Convergence of Hakka with Chinese in Taiwan

1.1 The Hakkas and Taiwan

The Chinese language spread through migration from the North of China to the Southern provinces where it developed its own local varieties, nowadays known as ‘Cantonese’, ‘Hokkien’ and other ‘fangyan’ (dialects). Later migrations, between the fourth and fourteenth century (Sagart and Chappell 2001) met already established populations and therefore newcomers settled in the less fertile mountainous regions between the economic centers and river systems (cf. Leong 1997). The newcomers in a space already occupied by earlier inhabitants speaking Sinitic varieties often faced a disadvantaged situation, e.g. as tenants (hence probably the name hag ga, 客家 = ‘guest people’), ultimately experiencing conflicts with the ‘Punti’ (i.e., the locals, now the ‘Cantonese’). The identity of being ‘Hakkas’ was formed on the basis of this conflict, discrimination and outcasting by other groups and fully surfaced in the nineteenth century (cf. Branner 2000: 43; Christiansen 1998), with a juxtaposed self-image of being hardworking, innovative and mobile, thereby reinterpreting their ethnonym ⟨hag ga⟩ ‘guest people’; at the same time, the emerging Hakka consciousness emphasized their Hàn adherence, probably seeking protection by adherence to the larger identity, and opposing Punti accusations.

The language of the Hakkas is a range of dialects related to Middle Chinese and classified as being closest to Gan and Cantonese, with other dialects (in the dialect continuum) showing similarities also with Hokkien. Specifically, incidents of negative representations of Hakkas in textbooks (in 1905, 1920) led to strong protests followed by the foundation of the Tsung Tsin association (1921) in Hongkong (Leong 1985, 311f.; cf. Christiansen 1998: 4). Hakkas played an important role in the building of modern China, but are also a substantial part of the overseas Chinese worldwide. The greatest number of overseas Hakkas emigrated in the twentieth century, upholding this Chinese-Hakka self-identification in their new homelands.

The Taiwanese Hakkas arrived from the seventeenth century onwards among a Hokkien-speaking majority (‘Hoklo’), experiencing once again conflicts with and discrimination from the majority which may have enforced their group identity and made them settle in specific, more marginal areas, often close to the then still “raw” (生番 shēng fān) or “mountain” (山番 shān fān) Austronesians (Carroll 2015: 237). This proximity made the Hakkas intermediaries between Hoklo and tribal cultures which led to a positive picture of Hakkas among the Austronesian groups. The negative attitude of the main culture towards Hakkas is documented in accounts from missionaries and travellers (e.g.: “the Hakkas—the outcaste class of China—fled to Formosa to escape persecution in the mother-country”; Montgomery McGovern 1922: 46). It has been observed that settling in the margins of centres occupying marginal economic niches and thereby becoming quite successful is a recurring theme in Hakka migration (cf. Oxfeld 2007).

In the twentieth century, language planning negatively affected both groups under Japanese (1895–1945) and Kuomintang (KMT) rule (1946–1987), trying to first Japanize, then Sinicize the population (cf. Heylen 2005). After the KMT mainlanders had occupied all leading positions in the country for decades, a (new) Taiwanese identity slowly formed. Since the 1980s, grassroots movements fought for the recognition of the historical heritage of the Hoklos and Hakkas as ethnic groups of Taiwan. The Hoklo-movements termed the Hoklo language ‘Taiwanese’ (as opposed to ‘Chinese’, cf. DeChicchis 1995) and thereby, among other things, marginalized the Hakkas once again, which led to more, partly radical protests (cf. Christiansen 1998: 18f.) in the quest for political participation. The “Hakka Basic Law” (2018) grants the Hakkas all rights as an ethnic group of Taiwan and recognizes Hakka as one of the national languages. Today, Hakka is used in public discourse, media and in schools in Taiwan, and the demographics are well-established (Hakka Committee 2016).

The Hakkas in mainland China and overseas traditionally emphasize their Hàn identity, while Hakkaness is seen as a piece of Taiwanese culture in Taiwan (cf. Hsiao and Lim 2007). The new politics of Taiwan (after ca. 1989) ultimately aimed at including all local cultures in the definition of Taiwan as a ‘diverse culture’; in this respect, the Taiwanese Hakkas had to shed their traditional association with Hàn culture and reorient themselves as “New Hakkas” and as “Taiwanese Hakkas” (cf. Wang 2007: 880f.), a unique identity different from other Hakkas (cf. Wilson 2009; Shen 2010; Tanoue 2012).

However, this ascension of Hakka identity is not reflected in language use, although the language is widely considered to be a marker of Hakka ethnic identity. Furthermore, Hakka newcomers do not seem to be included in this new national identity.

1.2 Linguistic identities

Identities are not fixed, but changeable, dynamic and with unclear boundaries, in spite of usually presenting as objective, fixed and definable – characteristics which may serve political purposes, ironically, by being modeled in a certain way. An identity emerges between individual and society, and established identity types are elements of a social reality (Berger and Luckmann 1991 [1967]: 195). Identity is construed as a characteristic of the individual, but it is based on shared, often inalienable group characteristics such as descent (race; cf. Tajfel and Turner 1986), culture (ethnicity, ancestry; cf. Stets and Serpe 2013), language competence, religion or other shared beliefs and identifiable collective interests. ‘Ingroup favoritism’ (cf. social identity theory) favors collaboration as a group and strengthens the difference towards the outgroup.

From a functional viewpoint, languages are sign systems for the communication between people. Although language needs to be learned, a native competence can act as an inalienable characteristic, in spite of its fluid nature including variation (dialects). A language is actually identified through an act of identity building, as this process requires the identification of a range of speech habits as being one phenomenon associated with a speaker community, i.e., an ingroup. From the identification onwards, this ‘language’ has a symbolic value. The sociolinguistic roles of languages have been described as overt or covert prestige (cf., e.g., Chambers and Trudgill 1998). A minority group (e.g., dialect speakers) usually assign overt prestige to the dominant language, and covert (ingroup-building) prestige to the non-prestigious, local vernacular.

Individuals have to cope with communicative situations irrespective of the symbolic (identitarian) value of languages. Traditional multilingualism in the form of diglossia (following Ferguson 1959) reserved complementary functionalities to each language which provides functional domains (cf. Ehala 2019) to every one of the languages. In the modern situation, a standard language fills out most or all functional communicative domains, so that (young) speakers acquire first a reduced variant of the smaller language, and then give it up altogether (cf., e.g., Dorian 1989). In recent times, a general globalisation of language use (with general education and focus on codified languages) leads to swift language shift from small, unwritten languages, to ‘national’, often ‘global’ languages. This development leads to a multilingualism of registers of a language which often includes ‘dialects’ or ‘accents’ into its range of sociolectal variation (cf. Kloss 1952, 1967) – with the covert, ingroup prestige of a social subgroup.

Modern educational and economic aspirations of individuals coincide with the demands of a ‘literate’ society where social and economic success depends on the ability to use an elaborated register of a standard language; smaller, uncodified, spoken languages usually lack an elaborated (‘literate’) register and therefore appear inappropriate for many high-level functions in a highly educated, complex modern society, and the promotion of their use appears as a paradoxical double-bind in practical terms. Language loss may be a sign of education and globalised communication, the language experiences a transition phase of grammatical reduction and “stylistic shrinkage” (cf. Campbell and Muntzel 1989) in ‘weak(er) speakers’, finally ‘rememberers’ of the language, before it will eventually disappear entirely. This is currently a sad international trend despite promotional measures.

1.3 The pros and cons of language promotion

While languages are usually just used, not discussed by the speakers, in the precarious situation of language loss, a ‘desire’ for the continuation of the smaller language will arise. The language becomes a symbol for an identity to be preserved. Language shift seems to be opposed by cultural activists mainly because it poses a problem to the ideological ‘justification’ of identity and is more difficult to overcome than the preservation of most other cultural characteristics, because linguistic competence is not easily acquired. For the same reason, language is a problematic identity marker which may get lost without dissolving the identity it represents.

Minorities often seem to develop additional markers, e.g., in clothing style, in order to ‘maintain’ their identity, for instance, the red clothes of the Austronesian groups in Taiwan. Similarly, they may demonstratively use the language on certain occasions, which is a symbolic, not a communicative use.

If a smaller language is promoted through education and media, it actually first needs to get ‘elaborated’ for this usage, and thereby standardized. Grammars and lexica are written, and this very process creates many new words and constructions. This leads to a major language change, usually approaching the smaller language to the dominant standard language which serves as a model: Lexemes will be borrowed (or loan translations will be produced), new grammatical structures will be adopted. This actually makes the minority language more connected to the dominating language. A standardization will also alienate uneducated native speakers.

1.4 Sociolinguistics of the Hakka language

The various linguistic distinctions of Taiwan are considered markers of a Taiwanese ‘identity in cultural diversity’, in reaction to former attempts for cultural homogenisation through the imposition of Guoyu, the Chinese standard language. Therefore, Taiwan takes great care in protecting languages, providing learning material, and keeping traditions alive. Ethnic identities are even important for tourism, pointing to, e.g., ‘Hakka townships’ of ‘Hakka villages’. Nonetheless, such efforts at the organisational level must meet everyday communicative practices, which is where problems will inevitably occur.

In Taiwan, Hakka is mostly spoken in specific areas which makes it a geographical variety (‘dialect’); on the other hand, social and geographical mobility lead to fast urbanisation, a situation in which Hakka speakers will be put in a different linguistic situation occupied by Guoyu and the colloquial Hoklo.

Hakka as a spoken minority language in a modern society can therefore mainly serve as a family and local ingroup language. The language practices of the Taiwanese Hakkas are described as oriented towards Guoyu, especially in the intergenerational transmission: “[…], Hakka people used significantly more Mandarin in talks with their own children than the Hakka language” (Yeh et al. 2004: 92). Contrary to Hoklo speakers, Hakkas often know Guoyu better than the Hakka language (Yeh et al. 2004: 85f.). Interlinguistic marriages (e.g., between Hakka and Hoklo speakers) usually lead to language shift away from Hakka. Therefore, in spite of the efforts performed through language planning and preservation efforts, it is not evident that this minority language would experience the exceptional development of Basque (a minority language with growing numbers of speakers).

A minority language may always experience a loss of speakers through language shift; 林衡道 1962 pointed out that in Yuanlin, Changhua county, there existed a group of Hakkas who had shifted from Hakka to Hoklo and termed them 「福佬客」hok(hô/ho̍h)-ló-kheh (cf. 陳逸君 2005). It is estimated that up to two million people may be Hokkien-speaking Hakkas who count as Hoklos according to their own self-identification.

On the other hand, there are Hakka speakers who arrived during the twentieth century. For example, more than 50,000 Chinese from Burma settled in Taipei, many in a dedicated quarter of the town (cf. 黃晶晶 101; Lu 2008), following Burmese buddhism and speaking Burmese with each other. They are visible in Mingalar Par (緬甸街; New Taipei) by the usage of the Burmese script in the public space. The Burmese and (YongDing) Hakka languages are spoken only by old people in an enclosed community; their offspring is acculturated to using Guoyu and Hoklo. But there are younger migrants who originate from Yunnan across Burma who joined the ‘Burmese’ community. Many of these migrants knew English, too, which helped them find jobs with foreign companies in earlier decades, and they sometimes moved to other continents for work migration.

Finally, Hakka was also spoken by Austronesian people (as L2), e.g. among the Saisiyat; the younger people now use Hoklo and Guoyu.

3.1 Peculiarities of Taiwanese Hakka

Taiwanese Hakka nowadays has its own accent and characteristics, but the varieties also originate from specific original dialects which are partly preserved; most people seem to speak Sixian or Hailu Hakka. It is therefore difficult to identify the various backgrounds of certain utterances. As one parameter, the frequent use of a diminutive -e is a characteristic of TW-HK (cf. 赖 2015). From an interview:

Differences such as TW-HK ⟨ziang⟩ and ML-HK ⟨liang⟩ ‘beautiful’ refer to different original dialects of HK (⟨jiang/liang⟩); TW-HK ⟨dong do⟩, ⟨qien⟩, ⟨ko i⟩, ⟨… ded hied …⟩ stands for ML-HK ⟨hau do⟩, ⟨lui⟩, ⟨ded⟩, ⟨… ded pied …⟩ ‘very many’, ‘money’, ‘can’, ‘fall off’, respectively. Words for new concepts are borrowed from Guoyu in TW, while being borrowed from English or Malaysian in ML-HK, e.g., TW-HK ⟨ceu sɨ⟩ ← CH 超级市场 ⟨Chāo(jí) shì(chǎng)⟩, but ML-HK ‘super(market)’.

TW-HK has ⟨de⟩ for 在zài ‘stay/at’, where ML-HK has ⟨coi⟩, but also ⟨de⟩ in some older speakers; however, sometimes TW-HK uses ⟨cai⟩ which seems to be CH influence; for instance, when asked to translate 我在家 ⟨wǒ zài jiā⟩, HK ⟨ngai coi/de vug ka⟩, ‘I am at home’, speakers produced the following varieties, with a variant ⟨lug ka⟩ for ‘house’:

(06)

在(屋)家。
ngai	cai	ga.
ngai	de	lug ka.
1	at	home.
I am at home.

Characteristics such as these are mostly dialectal differences with a few TW-HK preferences, as can be expected in spoken language data. Some other peculiarities reflect the influence of the other languages in the multilingual situation.

3.2 Influences on Hakka

Hakka is spoken in certain places in Taiwan, mainly by the older people, inside the communities. All speakers nowadays know Hoklo and Guoyu. As in all other minority language situations, Hakka is therefore influenced by their linguistic environment.

1. The speaker of the following examples is a 70+ lady from Zhutian, Pingtung living in Taipei together with her daughter and a son-in-law. Her native tongue is Hakka, her daughter also can use the language, however, she is speaking Guoyu most of the time, as her non-Hakka son-in-law resents the women talking a language he cannot understand. For this interview, she was addressed in Hakka and responded in that language. As is common for speakers of a minority language, her Hakka is lexically influenced by Guoyu and Hoklo, whereby Chinese loanwords are rendered in Hakka phonology, sometimes a Hoklo tone appears, and syntactic structures are often influenced by Guoyu. The next passages focus on her use of ⟨gong zuo⟩ and ⟨shang ban⟩.

ML-HK would simply say ⟨zo gung⟩ for ‘to work’ and would not accept ⟨song ban⟩ as a HK word, but understand it as CH 上班.⟨shàngbān⟩; The TW-HK dictionary (https://hakkadict.moe.edu.tw) does not list 上班.⟨song ban⟩, but has 下班.⟨ha ban⟩. The differences between the three languages being rather small, there seems to be a lot of convergence between the three ways of saying the words. The sentence structure seems to be following Guoyu which clearly is the dominant language in the multilingual competence of the speaker who is sometimes struggling to find the right variable from HK, HO, CH.—Here are two more mentions of ⟨gong zuo⟩:

Again, a part of the sentence is actually in Guoyu, and ⟨ngai e⟩ is a morphological change (simplification) in the possessive pronoun ⟨nga (ge)⟩. This phrase would be said in ML-HK as follows, with an established Hakka word for ‘Australia’ and a different word for ‘to work’:

2. Beside such conversations which already show heavy influences from CH and HO, speakers were asked to translate written Chinese sentences into Hakka, such as 这些游客全部都是来自中国的。‘These tourists all come from China.’ As one example, this resulted in the following response:

Again, as with the other test sentences, there is considerable influence from Guoyu on the utterance; the consultant translates HK ⟨ngin hag⟩ back into Guoyu ⟨you ke⟩; furthermore, she uses the copula 是 ⟨shi⟩ [sɨ] instead of ⟨he⟩ and the CH word 都 ⟨dou⟩ instead of HK ⟨du⟩. When this was pointed out to her, she insisted that ⟨dou sɨ⟩ is HK (which is CH 都是), and then proposed ⟨long sɨ⟩ (which is HO), finally ⟨long he⟩ (HO+HK). However, she correctly uses the HK word ⟨du⟩ in the following discussion:

In HK, one would expect ⟨he zung gued loi e⟩, the structure ⟨loi zɨ zung gued e⟩ seems to be an influence from the CH example sentence. These characteristics point to the possibility that the act of translating is too difficult or confusing for the elderly lady; she is competent in all three languages and can no longer identify the correct form when asked.

ML-HK would have ⟨le ge⟩ for ‘these’, ⟨rhiu hag⟩ for ‘tourist’ (loan translation from CH ⟨you ke⟩), and uses ⟨sung ma⟩ (← ML ⟨semua⟩) instead of ⟨quan bu⟩ for the lexeme ‘all’; furthermore, ML-HK would not use ⟨loi zɨ⟩ which is a loan from CH, but ⟨loi e⟩:

A high degree of language mixing and obvious dominance of Guoyu could also be seen in all speakers from the various regions; speakers are mixing Guoyu, Hoklo and Hakka in their speech.

3. The sample sentence 怎么会这样呢? ⟨Zěnme huì zhèyàng ne?⟩ ‘How could it be so?’ (cf. Malaysian English: ‘Why like that one?’) led to the following sentence (13), with a comparison to ML-HK in (14):

ML-HK has ⟨zo (mag)gai⟩ (lit. ‘do what’), ⟨iong ben/ban⟩, ⟨iam/iong men (iong)⟩ for ‘how’, and ⟨an iong⟩ for ‘like that’. The latter is different from the reported ⟨an nen⟩, it resembles HO ⟨an ne⟩, but is also phonetically close to HK. While TW-HK is influenced by CH and HO, ML-HK influences ML-CH: ⟨zo mo hui zhe yang zi de?⟩.

4. One more example is 今天停电。.⟨jīntiān tíngdiàn⟩, ‘There is / was / will be a blackout/power cut’, which was translated as follows (15), with ML-HK comparison (16):

Again, the TW-HK version is identical to the CH pattern; this sentence is also acceptable in ML-HK where it is, however, recognised as a calque from CH. Again, ML-HK (and the other Sinitic varieties) rather influence ML-CH by saying 今天没有电来 .⟨jin tian meiyou dian lai⟩, and Malaysian English ‘Today got no electricity’ – which are calques from (16). In accordance with typical language attitudes towards vernaculars and standards, for a ML-HK speaker, ⟨mau ten loi⟩ sounds more traditional than saying ⟨tin tien⟩.

To conclude, in a detailed analysis of TW-HK speech, we found many structural influences from CH and some lexical influence from HO which makes HK appear more like a phonological variety of CH. Many choices of the speakers appear to be only stylistic, as the structures are not unacceptable in other HK varieties (e.g. ML-HK). A careful examination of many utterances, however, proposes a strong preference for structures which are more similar to CH, the ‘roofing’ language (in the sense of Kloss 1952, 1967). At the same time, ‘traditional’ structures may be reinterpreted as an L (low) variant, due to the low prestige of the minority language.

3.3 Hakka in the media

Beside the interviews, Hakka media (which are conveniently available on Youtube) were watched. Of course, there are fully competent (older) speakers talking about their business, their craft, or other topics in the HK language. Often, however, educated younger (middle-aged) speakers show considerable influence from CH, or speak a mixed form of HK and CH, as in the following statement:

Such linguistic behaviour may be explained by the functional division of the varieties, with standard languages being connected to acrolectal language use (public speaking). When explaining facts on TV, it is a natural move to switch to the acrolectal variety. The same happens, for instance, with dialect speakers in Austrian dialect media programs—the speakers create a form of standard/dialect mixed language (cf. Vollmann 2021) which heavily relies on elaborated syntactic structures and non-dialectal words taken from the standard language.

3.4 Intergenerational transmission

Nowadays, Hakka is part of the school curriculum in Taiwan. In this research, consultants were asked about their subjective views on language transmission into the younger generations. All wished to transmit the language, however, the language was often no longer used much, and language preservation proved to be particularly difficult with the younger generations. It seems that the younger people are often ‘weak speakers’ (as is observed at stages of language decay). In the following passage, a woman (60+) from the north-west explains how she taught HK to her children, thereby recreating the bilingual situation in her report through code-switching:

This situation refers to times 30 years back. The mother wanted the children to speak Hakka, however, from the reported necessary effort, it can be understood that Hakka was not the first language of the children, but was transmitted as a symbolic value, as an identity. The language is optional and lacks an exclusive domain in which it is to be used. It can be deduced from this report that the children (now adults) will be weak speakers, not fully competent language users.

3.5 Memories of discrimination

The discrimination of the minority languages during the KMT rule (and before these times) seems to have left a lasting impression on the people, as it was often mentioned. The new situation of public recognition is appreciated by all speakers. As an example, a Xinzhu speaker summed it up as follows; as in the previous examples, the speaker switches to CH for longer stretches of the text.

The reasons for intergenerational language shift is often thought to be related to the oppression of the language; however, the linguistic disruption may relate more to the advent of mass media (which are effectively able to spread and promote superordinate languages), urbanisation and ‘interethnic’ communicative needs (e.g., in interlinguistic intermarriages, as opposed to traditional endogamous practices); for Hakka, these are the conclusions of Wang and Peng (2018), and the consultant making the above statement refers to the symbolic power and importance of Hakka media presence which provides overt prestige to the smaller language.

3.6 Promotion of the language and standardisation

In order to create textbooks, it is necessary to standardise the language. Spreading a standard language means that most speakers need to give up their personal variety and adapt to new linguistic norms which may create feelings of exclusion at the level of dialects and accents. In a TV discussion, one speaker made the following statement (in Chinese; only English translation given here):

This is a recurring problem in minority language preservation efforts: The speakers had to adapt to another standard language already, and then, in a second step, need to adapt to the standardisation of their spoken vernacular. In this way, every standardisation is a situation of language change, and that can prove to be harmful to an already weak language.