi31	mɑ31ny55tsɿ55
Nanhua Yi		fu33tɕo55	ʔɤ̱31dɤ̱33	zo31mɛ̱31fu33	mɛ31zo31xo33
Lisu		dzi21mu21	fu33za21	fu33	khu33
Sadu	Northern	fu33tɕo33mo21	tɕhi55mo21kɯ33	fu33	tɕhi55mo21kɯ33
Liangshan Yi (Niesu)		hɔ44ka33	ho33	dʑɯ21	ʑo33
Liangshan Yi (Nuosu)		fu44ka33	fu33	dʑi21	ʑu33
Luquan Lowu		fu33tɕo55	kɛ33do33	zo31mɛɹ31fu33	tɕhi31mu33tsɿ33
Luquan Nasu		fu33tɕo55	dʑǫ21do̱21	ne55fu33	tɕhi55χə21
Bijie Yi		fʊ33tɕu55	fʊ33m33	nɤ33tɕhi55	tɕhi55khɯ21
Dafang Yi		fɔ33tɕu55	fɔ33m33	ne33fɔ33	tɕhi55khɯ21
Wusa		fu13tɕo55	tɕhi55kɯ21	ʔɑ21me13tɕhi55	zu33tɕhi55khɯ21
Sani	Southeastern	fu33kɑ55	tʂʅ33tʂʅ55	fu33	ma21tʂhʅ33
Azha		mo33lɛ55nɔ44kə33po33	i44tshu31xʌ33	xɛ44lɑ55	zɔ33mɑ31xʌ33
Akha	Southern	mɛ51sɯ31	ø̱33m33ba55	su33mji55i55	(mji31ja31)sɤ31
Hani (Baihong)		ʑɛ55tɕho31	xa31mi31ɕɤ31	m55tsa31tsɛ55	xa31mi31ɕɤ31
Hani (Lüchun)		je55mo31	mi31sɿ̱33za31sɿ̱31sɤ31	su55mi55i55	mi31za31sɤ31

).

This semantic link is built upon the functional meaning of “to matchmake”, namely, “to arrange the marriage for someone”. The semantic developments are thus proposed as: (1) “to matchmake, or to arrange the marriage for someone” → “to marry”; and (2) “to matchmake, or to arrange the marriage for someone” → “to marry off a female”. Whether the two development paths can be merged into one is uncertain, due to the lack of sequential evidence about the meanings of “to marry” and “to marry off a female”.

The semantic evolution of the English verb marry provides a clear case of how a term can shift from denoting the actions of a marriage-arranging third party to describing the union of the couple themselves, namely “to matchmake, or to arrange the marriage for someone” → “to marry”. When borrowed from Old French marier in the early 14th century, the Middle English verb marīen primarily carried the sense of a parent or superior giving a subordinate in marriage. As documented in the Middle English Compendium, a publication of the University of Michigan (https://quod.lib.umich.edu/m/middle-english-dictionary/dictionary, accessed on 8 April 2026), one sense of marīen is explicitly defined as “of parents or superiors: to give (offspring, subordinates) in marriage; give (sb.) in marriage (to sb.); cause (sb.) to unite (with another) in marriage”. A representative quotation from c. 1325 illustrates this agentive, third-party perspective: Ich þe wole marie wel, mid þe þriddedel of mi londe, To þe nobloste bacheler… (“I will marry you well, with a third of my land, to the noblest bachelor…”), spoken by a father or lord arranging a match for his daughter. In modern English, the verb marry can function as an intransitive verb meaning “to take a spouse, to wed”, such as He first married at forty. See “marry (generally)” in Table 11.

Next, once a verb meaning “of parents or superiors: to give (offspring, subordinates) in marriage; give (sb.) in marriage (to sb.); cause (sb.) to unite (with another) in marriage” exists, sociocultural factors (e.g., patrilocal residence, patrilineal descent) will bias its use toward female objects. For example, based on my fieldwork observations, child marriage, which was still prevalent in eastern Liangshan in the 1990s, predominantly affected females. The biased use of the verbs in Central, Northern and Southeastern Loloish languages to denote the female-specific meaning, namely ‘to marry off a female’, is highly likely due to the patrilineal society of the Loloish people (see “marry a female off” in Table 11).

It is noted that all instances of borrowing involving Chinese terms 伐柯 (fákē) and 伐 (fá), highlighted in Table 11, fill the semantic gaps (i.e., “matchmaker”, “to marry (generally), “to marry a female off”) in the Loloish languages, leaving the native concept (i.e., “(man) to marry a wife” or 娶) unaffected.

Based on Koch (2012), the shift from “to arrange the marriage for someone” to “to marry” and to “to marry off a female” can be explained as a contiguous change, specifically a metonymic figure–ground effect within the same conceptual frame of MARRIAGE. In the donor language, the verb highlights a specific, contiguous element of the marriage frame: the arranging/match-making step. Since this step is contiguously related to the overall marriage outcome via a cause–effect or part-whole type of contiguity, speakers of the recipient languages can perform a figure–ground shift. Instead of highlighting the arranging element, the same lexical item is used to highlight the contiguous central outcome of that frame, i.e., the act of marriage. This results in a metonymic change from “to arrange the marriage for someone” to “to marry”. The semantic shift to “marry off a female” is another similar metonymic change. Within the marriage frame, a salient participant role often involves a female as the most affected figure. Through a figure–ground shift, the word becomes conventionally interpreted as foregrounding the female’s placement in the marriage. See similar theories about the motivations of semantic changes in Bloomfield (1933), Ullmann (1962) and Blank (1999).

Therefore, the semantic shift of the Loloish lexical forms from the source meaning (Chinese 伐 fá ‘to matchmake, to arrange the marriage for someone’) to ‘marry’ and to ‘marry off a female’ is both empirically attested (e.g., with English marry) and theoretically well-motivated.

5. Borrowing Time of “Matchmaker”

The term fákē (伐柯 ‘matchmaker, matchmaking’) and the related morpheme fá (伐 ‘to hew, to attack, to matchmake’) originated in Shījīng (or Book of Odes, 11th–7th century BCE). However, phonological adaptation suggests the two terms were borrowed into the Liangshan Yi language after the Chinese labials changed to labiodentals but before the completion of labiodentalization, i.e., around the time of Late Middle Chinese (ca. late Tang to Song period, or 9th to 13th century; see Baxter, 1992). The lower limit can be set as 1324 CE when Zhōngyuán Yīnyùn was written and documented the completion of labiodentalization.

To reconcile this mismatch between the term’s ancient literary origin and its borrowing timeframe, we examine periodized frequencies of fákē and fá using the large-scale annotated corpus of classical Chinese (http://ccl.pku.edu.cn:8080/ccl_corpus/index.jsp, accessed on 13 November 2025), also known as CCL corpus, developed by Center for Chinese Linguistics, Peking University (Zhan et al., 2003).

We aggregated the data into four primary periods, following established historical linguistic periodization (L. Wang, 2004; Pulleyblank, 1983, 1991; Baxter, 1992; Shen, 2020): Old Chinese (Zhou–Han, ca. 11th century BCE–220 CE; the classical formative era), Early Middle Chinese (Six Dynasties–Early Tang, 220–ca. 8th century, anchored by the Qièyùn system), Late Middle Chinese (Late-Tang–Song, ca. 9th century-1279 CE), and Early Mandarin (late Yuan–Qing, 1324–1911 CE; marked by vernacular resurgence following the ca. 1324 CE phonological benchmark).

The analysis begins with a periodized CCL query for the string 伐柯 fákē. For each period t, two frequency counts are extracted from the search results: $N_t\left(\mathrm{t}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}\right)$ meaning total number of retrieved occurrences of 伐柯 fákē (all senses), and $N_t$ (matchmaking) meaning the number of occurrences of “matchmaker, matchmaking”. The measure is the Matchmaking Share for period t, defined as:

$$MatchmakingShar{e}_t={\displaystyle \frac{N_{t\left(matchmaking\right)}}{N_{t\left(total\right)}}}$$

An occurrence is coded as matchmaking sense if and only if it satisfies both conditions (a) and (b) below:

(a) Marriage-context condition: The concordance window contains at least one of the marriage/mediation cues, e.g., 媒 ‘to matchmake’, 婚 ‘to marry (general)’, 嫁 ‘to marry a man’, 娶 ‘to marry a woman’, 妻 ‘wife’, 夫 ‘husband’, 亲/姻 ‘marriage, in-laws’, 婿 ‘son-in-law’, 男 ‘male’, 女 ‘female’, etc.

(b) Exclusion condition: The token should not be the use of the poem’s title, e.g., “《伐柯》曰…” meaning ‘The poem Cutting the Axe Handle says…’, “诗云伐柯…”meaning ‘the Shijing says “Cutting the Axe Handle” …’, and the context should not be the use of its literal sense, namely containing woodworking cues such as 斧 ‘axe’, 斤’ adze’, 斫 ‘to chop’, 薪 ‘firewood’, 木 ‘wood’, 材 ‘timber/material’, 削 ‘to shave/carve’, etc.

The results are summarized in

Table 12. Distribution of fákē and its matchmaking sense across different linguistic periods.

Linguistic Period	Dynasties Included	Corpus Size by Periods (Million)	${\mathit{N}}_{\mathit{t}}\left(\mathbf{Total}\right)$	${\mathit{N}}_{\mathit{t}}$ (Matchmaking)	Matchmaking Share	Normalized Frequency ${\mathit{N}}_{\mathit{t}}$ (Matchmaking/Million)
Old Chinese (OC)	Zhou–Han	7.81	30	2	6.67%	0.26
Early Middle Chinese (EMC)	Six Dynasties–Early Tang	9.45	70	2	2.86%	0.21
Late Middle Chinese (LMC)	Late-Tang–Song	74.32	562	25	4.45%	0.34
Early Mandarin (EM)	Yuan	8.48	5	0	5.16%	0.20
	Ming	29.21	235	6
	Qing	68.87	167	15

and visualized in Figure 5, which plots the matchmaking share of fákē across different linguistic periods.

The data reveals a U-shaped pattern regarding using the matchmaking sense of fákē. Although it emerged in the Old Chinese period with a normalized frequency of 0.26 and 6.67% of matchmaking share, the use of fákē declined significantly throughout the Middle Chinese era, dropping to 2.86% in Early Middle Chinese (see Figure 5). Because of the low frequency and exposure, it was unlikely for fákē to reach the communicative currency for interlingual borrowing.

However, from Late Middle Chinese, the matchmaking sense of fákē displays a significant rise, with 4.45% of matchmaking share, and continues to grow in Early Mandarin (i.e., 5.16%) (see Figure 5). It should be noted that the highest increase during the Late Middle Chinese period is primarily contributed by the increasing use of fákē as matchmaking in the Song Dynasty, namely 92% (23 out of 25 attestations) and the higher exposure made the borrowing of the matchmaking sense possible. See example (2) of using fákē as free words in the Song Dynasty.

(2)	不	疑	何	卜,	自	諧	鳴	鳳	之	和;
	not	doubt	how	divine,	self	harmonize	sing	phoenix	gen	harmony;
	‘Without doubt, what need is there for divination (of our fates)? We shall naturally harmonize in the harmony of singing phoenixes;’
	無	言	弗	酬,	願	守	伐柯		之	則。
	no	word	not	reply,	wish	keep	matchmaker		gen	rule.
	‘with no word left unanswered; I vow to abide by the rule of the matchmaker.’ (Liao Xingzhi, Letter in Response to the Liu Family’s Marriage Proposal, Shěngzhāi jí, Song Dynasty, 12th century)

The term fákē, used as formative in compounding—such as fákē-rén (matchmaking:person) ‘matchmaker’—is also attested. See sentence (3).

(3)	其	伐柯人		两	家	通报,		择	日	过	帖
	3sg.poss	matchmaking:person		two	family	inform		select	day	deliver	post
	‘Once the two families, through the matchmaker, communicate and exchange proposal cards on an auspicious day,
	各	以	色彩		衬	盘,	安	定帖		送	过
	each	with	color		decorate	tray	place	engagement letter		send	over
	‘and each side sends decorated trays and engagement letters,’
	方	为	定	论。
	then	become	final	decision
	‘then the marriage matter is considered settled.’ (Wu Zimu, Jiaqu, Mengliang Lu, Song Dynasty, c.1274)

The timing of the texts thus aligns with the phonological evidence situating the borrowing around Late Middle Chinese and no later than 1324 CE, the compilation year of Zhōngyuán Yīnyùn. Borrowing therefore was triggered not by fákē’s ancient origin from Shījīng, but by its subsequent rise in later dynasties. Although the usage frequency of fákē for ‘matchmaker’ is high in the Ming and Qing dynasties, our phonological evidence does not support Chinese borrowing later than Early Mandarin, namely after 1324 CE. Instead, the high frequency is the result of lexical archaization during the Ming and Qing dynasties, a social phenomenon of reviving the elegance of classic Chinese words and stratifying literary and colloquial registers (F. Wang, 2020).

The timeframe for fákē can be further corroborated through a parallel investigation of the morpheme fá (伐). Given the term’s high polysemy (‘to cut’, ’to attack’ and other meanings), we employed bidirectional proximity queries pairing fá with semantic markers including ‘matchmaker’ (媒), ‘marriage’ (婚, 姻), ‘to marry’ (嫁, 娶), ‘spouse’ (妻, 夫, 婿), ‘kinship’ (亲), and ‘gender’ (男, 女). The queries utilized the syntax X$100Y and Y$100X (e.g., 伐$100婚 and 婚$100伐), targeting co-occurrences within a maximum interval of 100 characters regardless of word order. The 100-character window was selected as an exhaustive range to ensure maximum recall, accommodating varied syntactic structures where fá and its contextual markers (e.g., ‘matchmaker’ or ‘marriage’) are not immediately adjacent. To ensure statistical integrity, all overlapping results from the queries were subsequently manually screened to confirm that the sense of each token was matchmaking-related. See

Table 13. Diachronic distribution of the matchmaking-related sense of the morpheme fá (伐).

Linguistic Period	Dynasties Included	Corpus Size by Periods (Million)	${\mathit{N}}_{\mathit{t}}$ (Total)	${\mathit{N}}_{\mathit{t}}$ (Matchmaking Sense)	Matchmaking Share	Normalized Frequency ${\mathit{N}}_{\mathit{t}}$ (Marriage Sense/Million)
Old Chinese (OC)	Zhou–Han	7.81	6191	2	0.03%	0.26
Early Middle Chinese (EMC)	Six Dynasties–Early Tang	9.45	13,579	2	0.01%	0.21
Late Middle Chinese (LMC)	Late-Tang–Song	74.32	65,488	38	0.06%	0.51
Early Mandarin (EM)	Yuan	8.48	361	7	1.94%	17.01
	Ming	29.21	35,414	35	0.10%
	Qing	68.87	35,603	1771	4.97%

.

As illustrated in Table 13 and Figure 6, similar to fákē, the morpheme fá (伐) meaning ‘to matchmake’ appeared in the Old Chinese period; but the frequency and exposure remained low during the Early Middle Chinese period, with 0.01% of matchmaking share and a normalized frequency of 0.21 (see Figure 6).

From Late Middle Chinese onward, the matchmaking sense of fá exhibits a clear upward trend, reaching 0.06%. It should be noted that this increase is similarly overwhelmingly driven by its usage in the Song Dynasty, which accounts for 94.74% of Late Middle Chinese attestations (36 out of 38), indicating a marked rise in frequency and exposure to support borrowing. The trend continues into Early Mandarin, where the matchmaking sense of fá shows a substantial expansion, reaching 2.54% of the share (see Figure 6). Although the frequency of this usage peaks in the Qing Dynasty, this later prominence should not be interpreted as evidence for a later-than-EM borrowing event, as discussed above. See example (4) and (5) of using fá as free morpheme in word formation in Song Dynasty.

(4)	適	見	其	婢	自	外	來,
	happen to	see	3sg.poss	maid	from	outside	come
	云	與	小	娘子	作伐。
	say	for	young	lady	do:matchmake
	‘(We) happen to see those maids come in from outside, who said they would act as a matchmaker for the young lady.’ (Guo Tuan, Kuiche Zhi, Song Dynasty, mid-12th century)

(5)	逵	为	其	子	叶寮	执伐,	娶	赵氏。
	Name	for	3sg.poss	son	Name	execute:matchmake	marry	Ms Zhao
	‘Kui acted as matchmaker for his son Yeliao to marry a young lady of the Zhao clan.’ (Chen Hu, Qi Jiu Xu Wen, Song Dynasty, mid-12th century)

Once again, this timeline supports the timing of the borrowing based on our phonological evidence. The continuing high percentage of using fá (伐) meaning ‘to matchmake’ in Ming and Qing dynasties is similarly due to the above-mentioned process of lexical archaization because our phonological evidence does not support borrowing later than Early Mandarin.

In sum, the quantitative evidence of using fákē 伐柯 and fá 伐 in the CCL corpus resolves the timing mismatch concerning the origins of the Chinese words and their phonological adaptation into Liangshan Yi. The textual resurgence of the two Chinese words by Late Middle Chinese and before Early Mandarin not only corroborates the phonological findings, confirming that the borrowing took place around the Late Middle Chinese period and no later than 1324 CE, but also suggests a narrower timeframe for the borrowing: namely, from the mid-10th century (since Song Dynasty) to the early 14th century.

6. Matchmaker Beyond Liangshan Yi Societies

As was discussed above, many Loloish languages borrowed Chinese 伐柯 fákē and/or 伐 fá (see Section 4.2). Table 11 in Section 4.4 displays that the borrowing of fákē, as well as fá, occurs primarily in Northern Loloish varieties6, with a few instances in Central and Southeastern Loloish varieties, and no occurrences in Southern Loloish. See the distribution of the borrowing in Figure 7.

This diffusion is historically rooted. Historical records indicate that Loloish-speaking elites were part of the ruling class in both the Cuan Kingdom (225–602 CE)7 of eastern Yunnan and the Nanzhao Kingdom (738–902 CE) of western Yunnan. The Northern Loloish languages are thought to derive from the Cuan (爨) Kingdom and its dispersal. For Central Loloish, it can be linked with the Nanzhao Kingdom and its expansion (Bradley, 2025). These rulers were heavily Sinicized and had closer relation with Chinese people than many other Tibeto-Burman speakers, and the close contact started as early as Jin-Northern and Southern dynasties-Sui, ca. 339 to 618 CE (Bradley, 2025).

Originally Han Chinese military settlers, the Cuan family assisted Zhuge Liang’s Southern Campaign, also known as the War of Pacification in the Nanzhong region, which took place in 225 CE (Fang, 1982). They then consolidated control over Nanzhong, encompassing modern Yunnan, southern Sichuan, and western Guizhou, around 339 CE and ruled for nearly four centuries. As the Loloish population, along with other smaller non-Chinese tribes, was dominant in the Nanzhong region, they were impacted by the warfare. However, due to their demographic predominance, through intermarriage and cultural assimilation, the Cuan people gradually adopted non-Chinese customs and languages, becoming what historical texts describe as “Cuan Barbarians”. At the same time, there were also the Loloish people embracing Cuan customs. Therefore, during the period from the Northern and Southern Dynasties to the Early Tang Dynasty (approximately 420–618 CE), the ethnic Yi people were also referred to as “Cuan people”. Moreover, Nanzhao’s aristocracy modeled its government on Tang China, creating Chinese-style ministries and civil-service exams, and explicitly invoked Confucian ideology (Daniels, 2021). Genetically, the ethnic Yi, along with the Qiang and Bai peoples, share a closer connection with the Han than do other Tibeto-Burman groups. This is evidenced by genetic studies of populations in the Tibeto-Yi Corridor—a region spanning Gansu, Sichuan, and Yunnan provinces along the eastern edge of the Tibetan Plateau (Z. Zhang et al., 2022). Therefore, such a Sinicized environment created the cultural context for Chinese marriage institutions and their terminology to diffuse among Loloish communities.

However, Nanzhao’s governance, as well as its succeeding reign Dali Kingdom (937 to 1253 CE), was weak in the south. Nationwide, because Chinese capitals have always been located to the north, the control of frontier lands throughout Chinese history has consistently spread from northern China following a north-to-south penetration pattern, beginning as early as the Qin State. Consequently, the more northern regions have always been better controlled than the southern regions, correlated with the distance from the central government. As Giersch (2006, p. 25) points out, Han Chinese migration and military settlements consistently struggled to reach the deep south, even as late as the Ming Dynasty; their progress was largely hindered by rugged terrain, endemic tropical ecology, and diseases. This enabled southern local societies to preserve their autonomous political and sociocultural institutions.

Regionally, the Nanzhao and Dali kingdoms, parallel non-Chinese states in Yunnan during the Chinese-governed Tang–Song period, functioned as early Southeast Asian “mandala” states which prioritizes influence over territory, personal loyalty over institutions, and the center over the periphery (Daniels, 2021, 2023). Nanzhao’s power center was located in the Erhai Lake basin (around present-day Dali). As the kingdom expanded southward, its direct governing capacity progressively weakened; in the southern periphery, Nanzhao relied mainly on indirect governance by recognizing existing local leaders as agents (Daniels, 2021).

The administrative structure of the Dali Kingdom, founded by Duan Siping (ca. 937–944 CE), resembled that of the expansive and vigorous Nanzhao Kingdom (ca. 738–902) (Daniels, 2023). Dali’s southern control was similarly weak. According to the Imperial Gazetteer of the Great Qing (vol. 389), Dali’s southern control had even shrunk. For example, the Yinsheng Military Command (modern Jingdong, Yunnan), established by Nanzhao, was captured by southern non-Chinese tribes and relocated north to Weichu (modern Chuxiong, Yunnan). The History of the Yuan Dynasty confirms that after Yinsheng fell, the land to the south was occupied by non-Chinese tribes, and the Dali rulers were never able to reconquer it.

Therefore, while the northern cultural groups of the Nanzhao and Dali kingdoms absorbed Han Chinese governance, they held only loose alliances with southern tribes and lacked the capacity for cultural assimilation. In other words, southern groups were more likely to maintain indigenous sociocultural practices and had little motivation to borrow the institutionalized practice of matchmaking or the word associated with that concept.

Finally, according to Haspelmath and Tadmor (2009), culturally motivated borrowing involves the importation of a cultural concept together with its lexical label. For example, Imbabura Quechua borrowed the Spanish words arrusa ‘rice’, riluju ‘clock’, and simana ‘week’, all terms for items that did not exist in the Americas prior to the European invasions. Chinese marriage customs traditionally involve a professional matchmaker; Loloish people, probably starting from the elites, emulating these rites, would naturally borrow both the institution and its name. Once introduced, the term and the practice could rapidly permeate through sister clans. In modern Loloish societies, besides the whole Liangshan Yi community in Sichuan, matchmakers play important cultural roles in other Loloish communities, such as Sani (fu³³kɑ⁵⁵, Southeastern Loloish in Kunming, Yunnan) (See matchmaking practice of Sani in X. Ma, 1951, p. 280), Nasu (fu³³tɕo⁵⁵, Northern Loloish in Luquan and Wuding, Yunnan)8, and Lolo (fu³³tɕo⁵⁵, Central Loloish in Nanhua, Yunnan)9.

7. Conclusions and Limitations

This study has demonstrated that the Liangshan Yi term hɔ⁴⁴ka³³ (Niesu) or fu⁴⁴ka³³ (Nuosu) for ‘matchmaker’, and ho³³ (Niesu) or fu³³ (Nuosu) for ‘to marry (generally)’, are both borrowed words from Chinese rather than as inherited lexical items. The argument rests on the convergence of morphological, phonological and lexical evidence, each of which independently disfavors an internal origin while jointly supporting a contact-based explanation.

The Liangshan Yi “matchmaker” is morphologically opaque: its formative morphemes cannot be decomposed into productive and meaningful units in the language. By contrast, the Chinese form 伐柯 (fákē), as the donor word, is morphologically analyzable: 伐 or fá: ‘to cut, to matchmake’, and 柯 or kē: ‘axe handle’, and exhibits clear semantic development into the meaning of “matchmaker”. This asymmetry in morphological transparency provides the first indication that the Liangshan Yi form was borrowed as a whole word from outside the language rather than created internally. Moreover, since another semantically related Chinese verb, 伐 or fá ‘to axe, to matchmake’, was also borrowed, the first morpheme of Liangshan Yi “matchmaker” becomes identifiable. However, after the borrowing, the original meaning of “to matchmake, to arrange marriage for someone” changed to “to marry” in Liangshan Yi: ho³³ or fu³³ ‘to marry (generally)’.

Lexical evidence, specifically through cognate comparison with other Sino-Tibetan languages, further supports this finding. No cognate set and comparable word formation for “matchmaker” are found, hence lacking the foundation for a reconstructed form across both Loloish and more distant Tibeto-Burman branches. Although phonetically similar forms appear in Northern, Central and Southeastern Loloish languages, such as Luquan Yi, Bijie Yi, Weishan Yi (Lalo), and Sani, no sound correspondence can be established. This suggests that the borrowing was likely from a common source, but that these Loloish languages had already split from one another by the time when the borrowing occurred. The absence of systematic correspondence and comparable morphology across genetically related branches constitutes strong evidence against the inheritance hypothesis.

Phonological evidence allows the borrowing to be situated within a specific historical interval. The adaptation of Chinese labiodental fricatives into Liangshan Yi, together with the treatment of vowel nucleus, points to a borrowing event that took place close to the completion of labiodentalization in Chinese but precedes its stabilization, or no later than 1324 CE, namely around the time of Late Middle Chinese to Early Mandarin (ca. 9th century to early 14th century). This timeframe is corroborated by the Chinese historical textual evidence from CCL corpus, which shows that the matchmaking-related senses of fá and fákē only become salient in Late Middle Chinese. The alignment of phonological adaptation and semantic resurgence strengthens the inference that the words and the concepts entered Liangshan Yi during the same period.

Beyond clarifying the origin of the lexical items, this paper has broader implications for the study of language and cultural contact. The linguistic form of 伐柯 fákē, as well as 伐 fá, serves as direct evidence that the associated marriage practice was borrowed from Han Chinese, a pattern in line with culture-motivated borrowing (Haspelmath & Tadmor, 2009). That is, matchmaking is not a traditional practice among Liangshan Yi community, nor in many other Northern Loloish groups and some Central and Southeastern ones; rather, it was introduced under the influence of Han Chinese marital customs between the 9th to 14th centuries, most likely from the mid-10th century (since Song Dynasty) to 1324 CE.

However, it remains unclear whether “matchmaker” in Liangshan Yi was borrowed through direct Yi-Chinese contact or indirectly via another Loloish language. So far, since we lack evidence to distinguish direct versus indirect borrowing, the present study can only rely on phonological evidence to infer the timing of the borrowing. Another limitation of the present study is that we lack systematic evidence to pinpoint the exact local Chinese dialect as the donor language. This is also one of our future goals of collecting more Chinese loanwords of greater historical depth, observing their phonological adaptation in Liangshan Yi, and reconstructing the phonological systems of the local Chinese dialect(s).

Finally, from a methodological perspective, this study highlights the potential value of drawing on linguistic evidence to investigate the intangible cultural practices of past peoples, a notable strength relative to archeology and genetics.