1 Introduction

Following the exponential expansion of the Internet and tremendous technological advancement since the 1990s, both the quality and efficiency of electronic communication has been significantly improved. Impact of this development can be felt in many facets of everyday life, for instance, calling someone or sending images – still or animated – over a distance is now much easier. The battle among transnational consortiums and providers of electronic gadgets to win over the hearts of end-users worldwide has been driving one technological breakthrough after another, with enhanced user-friendliness being the ticket to success. As a correlate of English being used as an international lingua franca, for example in marketing campaigns, one linguistic consequence of this development is the growing visibility, and versatility, of the letters i and e. Following the example of e-mail, an initial attempt to replace the longer electronic mail, the transnational corporate giant of e-gadgets, Apple Inc., decided to name its new products strategically with i: iPhone, iPad, iPod and iTunes. In effect, i has become a one-syllable substitute for internet, in the same way that e has taken the place of electronic where linguistic parsimony matters to the speakers/writers and their interlocutors. Among the more recent neologisms involving these two morphemes are e-learning, e-channel, iMedia, and iTouch (hyphen increasingly dispreferred, except when parsing may be a problem, e.g., e-gadget and e-channel).

Since advertising space is costly, marketing professionals are keen for their target clients to remember their products. It is therefore not surprising that marketing experts seek every possible means to get their message across concisely, hence the appeal of monosyllabic words and morphemes like e- and i-, but also compounds and abbreviations such as Facebook and the now defunct ICQ (‘I seek you’). A quick search of IT applications and gadgets – IT itself being another revealing example of this trend – yielded the following functions and product names: app (clipped from application), Chat, Dontalk, Snapchat and Talk. Long expressions tend to be abbreviated (e.g., GB, OS, SMS), especially if this results in an acronym pronounceable as a syllable (e.g., RAM and WAP). As of the time of writing, Samsung promotes a new smartphone function called kill switch, which allows the owner to deactivate the device if it is lost or stolen. All this suggests that monosyllabic English words have a strong appeal among marketing professionals of IT products.

In a corpus of texts collected from written Hong Kong Chinese newspaper columns (Li et al. 2014), which are characterized by adherence to vernacular-style and plenty of Cantonese-English code-mixing, we found a large number of monosyllabic English words (MEWs) which, following Muysken (2000), are mostly insertions into Hong Kong Written Chinese (HKWC, Shi 2006). This study is an attempt to account for the preponderance of MEWs in Hong Kong Cantonese and Written Chinese.

2 Theoretical framework in this study

In light of varying terminologies in research on language contact, we find it useful to adopt a set of terms that would subsume what is elsewhere referred to by various scholars as “code-switching”, “code-mixing”, “code-alternation” and “borrowing”, among others. Toward this end, we will follow Clyne’s (2003) terminological delineation as follows:

A ‘transfer’ is an instance of transference, where the form, feature of construction has been taken over by the speaker from another language, whatever the motives or explanation for this. ‘Transference’ is thus the process and a ‘transfer’ the product.

(Clyne 2003: 76)

Transfer or transference may take place at different levels – lexical, semantic, phonetic/phonological, prosodic, tonemic, graphemic, morphological and syntactic, and any combinations of these (Clyne 2003: 76). They free us from a concern, to what extent the transfers in question have been integrated (partially or fully, the latter being more like borrowings comprehensible to monolingual speakers), or are unintegrated into the recipient language, from ephemeral “nonce loans” that would fail to catch on as a result of low social acceptability in society (see Onysko 2007: 37–38 for a critical discussion), to frequently occurring insertions whose pronunciation approximates that of the source language. This point is especially important as much of the data reported in this study comes from written sources, where the English words inserted into Chinese texts cover a wide range, from “nonce loans” to frequently-occurring switches.

At the heart of Clyne’s (2003) explanatory framework is facilitation, a construct which he considers more precise and less controversial than the earlier concept of “triggering” (p. 162; cf. Clyne 1967, 1980). Clyne (2003) shows convincingly how, in a migration settlement context like Australia, different community languages of European (including via Latin America such as Italian and Spanish) and Asian origin undergo various linguistic changes under more or less the same language contact conditions, with English as the common nexus. Rather than universalist constraints on code-switching (e.g., Equivalence Constraint, Free Morpheme Constraint, Government Constraint, Conjunction Constraint), Clyne (2003: Ch. 3) demonstrates that code-switching data presented in previous models and analyses are more productively seen as strong tendencies resulting from one or more facilitating principles that account for “transversion” (term he prefers to code-switching) at the lexical, tonal/prosodic, and syntactic levels.

With the help of a wealth of data sets involving bilingual and trilingual speakers of various community languages as diverse as German, Dutch, Croatian, Italian, Spanish and Vietnamese, Clyne presents solid evidence showing how unintegrated transfers such as proper nouns in the embedded language, and bilingual homophones which have an identical or similar pronunciation as a matrix language counterpart facilitate transversion (Principle 1, “lexical facilitation”, pp. 162–175). Likewise, drawing on bilingual data from Mandarin and the Vietnamese communities, Clyne shows a similar mechanism at work at the tonal-prosodic level, where the matrix language “lexical items in a tonal language whose tone is identified with the pitch and stress of the non-tonal language in contact are liable to facilitate (though not necessarily cause) transversion” (p. 175; Principle 2, “tonal facilitation”, pp. 175–177). At the syntactic level, where syntactic structures occur or where contact-induced syntactic convergence has taken place, the points of convergence are often perceived as “triggers” or sites of switching, which allow the speaker to proceed in any of the languages in the language dyad or triad, thereby facilitating transversion (Principle 3, “syntactic overlap/transference/convergence [secondary facilitation]”, pp. 177–179). One instructive illustration involves “multiple transference” of the collocation for lunch, uttered by a second-generation German-speaker, in what Clyne calls anticipational (as opposed to consequential) facilitation of transversion:

Here, we see evidence of syntactic convergence to English, as shown in the choice of auxiliary haben (haben…gegangen) instead of sein ‘be’ (sind … gegangen), and the preposition aus, which was probably triggered by phonetic similarity with English out in out for lunch. It is termed “multiple transference” since out for lunch is not a single-word switch but a high-frequency collocation (cf. embedded language island, Myers-Scotton 1993). Clyne (2003) notes that syntactic overlap is only partial because the discontinuous structure, aux+participle, is maintained (i.e., it would have been total syntactic transference if the speaker had said Wir haben gegangen aus for lunch). Clyne comments that syntactic convergence and transference “function like other forms of convergence and overlap due to perceptual identification between items in the two languages as a potential facilitator of switching” (p. 178, emphasis in original). Perceptual salience is thus postulated as a theoretical premise behind facilitation. We will come back to this point later.

Additional compelling evidence comes from three data sets obtained from trilinguals: Dutch-German-English (DGE), Hungarian-German-English (HGE), and Italian-Spanish-English (ISE) (Clyne 1997). Under “Trilingual convergence” (Clyne 2003: 105–109), Clyne demonstrates that a strong “tendency for trilinguals to extend to the third language a feature shared by two of their languages is found at the lexical, semantic, syntactic, morphemic, phonological/prosodic levels” (p. 105), a feature characterized as “interlingual identification based on correspondences between two of the languages” (Clyne 1997: 95). Thus at the level of phonology, instead of pronuncia in homeland Italian, ISE trilinguals with Italian as L1 or L2 would pronounce it as pronunciazone, probably under the joint influence of Eng. pronunciation and Span. pronunciación. Similarly, homeland Ger. provinziell gave way to provinzial, aligning with Eng. provincial and Dut. provinciaal (Clyne 2003: 95). Where cognates exist, the prosodic pattern shared by two languages may exert pressure on the third. This is apparently why an HGE speaker pronounced the word for accent in Ger. as [’ɛksɛnt], which deviates from homeland Ger. Akzent [ak’tsɛnt] (compare: Eng.: accent [’aeksent]; Hung.: ékezet [’e:kɛzɛt], p. 108). Morphologically, among the DGE, homeland Ger. fused comparatives (e.g., normalste) gave way to analytic comparatives (e.g., meist normale; compare: Dut. meest normale; Eng. most normal, p. 107). At the syntactic level, to express the meaning ‘to like’, the English of ISE speakers shows a word order preference which is shared by Italian and Spanish, e.g., ‘The garden like it my wife’, where the experiencer (here, ‘my wife’) is placed at the end of the clause rather than being thematized as subject (compare: Il giardino piace a mia moglie; El jardin le gusta a mi mujer, pp. 106–107). Apart from the variety and sheer amount of solid evidence from different language dyads and triads with English as the common nexus, Clyne (2003) demonstrates convincingly that transference, often manifested as transversion (code-switching), is often motivated by overlaps or similarities in the linguistic subsystems of the languages in contact: lexical, tonal/prosodic, and syntactic.

After reviewing various language processing models to date, Clyne (2003: Ch. 6) draws implications and concludes that:

Each language constitutes a network. The networks are connected through items that are linked because such items (lexemes, tones) are (perceived to be) part of, or employed in, more than one language. Thus, using any item from a particular network is sufficient to activate the network (language) of which it is part or with which it is identified. There is also a secondary facilitation, where activation according to a similar procedure is further assisted by overlap in, and convergence of, grammatical structures that are the same. Transversion may be facilitated by anticipating a trigger-item or in consequence of one. (Perceived) overlaps in the lexicon and also in the prosody and syntax of the languages function as gateways to another network.

(Clyne 2003: 211–212)

Clyne’s hypothesis of tonal/prosodic facilitation (Clyne 2003: 175–177) is based essentially on the data sets collected from second-generation and young first-generation Mandarin-English bilinguals and first-generation Vietnamese-English bilinguals. In both cases, a strong correlation was found between the pitch level of the words immediately before a switch to English (Viet.: 85.46%; 33.41% in Tone 1 or high pitch, and 51.95% in Tone 2 or 3 mid pitch; Mand.: 96.49% of switches came after fourth (53, falling), half-third (35, falling then rising; and neutral). Clyne (2003: 175) argues that “[w]ords with these tones bring speakers into the tonal range which is also possible in English, i.e., which overlaps in the two languages”. This appears to facilitate transversion and transference from Vietnamese and Mandarin, respectively, into English.

Clyne’s notion of perceptual salience, the modus operandi behind facilitation of transference across languages, arguably underlies the theories in a few other prominent language contact studies. For instance, perceptual salience, which plays a crucial role in Field’s (2002) critical examination of hierarchies of borrowability based on the morphological typologies of the languages in contact, is subsumed in two complementary principles: Principle of System Compatibility (PSC) and Principle of System Incompatibility (PSI), which he applies to account for the extensive borrowing of Spanish into Modern Mexicano (Nahuatl). Field’s (2002) findings are neatly summarized by Comrie in the foreword as follows:

the borrowing language’s morphological typology – whether it is isolating, agglutinating, or fusional – will constrain the possibility of borrowing features from another language. An isolating language can borrow neither agglutinating nor fusional morphology. An agglutinating language can borrow agglutinating, but not fusional morphology. A fusional language can borrow both agglutinating and fusional morphology.

(Comrie 2002: x)

Facilitation mediated by perceptual salience is also clearly at work in graphic borrowing, which represents the focus of Hansell’s (2002) functional analysis of lexical borrowing. Hansell observes that graphic borrowing “requires not only that both languages be written but that they also share a common script. English can borrow graphically from French but not from Japanese while Japanese can borrow from Chinese but not Arabic, etc.” (p. 156, emphasis in original). Two of Hansell’s illustrations of graphic borrowing or transference are particularly instructive (2002: 157–158). First, the morpho-syllables 社 and 會, which had been borrowed from Classical Chinese into Japanese earlier, were combined to form a Japanese neologism 社會 (shakai) in the late nineteenth century to render the western concept of ‘society’ (written as 社会 in both kanji and simplified Chinese). This bisyllabic word was subsequently re-borrowed into Mandarin (shèhuì) for that modern meaning (compare borrowing of morphemes written in a different writing system: Ger. Automobil < auto- from Greek αυτο- ‘own, self’+Latin mobilis ‘moveable’; Eng. television via Fre. cognate télévision < tele- from Greek τηλε- ‘far away’+Latin vision- ‘sight’; Hartmut Haberland, p. c.). Second, Eng. boycott was transliterated in Hong Kong Cantonese as 杯葛 bui1 got3, ^[1] but the two morpho-syllables were later borrowed into Mandarin, albeit pronounced differently: bēigĕ. These examples lend strong support to Hansell’s view that the sharing of a common script facilitates or “expands the possibilities for interaction between languages, especially lexical borrowing” (2002: 154).

We believe Clyne’s (2003) twin postulate of perceptual salience and facilitation lends itself very well to explaining a large number of MEW (monosyllabic English word) insertions in Hong Kong Cantonese, which is arguably due to a community-wide perception of MEWs being functionally akin to Cantonese morpho-syllables. This claim logically entails linguistic evidence of the perceptual salience of the Cantonese morpho-syllable. Below, we will first present a list of MEWs separated by word class in our 1990s corpus of written data collected from informal sections of the Hong Kong Chinese quality press (Li et al. 2014). Then, to contextualize how MEWs are used in Hong Kong Chinese newspapers, we will outline the findings of a survey of reader response to one comic strip containing MEWs. Our key research question is: Roughly one in five English words inserted into Hong Kong Cantonese is monosyllabic, suggesting that MEWs are treated on par as Cantonese morpho-syllables. What linguistic evidence is there to facilitate transference?

3 MEWs in Hong Kong Cantonese mixed code: Corpus data in the 1990s

The preponderance of MEWs in Cantonese first came to our attention when processing data consisting of mainly Hong Kong Chinese newspaper columns collected in the mid-1990s – when the Internet was just beginning to become popular, and Chinese word-processing was technically rather difficult. The size of the corpus is about 600,000 Chinese characters. Data came from three main sources: Hong Kong Economic Times (香港經濟日報), Hong Kong Economic Journal (信報), and Ming Pao (明報). These sources may be broadly characterized as “quality press” (as opposed to “popular press”) material. A column typically contains no more than 400 characters, and the topic is usually thought to be of interest to readers (e.g., personal commentary on a recent news story) or within the columnist’s expertise. The columns, in the form of clippings, were collected randomly as they came to the attention of the first author; they were selected usually because there were one or two points of linguistic interest, of which one had to do with the insertion of some English element in Chinese. The clippings were sorted according to their points of interest. They were inputted into a database only recently. Since columns and other soft genres like adverts, cartoons and infotainment news stories are usually outside the scope or target of large-scale databases such as Linguistic Variation in Chinese Speech Communities (LIVAC, Tsou et al. 2011; http://www.livac.org/), our data is qualitatively different from mainstream Chinese databases in that, by virtue of text type and content, Hong Kong columnists are usually able to draw on vernacular-based norms more freely without meeting with editors’ disapproval. This makes for an interesting writing style, and space, where vernacular-based writing proliferates (Snow 2004). Such a writing style has a precursor dated from the 1950s known as saam1 kap6 dai2 (三及第, origin related to “imperial examination’s three top honours”, Cheung and Bauer 2002), where modern Chinese is blended creatively, sometimes unexpectedly and humorously, along with elements from classical Chinese and Cantonese (Wong 2002). This is the background against which insertion of English words of various lengths is seen by Hong Kong Cantonese speakers and readers as perfectly natural, which is rather different from prevailing norms for hard news stories.

With the help of two research assistants, who were instructed to proofread each other’s typed drafts by cross-checking the original clippings, all the monosyllabic English word (MEW) types and tokens were entered into an Excel file. Table 1 gives an overview of the word types listed alphabetically according to word class (Table 1; those bolded in red appear in five different texts or more).

Below are six examples of MEWs, two each from three word classes (N.: band, Line; V.: call, Talk; and Adj.: cool, HIGH – upper or lower case as in original), showing how they appeared in our Chinese corpus.

As shown in Table 2, the ratio of monosyllabic and polysyllabic words is 18: 82 (or 1: 4.56), suggesting that roughly one in four to five inserted English words, including letter words in the corpus is monosyllabic (acronyms pronounceable as single syllables like RAM and WAP are treated as monosyllabic; abbreviations are polysyllabic, e.g., CD and DJ are disyllabic; IBM is trisyllabic).

As illustrated in (2)–(7), the vast majority of MEWs (and polysyllabic English words, as in MELODY (2) and HAPPY (7)) occur as unintegrated single-word insertions – bare nouns, verbs or adjectives (Muysken 2000). There is no question that the matrix language in our corpus is Hong Kong written Chinese (HKWC) which, following Shi (2006), is characterized by considerable influence from Cantonese lexis and syntax, and from English to some extent. By virtue of these influences, readers tend to have the impression that HKWC follows the norms of speech (or vernacular style) rather than Mandarin-based standard written Chinese, which is the expected style for writing hard news stories, editorials, feature articles and other formal genres in public discourse.

4 Survey of reader response to a comic strip containing MEWs

As cartoons or comic strips typically feature social interaction in speech, it is not surprising that MEWs also figure prominently. One recent example consists of four panels, each of which contains one or more MEWs, as in (8) (text only; the four panels in the original appeared vertically, see Appendix 1 for the original).

(8)

三叔 Sm@rt P@use功能：唔望屏幕時video 會自動暫停播放……以上功能係咪真係咁 smart 呢？	試想想當你睇恐怖片嘅時候…	三叔驚你 miss 左…就幫你按 pause…	…等你擰返過嚟先繼續 play 俾你睇…
Panel 1Third Uncle’s Sm@rt P@use function: [if you] stop watching the screen, the video will automatically stop playing. Is the above function really that smart?	Panel 2Imagine when you are watching a horror film…“Wow! This part is ultra-horrifying!!”	Panel 3“Can’t fall asleep if [I] watch. [I] won’t watch!!” Third Uncle fears that you might have missed [that, and] helps you to press pause.	Panel 4When you turn your face to the screen again, [Third Uncle would] continue to play and let you see [it].“RRRAAAWWWRRR”

(‘三叔有幾 smart?’, by Summer&Muu, Headline Daily, 2013–4–8, p. 14)

The first two syllables in the title ‘三叔有幾 smart?’ (Saam1 Suk1 jau5 gei2 smart? ‘How smart is Third Uncle’) is an unmistakable allusion to Samsung, the brand product being thematized (trade mark in Chinese: 三星, saam1 sing1). Samsung is one of the most popular brands for mobile phones in Hong Kong; it is often jokingly referred to by the kinship term Saam1 Suk1 (三叔 ‘Third Uncle’). By tracking the user’s eye movements, apparently the new model of e-gadget would pause if the user is detected as not viewing, and would continue once the e-gadget user’s attention is restored. The cartoonist makes fun of this new function. Of interest to us is the fact that five MEWs are used in addition to video, of which three are verbs (miss, pause, play), one adjective (smart) and one noun (part).

To ascertain whether MEWs such as these are commonly used among Cantonese-English bilinguals in Hong Kong, a survey was carried out with about 400 students studying at the Hong Kong Institute of Education (84%) and the Hong Kong Polytechnic University (13%) from October to November, 2013. Aged between 18 and 21, the respondents were mainly first-year students (94%, the rest second-year). The gender ratio, female-male, was about 3:1. In terms of their majors, 36% indicated a discipline in the humanities, 35% in business, 23% in social sciences and 6% in science. Nearly two-thirds (74%) of the respondents were born in Hong Kong and the rest (26%) were born outside of Hong Kong. Besides, 82% of all respondents indicated that Cantonese was their first language.

An e-questionnaire was used as instrument to tap into our students’ awareness to what extent such MEWs were familiar to them, and whether they themselves would use them (Appendix 2). It consisted of 12 multiple choice questions. A small-scale pilot was carried out with over 10 students before the actual survey, and their feedback was used to fine-tune the wording of the questions. The survey was conducted in the respective lessons of the authors at the beginning or the end of our classes. The cartoon was first projected on the screen and the purpose and content of the questionnaire were briefly explained before the survey started. The respondents were asked to key in their choices via their mobile devices such as smartphones, tablets and netbooks. It took approximately 15 minutes to complete the e-questionnaire. All students were requested to send their completed questionnaire to a designated e-address. A total of 392 valid questionnaires were successfully collected.

After analysing the data using descriptive statistics, two main findings came to light. Firstly, the result shows that most of the respondents (74%) were familiar with the discourse and language use patterns of comic strips in Hong Kong Chinese newspapers such as the one in (8). The use of MEWs in the comic strip (8) did not present any literacy problems, except for the Mandarin-dominant respondents (19%) from mainland China. About half of them mentioned the reason(s) why they did not understand the expressions in the comic strip (e.g., don’t speak/read Cantonese; words are difficult). Interestingly, Cantonese was the cause of literacy problems rather than MEWs. Secondly, 95% of the respondents were able to provide examples illustrating how they themselves would use those MEWs, while 82% replied that the MEWs in the comic strip sounded natural to them in both spoken and written forms. Further, some 77% indicated that they would use those MEWs when communicating in Cantonese in their daily life. In short, the survey results provide empirical evidence that MEWs like those used in the comic strip (8) are indeed commonly found and used by Hong Kong Chinese students in their social interaction with others. The result of the survey suggests that for educated Chinese bilingual speakers, MEWs constitute an additional pool of linguistic resources for meaning-making. In what follows, we will briefly outline the context of Cantonese-English contact, the types of transference reported in the literature and how tightly they are integrated into Cantonese, before presenting evidence in support of the Monosyllabic Salience Hypothesis (MSH) in Cantonese.

5 The context of Cantonese-English contact in Hong Kong

Until 30 June 1997, Hong Kong was a British colony for over 150 years. Since the colonial period, in the formal curriculum English is introduced from primary to the end of secondary levels. In practice, English has been an integral part and, in some residential areas, a selling point of pre-primary institutions to attract pupils. Consequently, all children growing up in Hong Kong learn their ABC and some basic English vocabulary before formal schooling starts at primary. Under the nine-year compulsory education policy (Grade 1–Grade 9), which has been extended to 12 years from September 2012 (Grade 12), all secondary school-leavers have learned English for over 10 years, with those students (roughly 30% of every cohort) receiving English-medium education having considerably more exposure to English. Hence young people in Hong Kong have quite a bit of English as a meaning-making resource in addition to Cantonese. At the tertiary level, in general, English is an important subject and medium of instruction (MoI) for most disciplines in all of the eight government-funded tertiary institutions.

There is some evidence that when conversation topics related to school work and university life are invoked, Hong Kong students find it difficult not to code-switch to English. Li (2011) invited 43 students in Hong Kong and 65 in Taiwan to take part in an experimental study. At a briefing, students were asked to use only their dominant local language for one day (Cantonese in Hong Kong, Mandarin in Taiwan) and to report on “rich” experiences that happened to them. With the help of an e-template for jotting down ‘who speaks what to whom and when’ and expressions that they wanted to use but could not, students wrote a diary up to two pages reflecting on their feelings, giving details of one or two rich events that left them a deep impression. These diaries were then collected and salient events reported were extracted as stimulus material for more in-depth discussion at a two-hour focus group interview attended by students studying the same discipline (cf. Li and Tse 2002). The results revealed evidence of a “medium-of-learning effect” (MOLE); when reference is made to subject-specific content, technical concepts learned in English or school events (for Hong Kong participants in particular), they felt seriously inconvenienced by not being able to use English. As one would expect, the MOLE effect is more marked among Hong Kong students than their Taiwanese peers, probably because English is used less extensively as MoI in Taiwan.

6 Types of transference in Hong Kong

With Clyne (2003), we regard words or linguistic features from English as instances of transfer, irrespective of whether they are invoked following English pronunciation norms and traditionally analyzed as code-switching (some never occurred again – nonce loans), or closely integrated into the recipient language (i.e., loanwords). The process and types of transference may differ. Broadly, depending on the linguistic level, we can distinguish between phonological (including prosodic), lexical, syntactic, semantic, and graphemic transference. All of these types of transference have been reported in Cantonese-English contact research in Hong Kong, including Cantonese-English interaction among university students (Chan 2003, 2009a; Li and Tse 2002), Chinese newspapers (Li 2000a, 2000b), Canto-pop music (Chan 2009b), and transference of the “Sino-alphabet” (Hansell 2002) or letter words (i.e., letter names and abbreviations, Cheung and Bauer 2002; Li 2000a, 2000b) across different print and multimedia genres.

The frequent contact between English and Cantonese in Hong Kong has brought about the integration of a large number of English loanwords in Hong Kong Cantonese (Wong et al. 2009). These loanwords have led to an influx of “loanword syllables”. According to Bauer and Wong (2010), there exist a total of 78 such syllables in Hong Kong Cantonese, an expansion from 40 in 1997 to 49 in 2006 as documented in Bauer and Benedict (1997) and Bauer (2006) respectively. These loanword syllables are “non-occurring syllables or unused syllables which represent both accidental and systematic gaps in the syllabary” (Bauer and Wong 2010: 7). For example, the loanword syllable [wɛn⁵⁵], from the English word van, is commonly used in phrases such as [hʊŋ²¹ wɛn⁵⁵] and [lʊk²² wɛn⁵⁵] (literally ‘red van’ and ‘green van’, referring to two types of mini buses in the public transport system). The syllable [wɛn⁵⁵] did not exist in the Cantonese syllabary, which means that although the individual vowel [ɛ] and consonants ([w] and [n]) are phonemes of Cantonese, the combination of all three as [wɛn⁵⁵] was not used in any existing Cantonese word, until the loanword [wɛn⁵⁵] appeared. It figures prominently in the title of a recent Hong Kong film, which is often referred to in infotainment stories elliptically as《糹工 VAN》(hung4 wen1). Similarly, the loanword syllable [k^hɔ⁵⁵] used to be an unused syllable. It originated from the English word call and is only used in the loanword noun phrase [k^hɔ⁵⁵ gei⁵⁵] (‘beeper’ or ‘pager’) or as verb, as in [k^hɔ⁵⁵ ŋɔ²³] ‘call me’.

What is more intriguing is that some of these new loanword syllables do not conform to what is traditionally known as “the labial-dissimilation constraint”. This rule from Cantonese phonotactics predicts that labially articulated vowels do not combine with labially articulated final consonants (Bauer and Wong 2010: 18), but this rule may be relaxed in baby talk, onomatopoeia expressions and loanwords. The three new loanword rhymes that show such a feature are [ɔm], [œm], and [ɔp]. The rhyme [ɔm] can be found in the loanwords [fɔm⁵⁵] (for ‘form’ as in an application form) or [wɔm⁵⁵] (for ‘warm’). The rhyme [œm] is used in loanwords such as [fœm⁵⁵] (for ‘firm’, or ‘to confirm’) or [pœm⁵⁵] (for ‘permanent’). The third rhyme [ɔp] can be found in loanwords like [pa⁵⁵ t^hɔp⁵⁵] (for ‘bra top’) or [tsɔp⁵⁵] (for ‘job’).

Research since the 1980s has shown that individual English content words and expressions are often “mixed” into Hong Kong Cantonese and informal written Chinese (Chan and Kwok 1990). Consonant with previous research in other language-contact settings, nouns are more commonly transferred than verbs and adjectives. Most of the reported cases of lexical transfer are insertions (Muysken 2000), but a few verbs would trigger lexico-syntactic transference (Clyne 1991, 2003), where the V-O pattern in English is preferred to the verb-specific O-V pattern in Cantonese. For example, compared with (9), (10) is far more frequently used:

Today, since upload has been calqued as soeng6 zoi3 (上載) and become widely accepted, they are treated very much like lexical variants depending on the speaker, interlocutor(s), and the context at large. Similar examples are commonly found with other causative verbs such as undo, unstar, update and upgrade (Li 2000a: 317). The “code-mixed” pattern as in (10) is preferred probably for its simpler valency relations (Clyne 2003: 114), much like in Australia German-speaking migrants gradually shifted to remember(n) in their community language in place of their homeland German counterpart sich erinnern an (reflexive pronoun sich, preposition an, plus accusative). Similar evidence has been found among second-generation Croatian-English bilinguals, where remember is preferred to homeland Croatian sjeam+ACC, which requires a reflexive clitic se, plus the marking of the agent and patient in the nominative and genitive, respectively (Hlavac 2000; cited in Clyne 2003: 114). As Clyne has pointed out, such cases of lexico-syntactic transference may be due to the bilingual’s attempt to maintain grammaticality.

Semantic transference involves mapping the meaning of an English morpheme onto an existing Cantonese morpheme, resulting in expansion in the latter’s semantic scope. For instance, Shi (2006) has found that under the influence of English, the Cantonese verb fan1 hoeng2 分享 ‘share’ may be used in reference to negative experiences, a usage which is considered anomalous in Mandarin (fēnxiăng). Similarly, since the 1990s the morpho-syllable 芒 (mong1, as in mong1 gwo2 芒果, ‘mango’, e.g., ceng1 mong1 青芒 ‘green mango’; graphic variant mon) has been popularly used to refer to the ‘monitor’ (of a computer), resulting in semantic expansion or extension of 芒.

As for graphemic transference, perhaps the best-known example is the letter D (di1, possessive marker or nominalizer in Cantonese), which in informal written Cantonese is often preferred to the homophonous but considerably more complex (including in electronic communication) Chinese character 啲. Other English letters borrowed into Hong Kong Cantonese include B, D, E, K, T, Q, and X (Cheung and Bauer 2002). Still other letters have been found in more recent research (e.g., U for ‘university’, Chan 2011). As electronic communication gradually became more powerful and convenient to use, plenty of innovative examples of graphemic transference in Roman script may be found in all sorts of social media like ICQ and more recently facebook, twitter and whatsapp, for example, Romanized adjectives like hea (he3, ‘laid-back’ or ‘tardy’), chok (cok3, ‘suffocating’) and chur (coe2, ‘hard pressed for time’) (see Section 7.5). These are all Cantonese morphemes, whose written forms are clearly modeled on English (compare: heavy, choked and church). While it may not be easy to trace their origin, it seems safe to assume that such Cantonese morphemes written in Roman script first caught on in speech among young Cantonese-English bilinguals, before being popularized in their e-communication.

There is also some evidence of syntactic transference in Hong Kong Written Chinese (HKWC, Shi 2006). Thus, the pervasive use of the English structure ‘it is time (for someone) to do something’ has led to a widely used Anglicized structure in Hong Kong Chinese media, as in clauses beginning with si6 si4 hau6… (是時候…, ‘it is time to…’), sometimes with a locative expression occupying the subject position, which is not admissible in standard Chinese. For instance:

With regard to the crucial question, whether the patterns of lexical transference are more appropriately seen as code-switching or borrowing, Clyne (2003: 142–152) proposes three parameters to probe into the extent of integration: types, degree and stability. The types of transference outlined above suggest a fairly high level of integration; corroborative evidence may also be found in the two other parameters. In what follows, we will illustrate how tightly knitted MEWs are in Hong Kong Cantonese.

Regarding the degree of integration, sound evidence of close integration may come from the conversion of the source language word into a verb using the regular verb morpheme (e.g., Ger. -ieren: farmerieren ‘to farm’, literally ‘to farmer’; gärtnerieren ‘to garden’ among migrants in Australia, Clyne 2003: 111), or the use of plural or case marking in the recipient language. Cantonese being an isolating language, no such morphological evidence may be found; unintegrated English verbs and adjectives transferred into Cantonese typically appear as bare forms (i.e., free from any tense, person/number, or comparative/superlative marking). A high-frequency noun like fans (often capitalized as Fans) appears to carry the plural morpheme, whereas it is sometimes found in contexts where the number of ‘fans’ being referred to consist of just one person (jat1 go3 Fans, ‘a fans’; compare: Danish en hotdogs, ‘one/a hot dog’ and Ger. Keks (singular) < Eng. cakes, Hartmut Haberland, p. c.). The singular form is dispreferred probably because the word is invariably realized in speech as a bisyllabic word: fen1 si2 (often written as FAN 屎).

Nevertheless, there is ample evidence that English verbs and adjectives are commonly suffixed by aspect markers such as perfective -zo2 (e.g., out–zo2 ‘[the ball] is out’), experiential -gwo3 (e.g., pass–gwo3 ‘has passed’), progressive -gan2 (e.g., run–gan2, ‘is running [the program]’), and tentative aspect marked by verb reduplication, typically to soften a request (e.g., m4 goi1 bong1 ngo5 check check aa1, ‘please check [it] for me’). English adjectives may be affixed with the comparative marker di1 (ni1 go3 cheap di1, ‘this is cheaper’) or -gwo3 (nei5 fit gwo3 ngo5, ‘you are fitter than me’) and the superlative (nei5 zeoi3 friend, ‘you are the most friendly’). These illustrations show that MEWs can be inserted into Cantonese like other Cantonese morphemes. Also, the word class of an MEW insertion may shift, as when friend is used like an adjective.

Further evidence of close integration involves polysyllabic English words (PEWs) in syntactic frames such as the A-not-A (yes-no question) structure. For instance, in the Hong Kong white-collar workplace, it is very common to embed the adjective available in this structure (12) (see Section 7.4).

Close integration of MEWs is also evidenced in the common practice of bilingual punning, which is very common in Hong Kong advertising language. MEWs such as fun, high and phone are often blended into Cantonese to create additional semantic nuances (Li 2000a; cf. Li and Costa 2009; see Section 7.3). There is thus strong evidence that for Cantonese-English bilinguals in Hong Kong, MEWs constitute a pool of additional resources with regard to the rhetorical function of punning. Notice that in Cantonese, a syllable is the minimal segmental unit of punning, whatever the linguistic resource. Thus in the wake of the Edward Snowden affairs, a widely reported sub-syllabic pun like ‘Yes, we scan’ by European protesters, in mockery of President Obama’s election campaign slogan ‘Yes, we can’ during his first official visit to Germany in June 2013, cannot be replicated in Cantonese (or Chinese) because Chinese, and Japanese kana, “both permit phonological processing (…), but not of units as small as a phoneme” (Hansell 2002: 152–153).

As for the third parameter, stability, Clyne (2003: 146–147, 210) notes that speakers belonging to tight social networks tend to be more conservative toward maintaining community language norms, an attitude less adhered to when speaking to others. In Hong Kong, thanks to the implementation of 9-year compulsory education for children from primary level (recently extended to 12-year since September 2012), the level of English literacy among young people is quite high. Further, as Hong Kong is in the forefront of and testing ground for new IT gadgets, the use of electronic media is widespread and very popular among young people, from (earlier) ICQ and MSN to facebook, twitter and whatsapp. The ratio of cellular phones per person is also among the highest in the world. Thus linguistic innovations tend to catch on quickly, subject to their perceived trendiness and popularity among members of various networks of bilinguals big and small across different age groups. Thanks to the convenience afforded by various forms of social media and the many channels through which code-switching could be produced, displayed or shared, it usually does not take long for English words embedded in code-switching – pervasive and ubiquitous Hong Kong wide – to stabilize and become linguistic borrowings.

7 Evidence of monosyllabic salience in Cantonese

7.2 The average word length of Cantonese

Luke and Lau’s (2008) findings are consonant with the degree of monosyllabicity in Cantonese. In an experimental study of the relationship between homophony and internal morphological change in Chinese, Tsou (1976: 82) elicited narrative data in Mandarin and Cantonese using two stories: “Confucius Confused” and “The Boy Who Cried Wolf.” The results showed that “the proportion of disyllabic types is much greater for Mandarin than for Cantonese”, as exemplified in Table 6.

Table 6:

Disyllabic words in Mandarin versus monosyllabic words in Cantonese (Tsou 1976: 78–79).

Meaning in English	Mandarin	Cantonese
to quarrel (v)	zhēngchăo 爭吵	cou4 嘈
to perceive (v)	gănjué 感覺	gok3 覺
to play (v)	wánshuă 玩耍	waan2 玩
to say (v)	gàosù 告訴	waa6 話
be warm (adj.)	nuănhé 暖和	nyun5 暖
matter, affair (n.)	shìqíng 事情	si6 事

According to Tsou, such findings:

point to a very interesting correlation between the rise of homophony and internal morphological developments in the dialects. Thus while Mandarin has undergone a much greater measure of syllabic simplification than Cantonese, it has, as a compensating factor, developed further in the direction of a polysyllabic or disyllabic language.

(Tsou 1976: 82)

More recently, a basic word list with a stronger empirical grounding, the Leipzig-Jakarta Word List, has been made available by Haspelmath (2008) and his associates (Haspelmath and Tadmor 2009; Tadmor et al. 2010) as one of the outcomes of their World Loanword Database (WOLD) project. They invited scholars to indicate whether the 1,460 word meanings adapted mainly from Mary Ritchie Key’s Intercontinental Dictionary Series are indigenous words or loanwords in the languages they examined. This approach has many advantages compared with a largely intuitive word list. Controlled for meanings from 24 semantic fields (http://wold.livingsources.org/semanticfield), and analyzed for their borrowability score on a 5-point scale (1=“No evidence for borrowing”; 5=“Clearly borrowed”), each of the 1,460 meanings in the recipient language across 41 language projects generated a composite score to facilitate ranking. On the basis of the composite scores, a top-100 word list was produced (the Leipzig-Jakarta list, Tadmor et al. 2010: 239–241). Other lexicostatistics generated from this large-scale collaborative project include the “ranking of the languages with respect to the proportion of (clear or probable) loanwords in their vocabulary” (Tadmor et al. 2010: 230). Of the 41 languages, Mandarin ranks the lowest (1%) while English ranks fifth (41%). The 41 vocabulary lists, coordinated and maintained by the digital library of Max Planck Institute for Evolutionary Anthropology, can be accessed online for easy comparison. The approach in this well-conceived, empirically grounded project has substantially helped complement theory-driven work that focuses on researching linguistic constraints as a paradigm for better understanding factors which impact on the paths (directions) and rates (frequency and speed) of language change.

On the basis of the 2,120 entries on the Mandarin Word List (http://wold.livingsources.org/vocabulary/22), we worked out the Cantonese equivalents. When analyzing the Cantonese and Mandarin data, we found that often multiple entries may correspond with a given WOLD meaning (e.g., for meaning 1.212, ‘soil’, three entries are provided for Mandarin: tu3rang3 土壤, tu3di4 土地, and tu3 土). This happens to both Mandarin and Cantonese. For our purpose in this study, we excluded all those meanings in Mandarin and Cantonese with more than one entry, and analyzed only those with one entry. This process yielded 674 items in both languages (46.16% of the 1,460 word meanings on the WOLD list, Appendix 3). Of this subset, Mandarin shows a monosyllabic-polysyllabic (mono-poly) ratio of ca. 2:4 (228:446 > 51:100), while the corresponding ratio for Cantonese is ca. 3:4 (290:384 > 76:100) (Figure 1). A similar trend is found after the top 100 words in the Leipzig-Jakarta list are translated into Mandarin and Cantonese. Of the top 100 word meanings, 88 (mono-poly ratio ca. 9:1) and 73 (mono-poly ratio ca. 7:3) are monosyllabic in Cantonese and Mandarin, respectively (Appendix 3).

Figure 1:

Monosyllabic-polysyllabic ratios in Mandarin vs. Cantonese.

These ratios are consistent with Luke and Lau’s (2008) finding that Cantonese is more characteristically monosyllabic compared with Mandarin. We believe this provides further evidence in support of the Hong Kong Cantonese community’s predilection for monosyllabicity, especially verbs and adjectives.

7.3 Bilingual punning

MEWs constitute a pool of additional resources for bilingual punning, which is very common in Hong Kong Chinese public discourse, especially advertising. Words such as fun, high and phone are among the most often blended into Cantonese to create additional semantic nuances. For example, biliterate readers have no difficulty reading FUN FUN 鐘, fan1 fan1 zung1, as ‘every FUN minute’, where FUN, homophonous with 分 (fan1), yields the double meaning ‘fun’ and ‘minute’ (compare: 分分鐘, fan1 fan1 zung1, ‘every minute’, Li 2000a: 315; cf. Li and Costa 2009). Similarly, a writer who alluded to the community-wide fad of iPhones wrote i瘋 (ai1 fung1, ‘i crazy’), whose reference to iPhone was unmistakable for Chinese readers (Mandarin fēng, example from Chan 2011). A still more instructive example of bilingual punning is found on the home page of the Development Bureau of the Hong Kong Government:

(15)

‘由我建造’ BUILD 升培訓計劃

jau4	ngo5	gin3zou6	BUILDsing1	pui4fan3	gai3waak6
by	1SG	construct	BUILD rise	training	program

‘“Constructed by me” Build/rise/soar training program’

(Heading/slogan, http://www.buildhk.hk/tc/build_prospect/, see Figure 2)

The creative collocation ‘BUILD 升’ puns on 飆升 (biu1 sing1, ‘soar’), thus conveying an additional nuance that the training program is designed to groom master ‘builders’ and promises a fast track to a rewarding career. This message is reinforced visually by the advert, where the character 升 is artistically stylized like a high-rise building, with the angled first stroke to the top left shaped like an upward-pointing arrow (Figure 2). In the rest of the home page, the program is presented consistently within angled brackets as「Build 升」計劃 ‘Build/rise/soar program’.

Figure 2:

Slogan of the Development Bureau, Hong Kong Government, http://www.buildhk.hk/en/build_prospect/ (Accessed 18 April 2013).

Clyne (2003) notes that bilingual homophones are pivotal points which allow the bilingual to proceed in either the recipient language or the source language. We believe bilingual punning constitutes strong evidence demonstrating not only the speaker-writer’s bilingual awareness, but also a conscious attempt to exploit the semantic potential afforded by bilingual homophones.

8 Conclusion

Our starting point was the observation that many monosyllabic English words (MEWs) are freely inserted into Hong Kong Cantonese, in speech and informal written Chinese. In a corpus of informal writing collected from Chinese newspapers during the mid-1990s (Li et al. 2014) consisting of 600,000 characters, roughly one in four to five unintegrated insertions is monosyllabic. There is thus prima facie evidence suggesting that MEWs are treated collectively by the Hong Kong Chinese community on par like Cantonese morpho-syllables. This led us to a search in the literature for the theoretical grounding of this phenomenon. After examining Clyne’s (2003) analysis of language contact data in Australia, we believe his notion of facilitation – building on his earlier work on triggering (Clyne 1967, 1980) – lends itself very well as an explanatory framework of the preponderance of MEWs in Cantonese, written as much as spoken.

To argue on the grounds of perceptual salience as the basis for facilitation of cross-linguistic transference begs the question, what linguistic evidence is there to prove that transference is facilitated? Guided by this research question, we found a variety of linguistic evidence that point toward monosyllabic salience of Cantonese as a possible typological feature. These include:

1. (i)

   a tendency for polysyllabic English words to be truncated to monosyllables (Luke and Lau 2008), especially verbs and adjectives;

2. (ii)

   a shorter average word length compared with Mandarin (Tsou 1976), which is further evidenced in the World Loanword Database (WOLD, controlled for 1,460 word meanings) involving 41 languages (http://wold.livingsources.org/, Appendix 3), and the Leipzig-Jakarta word list (top 100 words based on WOLD, Appendix 4);

3. (iii)

   the truncation of the first syllable of a polysyllabic word embedded in the A-not-A structure for asking yes-no questions;

4. (iv)

   bilingual homophony, which is commonly exploited for bilingual punning, facilitating cross-linguistic transference thereby (Clyne 2003); and

5. (v)

   the creative coinage of Romanized Cantonese words, which tend to be monosyllabic.