2.1. Spatio-temporal metaphors

In Conceptual Metaphor Theory (Lakoff & Johnson, Reference Lakoff and Johnson1980), a conceptual metaphor exists when concepts from a source domain are employed in the conceptualization of a target domain. Even before Lakoff & Johnson’s seminal work, Clark (Reference Clark and Moore1973) had already described the two primary images used to conceptualize time in English, dubbed Moving Ego and Moving Time. In Moving Ego, the person in the present ‘approaches’ time events that are ‘ahead of’ her, that is, in the future; the past lies ‘behind’. In Moving Time, the stationary observer faces the future and watches as time events ‘approach’ her. Each time event has a ‘front’ and ‘back’ so that future events that are ‘ahead of’ other events are also closer to the speaker, whereas ‘behind’ events are further away. This explains the apparent contradictions in languages that have expressions implying a future-facing ego but associate ‘front’ with past in other expressions, for example, English (Lakoff & Johnson, Reference Lakoff and Johnson1980) or Mandarin (Yu, Reference Yu1998). Thus, English before, which indicates an earlier time but derives from a root meaning ‘front,’ represents a conceptualization of the ‘front’ side of the time events and does not imply that Ego is facing the past.

While temporal reference can be handled with many types of linguistic devices, including iconicity of sequencing, dependent verb forms, tense/aspect distinctions, etc., many of which do not particularly involve metaphor, the use of spatial concepts in at least some part of the conceptualization of time is a near-universal trend across the world’s languages (see, e.g., Haspelmath, Reference Haspelmath1997; Kouteva et al., Reference Kouteva, Heine, Hong, Long, Narrog and Rhee2019). The last several decades have seen the development of a vast literature on spatio-temporal metaphors, for example, Boroditsky (Reference Boroditsky2000), Boroditsky et al. (Reference Boroditsky, Fuhrman and McCormick2011), Boroditsky and Ramscar (Reference Boroditsky and Ramscar2002), Casasanto and Boroditsky (Reference Casasanto and Boroditsky2008), Núñez and Cooperrider (Reference Núñez and Cooperrider2013), and Moore (Reference Moore2011, Reference Moore2017), including interesting recent findings for Aymara (Núñez & Sweetser, Reference Núñez and Sweetser2006), Moroccan Arabic (de la Fuente et al., Reference de La Fuente, Santiago, Román, Dumitrache and Casasanto2014), Vietnamese (Sullivan & Bui, Reference Sullivan and Bui2016), Mandarin Chinese (Li & Cao, Reference Li and Cao2018), and Yupno (Cooperrider et al., Reference Cooperrider, Slotta and Núñez2022).

2.2. Spatio-temporal reference in Sinitic

Despite the diversity in phonology and other structural features across the Sinitic languages, the primary conceptual structure for temporal metaphors is generally consistent. Very similar temporal expressions utilizing the front-back axis and the up-down axis are used all across the Sinitic group, suggesting that these metaphors were inherited from an early ancestor language as the Sinitic languages developed. For the front-back axis, the relevant orientation comes from the morphemes qián ‘front’(前) and hòu ‘back’ (後/后). These can be used in expressions that point to a future-facing ego, like wǒmen qiánmiǎn de lù ‘the road in front of us,’ which can be interpreted as meaning ‘our future,’ but they occur more often in expressions where ‘front’ = ‘before’ and ‘back’ = ‘after,’ reflecting frames with time reference points. If event B is ‘in back of’ event A, it comes after it, as in (1). If no event A is specified, as in (2), the reference point is assumed to be either the time of speaking or a time retrievable from discourse context. Table 1 presents additional examples.

Table 1. Front-back axis in Mandarin

(Note: Mandarin pinyin transcription is used in the examples and discussion here. Bold font marks morphemes referencing a spatial axis.)

The literature on conceptual metaphor has discussed these mappings in Mandarin (e.g., Boroditsky, Reference Boroditsky2001; Yu, Reference Yu1998), but it is important to note that they are employed across the entire Sinitic group. For sake of space, only Mandarin examples are shown here, but corresponding examples from Cantonese, Wu, Hakka, and Taiwanese (Hokkien/Southern Min) are provided in the Appendix (Tables A.1 and A.2).

The up-down axis is used for indicating specific time periods with reference to the present, as in ‘next week’ or ‘last week.’ ‘Up-down’ expressions also indicate earlier or later segments of a time unit, as in ‘the start of the month’ or ‘the second half of the day’ (see Table 2). The member of each pair expressing the earlier time (‘last week’, ‘start of the month’) is indicated with shàng ‘up, above’ (上) or tóu ‘head’ (traditional: 頭, simplified: 头). The later time (‘next week’, ‘second half of the month’) is indicated with xià ‘down, below’ (下) or dǐ ‘base’ (底). Terms with zhōng ‘middle’ also exist for some of them: zhōngwǔ ‘noon,’ zhōngxún ‘middle third of the month.’

Table 2. Up-down axis in Mandarin

Corresponding forms for other Sinitic languages are provided in the Appendix (Tables A.1 and A.2), all of which are cognate with the Mandarin forms provided here, with the following exceptions: Cantonese uses méih ‘tail’ rather than xià in expressions for ‘end of the month/year,’ and Taiwanese has téng ‘peak’ where others have shàng. Unexpected forms for ‘morning’ and ‘afternoon’ were found for Wu, but my consultant was not able to gloss the individual morphemes. Taiwanese also has āu ‘back’ in the expression āu lé-pài ‘next week,’ and it does not use the expected cognate of Mandarin shàng in the word for ‘morning’ (tàu zà ‘noon early’).

2.3. Metaphor and language contact

Conceptual metaphor and language contact are situated in the minds of individual bilingual speakers, but their effects can be observed anywhere from the behavior of individuals in discourse contexts and psycholinguistic experiments to the lexicon and grammar of entire languages. The relationship between individual speakers and language systems is cyclical and dynamic: the structure of a language conventionalizes out of the aggregated usage patterns of individual speakers, and then that structure in turn shapes the cognitive development of present and future generations of speakers, and so on. This study surveys language systems as represented in grammatical descriptions and collections of texts and conversations, a first step which can generate hypotheses about what structures might exist in the cognition of individuals who speak these languages.

In the language contact literature, transfer of lexical material or grammatical patterns happens when a bilingual speaker accesses elements of their repertoire in one language while speaking in the other (e.g., Matras & Sakel, Reference Matras and Sakel2007; Thomason & Kaufman, Reference Thomason and Kaufman1988; Weinreich, Reference Weinreich1953). This often results in L1 features appearing in a speaker’s L2, but it can go the other way as well. In recent cognitive linguistics literature, the term conceptual transfer has been used to describe situations in which a speaker’s use of one language reflects the conceptual categories of another language, typically with L1 influencing L2 (e.g., Sharifian, Reference Sharifian2015; Wolf & Polzenhagen, Reference Wolf and Polzenhagen2009) in aspects like motion events (e.g., Brown & Gullberg, Reference Brown and Gullberg2011; Daller et al., Reference Daller, Treffers-Daller and Furman2011) or time/tense and emotions (Odlin, Reference Odlin2005).

The idea of conceptual metaphors being transferred has received less attention – for example, only one chapter in Callies and Degani’s (Reference Callies and Degani2021) volume on metaphor in World Englishes focuses directly on the transfer of metaphorical conceptualizations. In a study on Akan and English, Ansah (Reference Ansah2011) explores the metaphorical conceptualizations of anger and fear in linguistic expressions reported by monolingual Akan speakers and Akan–English bilingual speakers as compared to the literature on native/monolingual English. While subtle differences were found between the groups, bilingual speakers did not report using any Akan-specific mappings in their English. Mendes de Oliveira (Reference Mendes de Oliveira, Callies and Degani2021), on the other hand, does find linguistic and gestural evidence for metaphor transfer in an analysis of two video interviews with a bilingual speaker of Brazilian Portuguese and English.

A recently proposed framework called Cognitive Contact Linguistics (Zenner, Reference Zenner2013) combines cognitive linguistics and language contact. Its central objective is ‘to explore how the guiding principles of Cognitive Linguistics apply to the bi- or multilingual mind in its dynamic bi- and multilingual environment, how this feeds back to our general understanding of these guiding principles, and how we can as a result better grasp how the interaction between cognition and context results in contact-induced variation and change’ (Zenner et al., Reference Zenner, Backus and Winter-Froemel2019, p. 4). Here too, though, the idea of metaphor in contact has just begun to be explored, and so far has mainly been applied to varieties of English. Chapter 5 in Zenner et al. (Reference Zenner, Backus and Winter-Froemel2019) explains spatial expressions in Irish English in relation to metaphors of containment and support in Irish, and Chapter 6 finds variation in the cultural models of witch, woman, and homosexuality in British, Indian, and Nigerian English, which will lead to different mappings and entailments in metaphors involving these concepts. The present paper can be seen as broadly fitting within the program of Cognitive Contact Linguistics, aiming to extend its application beyond contact situations involving European languages and to further explore the idea of metaphors in contact.

Finally, in the psycholinguistics literature, Boroditsky’s (Reference Boroditsky2001) classic study on time metaphor in Mandarin and English does not frame the experiments as being primarily about language contact, but they do in fact deal directly with bilingualism and contact. When the Mandarin–English bilingual participants in the first two experiments showed faster reaction times in responding to ‘earlier/later’ prompts in English after being primed with vertical images, they were assumed to be accessing the metaphorical conceptual structure of their L1 (Mandarin) while performing tasks in an L2 (English). In the third experiment, language contact was simulated by teaching English speakers ‘a new way to talk about time’ using the English words ‘above/higher than’ and ‘below/lower than’ in Mandarin-style constructions for ‘before’ and ‘after.’ Even brief training with the novel metaphoric constructions yielded statistically significant effects for these artificially ‘bilingual’ participants.

The results surveyed above indicate that the transfer of spatio-temporal metaphors in the cognitive systems of individual bilinguals does happen. The utterance in (3), attested in casual conversation, indicates that the up-down mapping from the speaker’s Sinitic L1 is being accessed as he produces the sentence in his L2 English. The phrase ‘at the bottom of this month’ corresponds to the Mandarin equivalent zhè yuè dǐ ‘this month base’ (cf. Table 2 above), in which the concept of a lower physical part stands for the later part of a temporal unit.

If such transfers happen in the cognition and speech of a sufficient number of bilingual speakers of a language, the transferred metaphors could eventually become part of that language, so that speakers in later generations can acquire the borrowed conceptual mapping even if they themselves are not bilingual in the original model language.

There have been some suggestions of this in the literature on spatio-temporal metaphors in East and Southeast Asia, but it has not been pursued in detail. Radden (Reference Radden2011) describes spatio-temporal metaphors in Mandarin, Korean, Japanese, and Vietnamese, and the end of the paper briefly suggests that contact could be responsible for some of the similarities. Bisang (Reference Bisang1996) describes the grammaticalization pathway noun > class noun > relational noun > conjunctional noun, specifically discussing the ‘front’ and ‘back’ nouns in Chinese that developed functions as temporal conjunctions, and demonstrates similar functions relational nouns in Hmong, Vietnamese, Thai, and Cambodian. However, Bisang’s discussion does not raise the possibility of a contact-based account of the similarity, and Radden does not go into any detail beyond just suggesting contact as an explanation. This leaves us, therefore, with a gap to be explored.

3.1. The Northwest

Northwest China is an area of contact among Turkic, Mongolic, Bodic, and Sinitic languages. Historically, it was an area of Silk Road trade, but also an area of conflict as various groups migrated in and out of the region and competed for control. In the Qinghai-Gansu (Amdo) Sprachbund, languages from all four groups have converged in morphosyntactic, phonological, and lexical aspects. As Mongolic and Tibetan speakers acquired Northwest Mandarin centuries ago for trade and through intermarriage, local Sinitic varieties began to emerge which shifted to OV word order, developed case, number, and tense-aspect morphology, and reduced or lost tone (see, e.g., Dwyer, Reference Dwyer1992, Reference Dwyer1995; Peyraube, Reference Peyraube, Xu and Li2017; Sandman, Reference Sandman2021; Zhu et al., Reference Zhu, Chuluu, Slater and Stuart1997). A large number of Turkic speakers shifted to Mongolic in the 13-14th centuries, when the Mongol empire ruled China, which shaped the early development of local Mongolic varieties like Santa (Field, Reference Field1997, p. 7ff.). Increased bilingualism with Mandarin in recent generations has brought many Sinitic loans and constructions into the other languages (Field, Reference Field1997; Slater, Reference Slater2003).

Mangghuer, spoken in this region, was once considered a true ‘mixed language’ (Slater, Reference Slater1998), though more recent analysis sees it as identifiable as still clearly Mongolic (Slater, Reference Slater2003). In some of its constructions for temporal sequence, Mangghuer uses the spatial terms mieshi ‘front’ and khuonuo ‘back’ (see (4) and (5)), indicating a spatio-temporal metaphor using the front-back axis.

Bao’an Tu (Mongguor), another Mongolic language, shows evidence of front-back temporal metaphor with phrases involving the items ŋamada ‘behind/after,’ kuda ‘front/before,’ ɕintɕhada ‘behind/after,’ and jantɕhada ‘front/before,’ as in (6) to (8).

Santa, another Mongolic language spoken in Gansu and also further northwest, uses similar front-back expressions with meliə ‘front/before’ and quəina ‘back/after’ (see (9)). It also has a construction with a borrowed form of Chinese yǐhòu, as in (10).

There are plenty of Sinitic loanwords in these languages – for example, ji-shi-nian and zui in the Mangghuer examples, and dɑgɑi liuʂi niɑn and ixəu in Santa. However, there is no clear evidence that the conceptual metaphor itself was transferred from Sinitic. Front-back time metaphors are attested across the Mongolic family, including in varieties that have had much less bilingualism with Chinese until very recently. For example, in Khalkha Mongolian, ömnö covers both ‘before’ and ‘in front of’ (cf. Santa meliə) and xoyno is ‘behind/after’ (cf. Mangghuer khuonuo, Santa quəina) (Lubsandorji & Vacek, Reference Lubsandorji and Vacek2004). Dagur Mongolian in the northeast uses the morphemes emele and huaine similarly (Martin, Reference Martin1961). This indicates that the metaphors are inherited from Mongolic ancestor languages, not borrowed through contact.

The sources I consulted did not contain any evidence that these languages use the up-down axis in temporal constructions. The only instance of ‘last’ or ‘next’ that appeared in the collection of Mangghuer folktales used mieshi ‘front’ for ‘last’ (see (11)). Khalkha Mongolian uses ‘back’ for ‘next year’ and a Moving Time schema for ‘last year,’ as in (12).

Salar, a Turkic language spoken in the Qinghai-Gansu Sprachbund, also employs a front-back metaphor for time, using ardʒi for ‘behind, after’ and ili for ‘front, before’ (Dwyer, Reference Dwyer2007). The Salar term for ‘afternoon’ is ojlie soŋ-ɨ ‘noon bottom/end-3poss.’ While this construction shares some similarity with Mandarin xiàwǔ ‘down-noon,’ the temporal use of soŋ ‘bottom/end’ actually reflects Salar’s connection with Western (Oghuz) Turkic – cf. Turkish öğleden sonra ‘afternoon’) – and probably does not reflect an up-down time metaphor. According to Dwyer, the root originally meant ‘the end (of something)’ and was extended later to mean ‘bottom’ or ‘behind.’

Amdo Tibetan, a Sino-Tibetan language that is widely spoken in the Qinghai-Gansu Sprachbund, also has words with both front-back spatial meanings and temporal uses. In Amdo, sngonna means ‘in front’ and ‘before,’ while gzhugna means ‘behind’ and ‘after’ (Dpal, Reference Dpal2016).

The Sinitic languages spoken in this region have been significantly restructured through contact with Altaic and Tibetan languages, to the extent that some have been classified as ‘mixed languages’ (Zhu et al., Reference Zhu, Chuluu, Slater and Stuart1997). Temporal expressions using both spatial dimensions are attested in at least some of the descriptions. Example (13) shows a front-back expression from the Chinese variety spoken in Xunhua, Qinghai Province (cf. Mandarin zuìhòu). The variety spoken in Tangwang also has the expected front-back expressions related to qián and hòu (Xu, Reference Xu2017), and Gangou Chinese has hòu for ‘later’ (Zhu et al., Reference Zhu, Chuluu, Slater and Stuart1997).

In descriptions of other local Sinitic varieties, the typical Sinitic time metaphors are not as clear. In the Wutun variety, for example, the temporal adverbs godangma and wuzizi, both meaning ‘before, earlier’ (Sandman, Reference Sandman2016, p. 171), are not cognate with the qián terms attested elsewhere in Sinitic. Wutun has typical Sinitic spatial terms qanmian ‘in front of’ and bimian ‘behind’ (cf. Mandarin bèimiàn, synonymous with the spatial sense of hòumiàn), but they do not appear in any of the temporal expressions in Sandman’s description (see (14) to (16) for the ‘after’ expressions).

As explained above, though, the Mongolic and Turkic languages spoken in this region use the front-back axis for time. It seems unlikely that contact would lead the local Sinitic varieties to lose the mapping altogether. Sandman does not mention the etymology of jera, xenrada, and co, so I am not sure if they derive from ‘back/behind’ constructions or if they are strictly temporal terms.

The up-down axis, on the other hand, is not used for time in any of the non-Sinitic languages. One might wonder, then, if the local Sinitic varieties have lost this mapping under contact with languages that do not have it. From the evidence in the sources consulted, it seems likely that they have retained it. Example (17) shows an up-down expression in Xunhua Chinese (cf. Mandarin xiàge lǐbài), and Wutun Chinese has the word xongwu ‘afternoon’ (Sandman, Reference Sandman2016, p. 218), which is cognate with Mandarin xià-wǔ (‘below-noon’).

The Turkic language Uyghur is spoken further west, in what is officially called the Xinjiang Uyghur Autonomous Region of China, but also known as East Turkestan or simply ‘the Uyghur region’. For Uyghur, intense contact with Sinitic began relatively recently. It was not until the second half of the 20^th century that many Han Chinese began to immigrate (Baki, Reference Baki2012). Most Uyghurs were not bilingual in Mandarin until recent decades, and the number of Han Chinese migrants who would have learned Uyghur was fairly small.

Similar to the Mongolic languages, Uyghur uses front-back morphemes in temporal constructions – ald ‘front,’ burun ‘nose; before,’ and keyn ‘back.’ The combination of future-facing ego and Moving Time units with fronts and backs is the same as Chinese (both orientations are exemplified nicely in a single sentence in (19)), and the syntax is similar with the ‘front’ and ‘back’ words at the end of a preposed temporal clause. However, the constructions go back to Old Turkic (Erdal, Reference Erdal2004) pointing to origins unrelated to contact with Chinese.

Uyghur does make some use of up-down imagery for time, employing bash ‘head’ and ayagh ‘foot’ in expressions like those in (20). While there is some similarity to the Sinitic use of ‘head’ and ‘base’ (yuè tóu, yuè dǐ in Table 2), it is unlikely that the constructions were transferred, as they are also used in other Turkic languages that have had little or no contact with Sinitic (cf. Kazakh törtinşi aydıñ basında ‘at the beginning of the fourth month,’ sekseninşi jıldardıñ ayağında ‘at the end of the 80s’). Uyghur also uses bash ‘head’ and ayagh ‘foot’ in verbal forms like bashlimaq ‘to begin’ and ayaghlashmaq ‘to finish.’ The ‘head’ verbs go back to Old Turkic, and to my knowledge, these specific uses of ‘head’ and ‘foot’ are not found in the Sinitic languages.

Uyghur does not use the up-down axis for any of the expressions where Sinitic languages do, employing either front-back terms as in (21) or Moving Time expressions.

Sibe is a Tungusic language originally from Manchuria but currently only spoken in the northwest of the Uyghur region. Sibe speakers were transferred there to resettle the area after the Qing empire’s genocide of the Dzungar population in the 1750s. In Sibe, the word aməɹ in (22) comes from the root ama ‘back,’ but the corresponding root jule ‘front’ is not typically used for ‘before.’ Instead, we see oɴoɹ, a borrowing from Mongolian, as in (23). However, the ‘front = before’ mapping survives in a few fixed forms, such as julge-i fon-de ‘long ago, once upon a time’ (front-gen time-dl) (Zikmundová, Reference Zikmundová2013).

To summarize, front-back time metaphors are common in the languages of the Qinghai-Gansu Sprachbund. However, while there is some lexical borrowing, there is no clear evidence for transfer of the metaphor. Additionally, none of the non-Sinitic languages here use up-down temporal constructions that could be traced to contact with Sinitic. The same is true for Uyghur and Sibe, spoken farther west.

3.2. The Northeast

To the northeast, Japanese and Korean had many centuries of contact with Chinese. The practice of writing was learned from the Chinese, and Classical Chinese was used as a high-status register for government and scholarship. While Japanese and Korean speakers eventually developed orthographies more suitable for their own languages, written Chinese was in use through the 19^th century in Korea, and heavily Sinicized formal registers of Japanese were in use well into the 20^th century (Loveday, Reference Loveday1996).

In Section 2.2, above we saw that the Sinitic languages use the up-down axis in expressions for ‘next’ and ‘last,’ for parts of the month, and for ‘morning’ and ‘afternoon.’ Japanese and Korean both use the up-down axis for time. Both have borrowed temporal expressions with the Chinese morphemes shàng and xià. Here, we see the strongest case for metaphor transfer among the languages surveyed for this paper.

In Japanese, the month can be divided into the first third, middle third, and last third using jō ‘up’ and ge ‘down,’ as in (24). Similar expressions are used for volumes of books in a series and sections of poems. These terms are borrowed from Chinese and are written with kanji. Jō and ge are quite different phonetically from Chinese shàng and xià, but they are indeed Sino-Japanese pronunciations of the kanji characters 上 and 下 (cf. 上下 jōge ‘up and down; above and below’). Versions of the poem terms also exist with native Japanese morphemes kami ‘up’ and shimo ‘down’ instead of jō and ge (Nelson, Reference Nelson1962).

Similar terms for parts of the year also involve the up-down axis, as in (25). The constructions here are from Chinese shàngbànqī 上半期 and xià(bàn)qī 下(半)期, and are written with these characters (Nelson, Reference Nelson1962), but in spoken form the initial morphemes are the native Japanese morphemes kami and shimo.

The morphemes jō and ge are also associated with up-down spatial orientation, as in (26). Borrowed Chinese xià is also pronounced ka, as in kakō ‘descent, fall, drop’ (下降) and kahō ‘lower part’ (下方).

For ‘next’ and ‘last,’ Japanese also uses borrowed Sinitic morphemes, but the terms do not use the up-down axis. ‘Last month’ is sen-getsu (‘first-month,’ cf. Chinese xiān yuè 先月), ‘last year’ is kyo-nen (‘go-year,’ cf. Chinese qù nián 去年), and ‘next year’ is rai-nen (‘come-year,’ cf. Chinese lái nián 来年).

The examples above in (24)-(26) indicate that a transfer of the up-down time metaphor occurred. Sino-Japanese jō and ge are used in both spatial and temporal expressions. Further evidence for the transfer comes from the fact that the native Japanese morphemes kami ‘up’ and shimo ‘down’ are also used in temporal expressions. It seems unlikely that the temporal use of the native morphemes existed pre-contact, since in all the temporal expressions I have seen involving kami and shimo, the other morphemes are clearly borrowed from Chinese. Additionally, a cross-linguistically rare feature like temporal use of the up-down axis is less likely to have originated independently in two neighboring languages than to have transferred through contact.

Korean makes use of the up-down axis via the Sino-Korean morphemes sang and ha, from Chinese shàng and xià. Example (27) lists several Sino-Korean words for parts of time units, which come from Chinese shàng/xiàxún (上/下旬), shàng/xiàwǔ (上/下午), shàng/xiàpiān (上/下篇), and shàng/xiàbànqī (上/下半期). In addition to these, se-mit ‘end of the year’ (lit. ‘year-lower’; Radden, Reference Radden2011, p. 6) uses native Korean morphemes.

As in Japanese, ‘up’ and ‘down’ are not used for ‘next’ and ‘last’ in Korean. For these functions, the non-spatial native morphemes taum ‘next’ and jinan ‘last’ are used, as in taum tal ‘next month’ and jinan tal ‘last month.’

Sang and ha are associated with spatial orientation in other Sino-Korean words, such as sangseung ‘rise, climb, increase,’ hagang ‘descent,’ jiha ‘underground,’ jihacheol ‘subway,’ jihado ‘underpass.’ The ‘up/down’ semantics of sang and ha are metaphorically extended in other typical ways, as in hyangsang ‘improvement,’ isang ‘greater than,’ and iha ‘less than.’ As in Japanese, then, we have up-down morphemes functioning in both spatial and temporal expressions, so this represents a likely case of metaphor transfer. The metaphor spread through borrowed lexical items and then native morphemes were substituted into borrowed constructions or used to create new items on the model of the Sinitic ones. Metaphor transfer via borrowed words makes sense as the contact primarily involved Japanese and Korean speakers learning Chinese (L2 > L1 transfer); if it had been large numbers of Chinese learning Japanese and Korean (L1 > L2 transfer), the metaphor-based constructions might have been copied even without borrowed words.

Japanese and Korean also use the front-back axis for time, and borrowed Sinitic morphemes are used in some of the relevant expressions, but there is not clear evidence that the mapping itself was borrowed from Sinitic. The borrowed morphemes are used alongside native morphemes for ‘front’ and ‘back’ that function in both spatial and temporal domains. For ‘after,’ Japanese uses the native morpheme ato ‘back’ more colloquially as in (28), but go (from Chinese hòu) is also used for temporal clauses as in (29) and in some temporal words or phrases like sono-go ‘subsequently’ and shoku-go ‘after a meal’ (Kaiser et al., Reference Kaiser, Ichikawa, Kobayashi and Yamamoto2013, pp. 83, 130). Similarly, native mae ‘front’ is used for ‘before’ more frequently than the Sino-Japanese zen (from Chinese qián), which occurs mostly in borrowed lexical items like chokuzen ‘immediately before,’ izen ‘earlier,’ or jizen ‘beforehand’ (Kaiser et al., Reference Kaiser, Ichikawa, Kobayashi and Yamamoto2013, pp. 232, 495, 631).

In Korean, the situation is similar. The Sino-Korean loans cen/jən ‘before’ and hwu/hu ‘after’ (from qián and hòu) are used alongside the native morphemes ap ‘front’ and twi ‘back’. The borrowed and native morphemes seem to be used in equally diverse contexts, including doublets like twi-nnal / hu-nnal ‘at a later date’ (Radden, Reference Radden2011, p. 23) and sam nyen twi-ey / twu sikan hwu-ey ‘in three years / in two hours’ (Haspelmath, Reference Haspelmath1997, p. 164), and temporal clauses as in (30) and (31). The borrowed form of qián also appears in jənjənal ‘front-front-day; day before yesterday’ (Radden, Reference Radden2011, p. 26).

It is clear that the lexical items for ‘front/before’ and ‘back/after’ were transferred from Chinese into both Japanese and Korean. In some instances, the borrowings are individual lexical items whose component morphemes are all Sinitic, like Japanese izen ‘earlier’ (Chinese yǐqián) or Korean hu-dae ‘future generation’ (Chinese hòu dài). In other cases, especially in Korean, the borrowed morphemes are thoroughly integrated into the grammar, and represent a standard way of expressing that function. Additionally, spatial uses of the morphemes exist in both languages, as in (32).

However, considering how common front-back time is cross-linguistically, and also considering that there are native morphemes in both languages that cover both spatial and temporal anterior/posteriority, it is likely that the use of front-back time expressions predated contact with Chinese. If pre-contact stages did not have the mapping, or if related languages not in contact with Sinitic did not have it, we might argue that the constructions with native morphemes were copied from the Chinese model. However, for both Japanese and Korean, there are no written records before contact with Chinese, and there is also no way to compare with closely related languages, since Korean is an isolate and the only relatives of Japanese (the Ryukyuan languages) were also in contact with Chinese. The conclusion, therefore, must be that contact has affected the lexical expression of front-back time, but there is not a strong case for a transfer of the metaphorical mappings.

Manchu, a Tungusic language spoken to the northeast of China, also had intense contact with Sinitic. The Manchu empire ruled China in the 17^th‑early 20^th centuries (the Qing dynasty), but it turned out that Chinese culture and language exerted significant influence on Manchu rather than the other way around. Manchu speakers quickly began shifting to Chinese, and despite maintenance and revitalization efforts, the language has relatively few speakers now (Gorelova, Reference Gorelova2002). In Manchu, the front-back axis is partially used for temporal expressions (see (33), and cf. (22) above for Sibe), but there is no evidence of borrowing from Sinitic. I did not find any examples of the relevant expressions for up-down time in the sources consulted for Manchu.

To summarize, both Korean and Japanese have borrowed Chinese expressions involving both the front-back and up-down axis. The front-back mapping may have been present before contact, but the up-down one can be analyzed as having transferred from Sinitic to both Korean and Japanese. Manchu has a front-back conceptualization of time, but it did not emerge through contact.

3.3. The South

To the south, the region of Vietnam was under Chinese administration from the early second century BCE to the early 10^th century CE. Classical Chinese was used as the written language for administration and education, and an adapted system of characters was devised for writing Vietnamese in the 11^th century. Massive lexical borrowing from Chinese occurred in several waves; it is likely that the greatest amount of everyday bilingualism happened after Sinitic-speaking immigrants from Fujian and Guangdong moved south into Vietnam in the 17^th century (Ngo, Reference Ngo2021).

Vietnamese uses trước ‘front’ and sau ‘back’ for ‘before’ and ‘after,’ as in (34) and (35). However, it is likely that the use of the front-back axis for time existed in Austroasiatic before contact with Chinese. Khmer, which belongs to the same branch of Austroasiatic as Vietnamese but has not had significant contact with Chinese, has a ‘behind ~ after’ mapping in the word kraoy (Haiman, Reference Haiman2011, p. 173), which goes back to Old Khmer (Jenner & Sidwell, Reference Jenner and Sidwell2010, p. 38).

In expressions for ‘next’ and ‘last,’ Vietnamese does not use the up-down axis, instead making at least partial use of the front-back axis, as in (36).

In contrast with Vietnamese, speakers of Hmong lived in relative isolation in what is now southern China until Chinese settlers moved into their territory in the late 17^th century, prompting armed conflict and subsequent migration (Culas & Michaud, Reference Culas and Michaud1997, p. 215). Some Hmong speakers migrated to Vietnam starting in the 18^th century, to Laos in the 19^th century, and Thailand in the 20^th (Kunyot, Reference Kunyot1984, p. 4). The communities in southern China have had continued contact with Sinitic languages, while those outside of China have not.

The front-back mapping is attested in Hmong varieties spoken in both China and Thailand. In Hmong Njua (Green Miao), spoken in Thailand, the temporal conjunctions are at the beginning of the dependent clause (see (37) and (38)).Footnote ¹

Sposato’s (Reference Sposato2015) description of the variety known as Xong spoken in Hunan and Guizhou provinces of China also shows the use of ‘in front of’ and ‘behind’ in temporal constructions, as in (39) and (41), but with notably different syntax. In (39), the clause-linking device is at the end of the temporal clause, as in Chinese, not the beginning, as in Hmong Njua. The overall syntax is exactly parallel to the Mandarin equivalent (see (40)), and it includes equivalent constructions involving the use of ‘hold’ as an object marker and ‘complete’ as an aspect marker.

Xong also has a number of front-back time words that are clearly borrowed from Chinese – ix.houf from yǐhòu 以后 ‘after,’ ranf.houb from rǎnhòu 然后 ‘then,’ and zeib.houf from zuìhòu 最后 ‘final’ (note that the final orthographic letter of each syllable indicates tone, not a phonetic segment). The construction in (42) parallels the Mandarin construction liáng-sān-niǎn yǐhòu 两三年以后 ‘two-three-year later.’

Heal’s (Reference Heal2020) sketch of a Hmong variety known as Mashan Miao, spoken in southern China (Guizhou), includes few examples of temporal expressions, but it appears that this variety has borrowed the morpheme hòu ‘after’ from Chinese as hob (recall that the b here represents tone, not a phonetic segment). The construction at the end of the sentence also uses hob, and while the free translation reflects a more natural English syntax, the Hmong seems to involve a temporal clause (‘…after returning, came to the house’). In (44), the temporal adverbial at hob ‘after’ looks like a possible borrowing of Chinese yǐhòu.

‘Front’ and ‘back’ in Heal’s description are nzouk ndaek and nzouk huob, respectively. It is not clear to me whether they are native morphemes (perhaps cognate with Xong geud-neul and geud-zheit) or borrowed, or whether they are used in any temporal constructions.

Considering expressions where Sinitic languages use the up-down axis, the Hmong Njua terms for ‘next’ and ‘last’ do not reference either spatial axis (see (45)). ‘Last month’ involves a Moving Time metaphor, but ‘next month’ does not appear to. None of the available data for ‘afternoon,’ ‘morning,’ and parts of the month suggest borrowing from Chinese, either in the lexical items or the conceptual structure.

For Qiang, which comes from the Burmo-Qiangic branch of Sino-Tibetan and is spoken in southwest China, contact with Sinitic was attested in ancient times, but widespread bilingualism was limited to the last century. Historically, only men who left their home villages for work needed to learn Chinese (and one assumes very few Chinese would have learned Qiang), but since the mid-20^th century, most Qiang people have begun shifting to Mandarin (LaPolla & Huang, Reference LaPolla and Huang2008, pp. 3–5).

In Qiang, ‘before/after’ temporal clauses do not use the front-back axis, but ‘front’ and ‘back’ are used for adverbials meaning ‘long ago’ and ‘afterwards,’ and also in at least some expressions for ‘next’ and ‘last,’ as examples (46) to (49) demonstrate. The ‘front/back’ morphemes do not appear to be borrowed from Chinese qián and hòu. The item tɕi-steke-le ‘the last one’ in (49) is parallel to Chinese 最后一个 zuì hòu yígè ‘most back one-cl,’ that is, ‘last,’ and the morpheme tɕi ‘most’ could conceivably be a borrowing of Chinese zuì 最 ‘most’. Modern Mandarin also uses adverbs involving qián and hòu for ‘before/long ago’ and ‘later,’ but the morphological structure of the Qiang words (e.g., steke-tɑ behind-loc ‘later’ [LaPolla & Huang, Reference LaPolla and Huang2008, p. 113], qe:^ɹ-ŋuəȵi before-top ‘earlier, before,’ [p. 73], qe:^ɹ-tɑ front-loc ‘in the past’ [p.171]) does not match the structure of the corresponding Mandarin words.

In Taiwan, the indigenous languages represent great diversity, with 20 or so languages constituting nine of the ten primary branches of the Austronesian family. (The other branch, Malayo-Polynesian, contains the other 1200+ languages of the family). Speakers of the indigenous languages had no contact with Sinitic until Hokkien (Southern Min) and Hakka speakers from the Fujian region of the Chinese mainland arrived in the 17^th century (Lin, Reference Lin2015). Migration to Taiwan was subsequently banned for almost two centuries until the end of the Qing dynasty, leaving a relatively self-contained contact situation. It was not until the second half of the 20^th century (after 50 years of Japanese control) that Mandarin entered Taiwan en force. Most of the indigenous peoples are now shifting to Taiwan Mandarin, but this is a recent development.

Lee (Reference Lee2016) gives a detailed analysis of the expressions and concepts for temporal relations in Kavalan, and briefly surveys six other related languages. Among them, several use spatial dimensions in conceptualizing time, but others do not. Kavalan uses ‘in front of/behind’ for ‘before/after’ with nominal time referents, including nominalized verbs (see (50) and (51)).

Isbukun Bunun also uses the front-back axis for time, as in (52) and (53).

Three other Taiwanese languages have an asymmetry in their use of spatial terms in temporal constructions. For Tsou, Pan (Reference Pan2007) reports variation between speakers in the use of spatial terms in temporal constructions. Tsou uses the specifically temporal markers n’a/na’a or auyu ‘firstly, at first’ and -epungu ‘finish’ for ‘before/after’ clauses. With nominal time referents, the spatial terms tan’esi ‘here, in front of’ and ta’esi ‘there/behind’ can be used as alternatives to the specifically temporal auyusi ‘first, early’ and ataveisi ‘at last, finally,’ as in (54) and (55). However, while all the speakers for Pan’s study accepted the temporal use of ta’esi for ‘after,’ the use of tan’esi for ‘before’ was only acceptable for speakers living in the town of Tfuya, while speakers in three other locations did not accept it as grammatical (Pan, Reference Pan2007, p. 88). In the languages Amis and Puyuma, on the other hand, ‘front’ constructions are used for ‘before,’ but ‘finish’ constructions for ‘after’ (Lee, Reference Lee2016). Saisiyat, Rukai, and Paiwan, three other Austronesian languages of Taiwan, are not reported to have any spatial constructions for ‘before/after.’

This variation might be evidence that these languages did not historically use the front-back axis for time, but they are beginning to do so now under increased contact with Taiwan Mandarin. If that is so, it would be a case of metaphor transfer in progress. On the other hand, as each of these languages comes from a different primary branch of Austronesian, the variation could simply be due to diversity within the family. Even if the front-back metaphors emerged independent of Sinitic contact, though, the extension of tan’esi ‘in front of’ to match the temporal use of ta’esi ‘behind’ in Tsou has probably been facilitated by contact with Mandarin (Pan, p.c.).

Looking at the up-down axis, there is no evidence of transfer from Sinitic. Expressions for ‘last month’ and ‘next month’ in Tsou use the terms auyu ‘first/early’ and faova ‘new’ (Pan Reference Pan2007, p. 69). In Kavalan, while lipay ‘week’ is borrowed from Chinese, the construction for ‘last week’ uses a term for ‘day before yesterday,’ (see (56)), which is unlike the Sinitic construction.

Finally, Wulai Atayal actually has the opposite up-down mapping from the Sinitic languages. The spatial term zik ‘below’ is used for ‘before,’ and βaβaw ‘above’ for ‘after.’ Moreover, the expressions employing this up-down mapping are similar to the ones where Sinitic languages use the front-back axis. For example, zik is used in expressions for ‘the day before,’ ‘two days in advance,’ and also in ‘before’ clauses (Lee, Reference Lee2016, p. 128).

To summarize, Vietnamese, Hmong, and Qiang all use the front-back dimension for time, but none of them borrowed the metaphorical mapping from Sinitic. The time expressions in Hmong varieties spoken in China show considerable lexical borrowing and syntactic restructuring, but the front-back mapping is attested in the Thailand variety that has had less contact with Sinitic. In Taiwan, several languages do not use the front-back dimension in temporal expressions, but others do. This represents potential evidence for metaphor transfer via contact with Taiwanese or Taiwan Mandarin. None of the languages surveyed in this section use the up-down dimension for time, except for Wulai Atayal, whose mapping is the opposite of the Sinitic one and thus does not constitute evidence of convergence.