Variation in the Amplifier System Among Chinese L2 English Speakers in Australia

Abstract

This study investigates the English adjective amplifier system of eleven Mandarin Chinese L2 speakers of English residing in Australia compared to a sample of ten native Australian English (AusE) speakers from the AusTalk corpus. Employing a variationist framework, we find that the L2 speakers employ a markedly overall higher rate (50.2%) of use of adjective amplifiers than AusE speakers (34.8%). This has been shown to be a common phenomenon among L2 speakers, who have a smaller range of adjectives at their disposal, and thus “over-use” amplifiers. However, we also argue that the propensity for amplifier–adjective bigrams in Mandarin Chinese transfers to their L2 English. The results show that Chinese L2 speakers use very more than really, whereas really is more frequent than very in AusE, suggesting that the L2 speakers may be lagging behind in this previously-reported change in AusE. The results also show that higher rates of English proficiency and length of residence in Australia result in more Australian-like amplifier behavior among the Chinese L2 group. The present paper can provide meaningful insights for future language teaching and learning in classroom and naturalistic settings, revealing potential for the instruction of more authentic language among L2 English learners.

1. Introduction

In a globalized world, sociolinguistic competence, or the ability to be aware of sociolinguistic variation and apply this awareness to language production (Canale & Swain, 1980; Geeslin & Long, 2014, p. 6), is essential in language teaching and learning. However, non-native English learners and teachers can experience challenges in developing this kind of competence (Izumi, 1996; Mede & Dililitaş, 2015), which can have fundamental implications for the attainment of native-like proficiency (Kanwit, 2022, p. 33). A pertinent question relating to the L2 acquisition of sociolinguistic competence is how capable L2 speakers are of replicating and acquiring variation and change in progress in the L1 community (see e.g., Schleef et al., 2011; Davydova, 2019).

This paper focuses on L2 acquisition of linguistic variation in the use of English amplifiers (cf. Quirk et al., 1985). Amplifiers such as very, really, and so, also referred to as degree adverbs, are defined as having a raising and “scaling up” effect on adjectives (Bolinger, 1972, p. 17) and are frequently used in everyday spoken English. Amplification is a highly dynamic and “unstable” variable (Davydova, 2023, p. 3). While really is becoming the leading variant in the adjective amplifier system of native speakers, very is declining. As with many situations regarding language change, this results in synchronic variation as forms “compete” with each other (see Tagliamonte, 2006, p. 11). This results in a challenging set of targets for language learners. Do they choose the new, incoming variants, or the traditional variants?

Ongoing change in the English adjective amplifier system has been studied in a range of native English varieties (see Section 2.2 for a detailed overview). Meanwhile, previous research has examined the use of adjective amplifiers among non-native English speakers (Section 2.4). However, to date, there has been no variationist sociolinguistic study of amplifiers among Chinese L2 English speakers in Australia (or L2 speakers of any other L1 background, for that matter), despite ever-increasing numbers of Chinese international students, who constitute an essential immigrant group in Australia (Tao, 2023). Further, with the exception of Davydova (2023), who studies the distribution and patterning of English adjective amplifiers in German learner English and the effect of three sociolinguistic factors (linguistic identity, naturalistic exposure, and gender) on the use of really, very, and so, none of the studies in L2 contexts have adopted a fully accountable variationist sociolinguistic approach, unlike studies in L1 variation in amplifiers (e.g., Ito & Tagliamonte, 2003). Meanwhile, a range of linguistic-external constraints including language proficiency (LP) and length of residence (LoR) have also not been thoroughly examined in previous research, despite these factors being central to the research on how L2 speakers participate in linguistic variation more broadly (see, e.g., Geeslin, 2022). The present paper will address these research gaps.

Applying the variationist sociolinguistic framework, the current study investigates the English adjective amplifier system of Chinese L2 English speakers residing in Australia, incorporating the social factors of (self-reported) gender, language proficiency, and length of residence in Australia. We analyze quantitative data (frequencies and percentages of amplifiers) from sociolinguistic interviews (cf. Labov, 1984) with 11 Chinese students at an Australian university. The results are compared with those of ten native AusE speakers from the AusTalk corpus (Burnham et al., 2011).

2. Literature Review

2.1. Amplifiers

Intensifiers are a particular type of adjective modifier, corresponding to adverbs of degree (Allerton, 1987, p. 16), and ranging from very low intensification to very high. While downtoners1 scale downwards from an assumed norm, amplifiers, which are the focus of this study, have a raising effect (Quirk et al., 1985). Amplifiers can be viewed on a continuum from boosters (e.g., very, pretty, really) to maximizers (e.g., totally, extremely, perfectly) (Quirk et al., 1985, pp. 590–591).

Table 1. Main amplifier variants in AusE.

English Variety	Amplifier Variants
AusE (Schweinberger, 2021)	Really, very, so, pretty, bloody, absolutely, totally, completely, extremely, particularly, true, actually, awfully, genuinely, incredibly, real, strongly

provides a list of the main amplifier variants identified in Schweinberger (2021) for Australian English (AusE), which provide the principal points of reference for this paper.

It is crucial to determine the syntactic position of the adjective heads with which amplifiers occur (cf. Ito & Tagliamonte, 2003; D’Arcy, 2015; Schweinberger, 2021). Amplifiers can modify nouns in attributive position, in which the noun being amplified is positioned after the amplifier (1a–b), and in predicative position, where the noun being amplified is positioned before the amplifier (1c–d).

(1)	a.	I think that would be a very good try. (participant B6)
	b.	So I think the market is- is a very important factor here. (participant B4)
	c.	It’s still very hard for me. (participant A5)
	d.	They’re all very delicious. (participant B7)
		(nouns underlined; amplifiers and adjectives bolded and italicized)

2.2. Variation in the Amplifier System of Native English Speakers

A range of research has centered on the amplifier system in spoken native English varieties, e.g., AusE (Schweinberger, 2021), BrE (Ito & Tagliamonte, 2003; Xiao & Tao, 2007; Aijmer, 2018), Canadian English (Tagliamonte, 2008; Tagliamonte & Denis, 2014), New Zealand English (D’Arcy, 2015), etc. Much attention has been given to the distribution of amplifier variants. A significant and consistent trend is found in apparent time studies: really is increasing, while very is declining, and the observed change “does not proceed in a haphazard manner but […] is highly systematic” (Schweinberger, 2021, p. 4).

Ito and Tagliamonte (2003) find that very is the most frequent amplifier (38.3%) among native BrE speakers, and 68.5% of all amplifiers are subsumed under two variants: very and really. Furthermore, a notable pattern is found in apparent time data: for BrE speakers aged over 35, very is the most popular amplifier, but the variant is declining among younger speakers (17–34), where really is becoming the preferred variant (Ito & Tagliamonte, 2003, p. 267). Similarly, D’Arcy (2015) reports that very is the most frequent form across 100 years (1850s–1950s) in New Zealand English (NZE), but really overtakes very among NZE speakers born after the 1950s (p. 467). Tagliamonte (2008) also finds that very is the most popular variant in Toronto English speakers over 50, but it declines rapidly thereafter, particularly among young Toronto English speakers (under 30) (p. 372). By contrast, really increases sharply from the oldest to the youngest group, while so is the second most frequent amplifier among the young speakers (Tagliamonte, 2008, p. 372).

With respect to linguistic-external constraints identified across the English-speaking world, the situation is complex (Schweinberger, 2021, p. 5). With regard to gender, Fuchs (2017) finds that men consistently employ amplifiers less frequently than women in the Spoken British National Corpus, while Ito and Tagliamonte (2003), D’Arcy (2015), as well as Tagliamonte (2008), find no consistent gender differences in spoken data from other native varieties. Further, Tagliamonte (2016) reports the pattern of gender differentiation found in Canadian English and television series, whereby “the pattern of females using more so has persisted” (p. 92), and using pretty is found to be the mark of being young and male (p. 90). Previous studies have also shown that educational background conditions the use of amplifiers to some extent. Higher education has been found to be correlated with higher use of really in both middle-aged and younger speakers (Ito & Tagliamonte, 2003, p. 275). Xiao and Tao (2007) find that BrE speakers with higher education generally employ amplifiers more frequently (p. 256).

2.3. English Amplifiers in Australia

Drawing on the spoken private dialogue section in the Australian component of the International Corpus of English (ICE-Aus), Schweinberger (2021) investigates the amplifier system, finding that really has replaced very to become the dominant amplifier variant used in both syntactic positions (attributive and predicative) among young AusE speakers (aged 17–25), while so is the second most popular amplifier in predicative contexts (p. 14). He uses a lexical diversity (LD) score to investigate the collocational profile of amplifiers, which “is calculated by dividing the number of adjective types a given amplifier co-occurs with by the number of tokens of that amplifier” (p. 11; see also Section 3.6). Schweinberger (2021) argues that examining the LD scores of amplifiers is more meaningful than focusing on the semantic categories of adjectives (which have been examined in previous studies, e.g., Ito & Tagliamonte, 2003), as the results of Boruta analysis2 show that examining the semantic categories of adjectives is not a meaningful predictor for the use of really in AusE (p. 17).

Schweinberger (2021) reports a low LD score (0.3) for really among younger AusE speakers (17–25), but a higher score (0.7) among older AusE speakers aged 41–80 (p. 15). In other words, older speakers employ really to modify a greater range of adjective heads than younger speakers do. Additionally, a covarying co-lexeme analysis illustrates that, compared with other adjectives, really and good (the most frequent adjectives in the AusE data) significantly collocate with each other among the young AusE speakers, which indicates “specialization with a focus on high-frequency adjectives” (p. 16). By contrast, so occurs less frequently with good than would be expected by chance (Schweinberger, 2021, p. 16). For comparative purposes, the present paper also includes the use of a LD score, which has not yet been employed in previous studies of L2 learners’ use of the amplifier system in English. Further, as we do not aim to examine the delexicalization process of intensifiers (see, e.g., Ito & Tagliamonte, 2003; Davydova, 2023), traditional linguistic-internal variables (e.g., adjectival frequency and emotional value) are not included.

2.4. Amplifier Variation Among L2 English Learners

Amplification among non-native English speakers has been the subject of much linguistic research (e.g., Liang, 2004; de Klerk, 2005; Qi, 2006; Dunn, 2012; Edmonds & Gudmestad, 2014; Su, 2016; Hendrikx et al., 2019; Schweinberger, 2020, 2024). The most relevant paper to the current study is Liang (2004)’s study of Chinese L2 English learners. Comparing two spoken English corpora, Liang (2004) finds that Chinese L2 learners employ a much larger proportion of intensifiers (including amplifiers and downtoners) by normalized frequency, as compared to BrE native speakers, which indicates a tendency for “overstatement” among the L2 learners (p. 108). Moreover, very is found to be the most frequent variant, and it is much more popular than really, the third most frequently used amplifier variant (Liang, 2004, p. 108).

Schweinberger (2024) analyzes adjective amplifier systems in Hong Kong English (HKE), Indian English (IndE), and Philippine English (PhiE)3 with spoken English data from ICE to assess if really or so is replacing very as the dominant adjective amplifier in these outer circle Asian varieties. The quantitative results show that the amplifier systems of HKE and IndE are consistently dominated by very, which mirrors the patterns found in native varieties before really starts replacing very (p. 13). In contrast, so and really have replaced very as the dominant amplifier in PhiE. Schweinberger (2024) also reports that more than 90% of adjectives are amplified by very, so, and really across these Asian Englishes (p. 9). Regarding the effect of language-internal and-external factors, Schweinberger (2024) finds that amplifier choice in HKE and IndE is primarily determined by internal constraints (i.e., adjective type, syntactic context, and semantic category). Meanwhile, ongoing change in PhiE from very to really and so is driven by social factors (i.e., age and gender) (p. 14).

Dunn (2012), taking a variationist sociolinguistic approach, compares the use of amplifiers among five L1 and five L2 English speakers from three backgrounds (Mandarin, Spanish, and Vietnamese) at a university in the U.S. The rates of amplification are reported across the speaking groups, with a minor but visible difference (44.3% for the L2 group; 39.4% for the native group). Dunn (2012) finds that very is the most frequently used variant in the amplifier system of the L2 English learners, followed by so and really (p. 27). Dunn (2012) also finds that L2 English speakers “used common boosters, such as very, really, and so, at higher overall rates, but produced a smaller variety of amplifier lexemes” than the native English speakers (pp. 27–28).

Within the framework of variationist sociolinguistics, Davydova (2023) studies the amplification rate of 53 advanced German learners of English and finds a higher rate of amplification (50.4%) in the German L2 data compared with L1 English, such as British English (24%, reported in Ito & Tagliamonte, 2003), which echoes Liang (2004) and Dunn (2012). Davydova (2023, p. 10) also reports that really, very, and so stand out as frequently used variants. Moreover, she finds that male speakers overwhelmingly prefer the use of very and pretty, while really and so tend to be more frequent among female speakers (p. 11).

To summarize this previous research, L2 English speakers, with some differences across language backgrounds, employ amplifiers at higher rates in comparison to native speakers, and so, really, and very are the most popular variants for both L1 and L2 English speakers. Consequently, this paper also centers on these three amplifiers. Moreover, the use of pretty will be included when we investigate the effect of external factors, as this has been found to be prevalent among male English speakers in particular (see, e.g., Tagliamonte, 2016; Davydova, 2023).

2.5. The Collocation Patterns of Amplifiers Among L2 English Learners

Collocation patterns have been extensively researched in the context of the amplifier system of L2 English learners (e.g., Liang, 2004; de Klerk, 2005; Edmonds & Gudmestad, 2014; Hendrikx et al., 2019; Schweinberger, 2020). First of all, the reduplication of a (basic) amplifier has been widely observed among L2 English learners, such as very very good (cf., Liang, 2004; de Klerk, 2005; Hendrikx et al., 2019; Davydova, 2023). Davydova (2023) argues that German L2 learners employ double intensification to “increase the pragmatic force of an already established variant” (p. 11). Liang (2004) also suggests a possible explanation: “In cases where a maximizer is optimal, learners tend to go one step down the intensifier scale and resort to the booster word very”, but “they are well aware that very does not warrant them their intended meaning” (p. 112). As a consequence, the booster very is used twice, or even three times, to achieve the desired effect. According to de Klerk (2005), this is because L2 speakers lack a range of lexical choices (p. 16), although this kind of construction is also attested in L1 varieties (Méndez-Naya, 2017; Merx, 2018). Another feature that has been found among Chinese L2 learners of English is the use of amplifiers to modify non-gradable adjectives, such as *very perfect (or equal, exhausted, or acquainted), which shapes the “non-nativelikeness” of Chinese EFL learners’ speech (Liang, 2004, p. 111).

Other research has shown that external constraints, such as language proficiency, have an impact on amplifier collocations among L2 English speakers. Edmonds and Gudmestad (2014) find that advanced L2 English learners and native speakers perform similarly on a collocation judgment questionnaire and fill-in-the-blank production task for amplifiers, while different patterns are found among L2 speakers with lower English proficiency. Similarly, Hendrikx et al. (2019) report that longitudinal foreign language input in English leads to a progressive tendency toward more target-like use of intensification among non-native English speakers. It must be noted, however, that the effect of social and contextual factors (e.g., gender, English language proficiency, length of residence in English-speaking country) on the frequency and distribution patterns of various amplifiers has not been extensively studied in the existing literature.

This study aims to address the research gap stated above by investigating the acquisition of amplifiers by Chinese L2 learners of English in Australia. The L2 English speakers’ use of amplifiers, and particularly popular variants such as really, is compared with amplifiers used by ten native AusE speakers from the AusTalk corpus (Burnham et al., 2011). In doing so, three research questions guide this paper:

• Do Chinese L2 English learners in Australia follow the distribution pattern of AusE in the use of English amplifiers?
• How do Chinese L2 English learners compare with AusE speakers in their collocation patterns of the amplifiers so, very and really, that are currently prevalent in English varieties?
• Do the factors of (self-reported) gender, length of residence in Australia, and language proficiency influence Chinese L2 English learners’ use of so, very, pretty, and really?

3. Methodology

3.1. Research Setting and Participants

The current study recruited eleven Chinese L2 speakers of English (six female and five male) as participants with full human ethics approval. Their profiles are presented in Appendix A. The participants all come from Mainland China. They listed “Chinese”, “Mandarin”, or “Standard Mandarin” as their mother tongues in a questionnaire, with English recorded as the second language of all participants. Their ages range from 23 to 26 (mean age: 24). All participants were either completing or had completed master’s programs at university, with all but two (participants A4 and B5; see also Appendix A) in the area of applied linguistics.

All participants had met the language requirement for entry into a master’s program at their university: an overall band score of ≥6.5 in the International English Language Testing System (IELTS) language test, or an equivalent result in another recognized English language test. Nine participants provided an IELTS test band score, with ranges from 6.5 to 7.5, and two participants provided equivalent scores from the Pearson Test of English (PTE). Following Choi and Diskin-Holdaway (2022), participants were divided into two proficiency groups (see Appendix A): a lower proficiency group, consisting of those who scored an IELTS test result of 6.5 or equivalent (N = 5), and an upper proficiency group, consisting of those who scored an IELTS test result of 7 and 7.5 or equivalent (N = 6). The reason for this cutoff point is that participants with a score of 6.5 (“competent users”, IELTS, n.d.-a) are classified at the B2 level and as an independent user in the Common European Framework of Reference for Languages (CEFR; see IELTS, n.d.-b), whereas those with a score of 7 and 7.5 (“good users”) are classified at the C1 level and as a proficient user in the CEFR, which we believed presented a meaningful difference.

It was hypothesized that length of residence (LoR) in Australia would increase the likelihood of acquisition of the popular variants (e.g., really and so) in AusE, so it is included as a factor in this study. Participants’ LoRs range from 1.5 to 66 months (mean length of residence: 18 months). The eleven Chinese L2 English speakers are divided into two LoR cohorts (see Appendix A): ≤12 months (“short”; 1.5–9 months, N = 4, SD = 3.1) and >12 months (“long”; 15–66 months, N = 7, SD = 17.6), as this presented a natural split in the data. There was some overlap between upper English proficiency and long LoR, and lower proficiency and short LoR (see Appendix A).

3.2. Instrument

We used a sociolinguistic interview (Labov, 1984) to obtain the natural speech of the Chinese L2 learners, which offers “examples of variation as evidence for linguistic change” (Becker, 2013, p. 107). The interview protocol is modified from Chen (2021) and Diskin (2015). There were 15 main questions, covering five topics in a semi-structured format: life, study experience, work, socializing, and language in Australia. The research objective—English adjective amplifiers—was not shared with the participants during the interview. Following the interview, participants filled out a questionnaire (see Appendix B; modified from Chen, 2021) designed to collect sociodemographic information (e.g., age, gender, mother tongue), length of residence, and information on language exposure/contact (including social networks) of the L2 English speakers.

3.3. Procedure

The Chinese L2 data were collected in April 2023. Participants engaged in sociolinguistic interviews with the first author either face-to-face or online via Zoom. All interviews were one-on-one and recorded either by the speech recognition software Otter.ai on a laptop, or by a voice memo application on a mobile phone. The length of the sociolinguistic interview ranged from 24 to 40 min (mean length = 29 min). After the interview, a copy of the questionnaire was sent to the interviewee via WeChat or email. Before participants filled out the questionnaire, the first author provided explanations of the different sections. The questionnaire had no time limit, and all the questionnaire files were to be sent back with clear answers within one day. The interview audio recordings were transcribed by Otter.ai, while a broad orthographic transcription convention and anonymization process4 were applied. The final transcribed output was compiled into a small-scale corpus of 24,710 words.

3.4. Australian English Data

In the present project, AusE data from the AusTalk corpus (Burnham et al., 2011) are used as a reference to compare with the L2 Chinese group. AusTalk is a large audiovisual corpus of modern AusE that contains spoken data recorded between June 2011 and June 2016 from 15 different locations in all Australian states and territories. The spoken data of 10 young AusE speakers were selected from the AusTalk corpus for their demographic comparability with the Chinese L2 data. All the native AusE speakers listed “Australian” as their cultural heritage. The ages of the participants range from 20 to 27 (mean age: 25), and they all were currently completing, or had been conferred with, bachelor’s degrees or graduate diplomas. This study analyzes 20 files (approx. 35,000 words) from AusTalk containing the orthographic transcriptions of story retelling and interview tasks where spontaneous speech was elicited. As our AusE sample was imbalanced with regard to gender (eight female versus two male participants), we do not analyze this sociolinguistic factor for the AusE group.

3.5. Coding Process

The principle of accountability is the cornerstone of variationist sociolinguistics (Labov, 1972; Tagliamonte, 2006; D’Arcy, 2015), which holds that “every variant that is part of the variable context, whether the variants are realized or unrealized elements in the system, must be taken into account” (Tagliamonte, 2006, p. 13). Our study is restricted to adjective heads, which enables us to approach the use of amplifiers with a consistent denominator—all English adjectives (Ito & Tagliamonte, 2003, p. 263). This accountable methodology includes all amplifiable adjectives—instances where an amplifier could have occurred, but did not, as well as all amplified adjectives (see D’Arcy, 2015, p. 458).

In accordance with the methodology outlined in Ito and Tagliamonte (2003), as well as D’Arcy (2015), we also exclude contexts that either do not permit amplifiers or block variation. Firstly, comparatives and superlatives, the “effectively categorical zero contexts”, were not included (D’Arcy, 2015, p. 459). Such contexts are limited to select amplifiers, such as the very best (because *the really best is not grammatically possible). As shown in (2a–c), cases in which the adjectives work as classifiers (bolded and italicized) were also excluded. Additionally, the current study did not analyze the contexts in which the variants do not function as the modifiers of adjectives, as in (3a–b).

(2)	a.	I started as an international student here in Australia out in October 2017. (participant B1)
	b.	I become more- I became more confident in academic writing. (participant A2)
	c.	Because it’s a compulsory subject so I have no choice. (participant B1)

(3)	a.	So (=conjunction) first one. (participant A3)
	b.	It’s really (=truly) a subjective question. (participant B1)

Moreover, if adjectives would “fall immediately under the scope of negation”, as in (4a–c), these negative contexts were likewise excluded (see D’Arcy, 2015, p. 459). According to Ito and Tagliamonte (2003), negation alters the meaning of amplifiers and cannot express a higher degree, thus losing their function (p. 264).

(4)	a.	And it was- it was overall pleasant journey because- even if the weather was not so good. (participant B2)
	b.	But that’s not really convenient. (participant B7)
	c.	It is not very convenient in Australia. (participant B6)

The first author coded and noted all amplifier–adjective collocations in the data (167 tokens for Chinese L2; 163 tokens for AusE), along with occasions where an amplifier could have occurred, but did not (166 tokens for Chinese L2; 306 tokens for AusE). Adjective heads and syntactic functions (attributive or predicative) were also included in the coding process. Then, the second author verified a subset of approximately 10% of the data (84 items) from both L1 and L2 speakers for whether the tokens were amplified, amplifiable, or not amplified. An initial inter-rater agreement of 95.2% was achieved, and following discussions over some more challenging contexts, a revised rate of 100% agreement was reached.

3.6. Data Analysis

Descriptive statistics were used to examine differences in the frequencies and percentages of all amplifiers. To investigate the collocational profile of three popular variants—really, very, and so, the lexical diversity (LD) scores for their collocations were also calculated, as shown below. The LD score ranges from 0 to 1: the higher the LD score, the higher the degree of lexical diversity.

LD = N _{Types of adjectives an amplifier co-occurs with}/N _{Tokens of an amplifier}

We also examined the effect of gender, language proficiency, and length of residence. It must be noted, however, that due to the relative sparsity of our sample (11 Chinese L2 participants and 10 AusE speakers), we focused on descriptive statistics rather than inferential statistical analysis of extralinguistic factors. As a result, these descriptive findings should be interpreted with caution and as a preliminary glimpse into the patterning of these data across the two groups.

4. Results

4.1. Overall Distribution of Amplifiers

The analysis of the Chinese L2 data reveals 167 amplifiers, and 166 non-amplified amplifiable adjectives, resulting in an amplification rate of 50.2% (

Table 2. Overall distribution of amplification in the Chinese L2 corpus and AusE.

	Amplified		0 Amplification
Chinese L2	50.2%	N = 167	49.8%	N = 166
AusE	34.8%	N = 163	65.2%	N = 306

). By contrast, only 34.8% of amplifiable adjectives were amplified in the AusTalk data, which had 163 amplified adjectives and 306 non-amplified amplifiable adjectives.

The frequencies and percentages of all amplifier variants were compared across the L1 and L2 groups (

Table 3. Frequencies and percentages of variants in the Chinese L2 corpus and AusE.

Variant	Chinese L2 %	Chinese L2 N	AusE %	AusE N
Very	29.3%	49	30.1%	49
Really	24.6%	41	37.4%	61
So	20.9%	35	6.7%	11
Pretty	18%	30	20.3%	33
Bloody	0	0	0	0
Too	3.6%	6	0.6%	1
Totally	1.2%	2	0	0
Especially	0.6%	1	0	0
Super	0.6%	1	0	0
Extremely	0.6%	1	0.6%	1
Perfectly	0.6%	1	0	0
Completely	0	0	2.5%	4
Absolutely	0	0	1.8%	3
Total		167		163

). First of all, the data reveal that very and really are the most popular amplifiers in the Chinese L2 group and AusE. Moreover, we find that the Chinese L2 speakers use very at a somewhat higher rate (29.3%) than really (24.6%). In contrast, really is found to be the most frequently used variant in AusE, with a difference between the proportions of really (37.4%) and very (30.1%).

It is noteworthy that so is the third most frequently used variant in the amplifier system of the Chinese L2 group (20.9%), but it is a marginal variant in AusE (6.7%). Overall, 74.8% of English adjective amplifiers employed by the Chinese L2 group consist of really, very, and so, while really, very, and pretty are used 87.8% of the time by the AusE group.

The remaining amplifiers across both corpora comprise a wide range of different variants. Overall, the L2 group employs a somewhat larger lexical range of amplifiers (N = 10) compared with the AusE group (N = 8). The variants totally, especially, super, and perfectly were unique to the L2 group, whereas completely and absolutely were unique to the AusE group. Furthermore, there were no examples of bloody in either corpus, despite it being reported as the fifth most popular variant in the ICE-Aus corpus (Schweinberger, 2021, p. 9).

The percentages of amplifier variants collocating with predicative and attributive adjectives are illustrated in

Table 4. Percentages of amplifiers in predicative and attributive contexts in the Chinese L2 corpus and AusE.

	Variant	Chinese L2 %	Chinese L2 N	AusE %	AusE N
Predicative	Really	22.2%	28	35.1%	45
	Very	21.4%	27	29.7%	38
	So	27.8%	35	8.6%	11
	Pretty	20.7%	26	21.1%	27
	Other	7.9%	10	5.5%	7
	Total		126		128
Attributive	Really	31.7%	13	45.7%	16
	Very	53.6%	22	31.4%	11
	So	0	0	0	0
	Pretty	9.8%	4	17.2%	6
	Other	4.9%	2	5.7%	2
	Total		41		35

. Overall, there are far more adjectives amplified in predicative position (126/167 = 75.4% for the L2 group; 128/163 = 78.5% for the AusE group) than in attributive position (41/167 = 24.6% for the L2 group; 35/163 = 21.5% for the AusE group).

The difference in proportions of the use of so in predicative context between the L2 group (27.8%) and the AusE group (8.6%) is particularly noteworthy, although this is also bolstered by the overall preference for so among the L2 group. Really is used more often in predicative position by the AusE group (35.1%) as compared to the L2 group (22.2%). We also find that more than half (53.6%) of the attributive adjectives in the L2 group are amplified with very, while a much lower proportion is found in AusE (31.4%). The predominance of very in the attributive position for the L2 group represents the most marked difference in the attributive/predicative ratio across the corpus (53.6% versus 21.4%). Further inspection of the collocations with very in attributive position in the L2 group revealed no particular specialized pattern. Finally, really is the dominant variant in the attributive context for the AusE group (45.7%), but the second most frequent for the L2 group (31.7%).

4.2. Collocation Patterns of the Main Amplifiers

The LD scores of the three dominant amplifiers—really, very and so—were calculated across speaker groups to examine the collocation patterns of adverb (amplifier)–adjective bigrams (

Table 5. Lexical diversity scores of really, very, and so in the Chinese L2 corpus and AusE.

Variant	Group	Adjective Types N	Variant Frequency N	LD Score
Really	Chinese L2	28	41	0.68
	AusE	37	61	0.61
Very	Chinese L2	33	49	0.67
	AusE	38	49	0.78
So	Chinese L2	29	35	0.83
	AusE	10	11	0.91

). Overall, the amplifier so yielded higher LD scores than really or very, with the AusE group showing the highest score for this variant at 0.91. However, so only appears 10 times in the AusE corpus, and according to Schweinberger (2021), LD scores are likely to be skewed by low frequencies (p. 14). The LD score of really is slightly higher in the Chinese L2 corpus (0.68) as compared to the AusE group (0.61), but the LD score for very in AusE (0.78) is higher than in the L2 group (0.67), albeit with identical frequencies. In addition, we find that the LD score of very (0.78) is higher than that of really (0.61) in AusE, while these two variants yielded similar LD scores (0.67 for very; 0.68 for really) in the Chinese L2 corpus.

An analysis was conducted of the main collocations with really and very (

Table 6. Major lexical collocations of really and very in the Chinese L2 corpus and AusE.

	Chinese L2		AusE
Really	Good	19.5%	Good	27.9%
	Different, depressed *	7.3%	Nice, great	4.9%
	Hard, interesting	4.9%	Bad, beautiful, fun, hollow, lovely	3.3%
Very	Good	14.3%	Nice	10.2%
	Different	8.2%	Good	8.2%
	Delicious, simple	6.1%	Cute, old	6.1%

* Adjectives listed on the same line each yielded the same proportions of use.

). The most popular collocation for really was really good, accounting for 27.9% (17/61) of all lexical collocations in AusE and 19.5% (8/41) in the L2 group. The collocation very good was the most frequently used among the L2 group (7/49 = 14.3%), followed by very different (8.2%), very delicious, and very simple (both 6.1%). In AusE, very nice has a slightly larger proportion (5/49 = 10.2%) than very good (4/49 = 8.2%).

Collocations including two or three repetitions of the amplifier are accounted for equally across both corpora (five cases in each), and at very marginal rates (5/167 = 3% for the Chinese L2 group; 5/163 = 3.1% for the AusE group). The reduplication of really (e.g., 5a) is identified four times in the Chinese L2 group, and there is one reduplication of very (5b). We also find two instances of the reduplication of really and three of very in AusE (e.g., 5c–d).

(5)	a.	It was good- really, really good. (Participant A3)
	b.	I mean, although professors and lecturers they’re very, very patient and very
		friendly to us. (Participant B2)
	c.	No, actually they’re really, really tight and uncomfortable and rubbery.
		(AusE speaker 2–959)
	d.	And as I was going up there, one of the very, very, very drunk English woman
		[sic]- turned around. (AusE speaker 3–794)

4.3. Analysis of the Effect of Linguistic-External Constraints

4.3.1. Comparison Between Gender Groups

First of all, as shown in

Table 7. Overall distribution of amplification in the female and male Chinese L2 groups.

	Amplified		0 Amplification
Female	53.8%	N = 106	46.2%	N = 91
Male	44.9%	N = 61	55.1%	N = 75

, 53.8% of amplifiable adjectives were amplified in the female Chinese L2 group. This figure is higher than the amplification rate of the male group (44.9%), by 8.9%.

The distributions of the four major variants (really, so, pretty, and very) for the male and female L2 groups are illustrated in Figure 1. First of all, we find that very and so are popular in both the female and male group. Moreover, really tends to be more frequently used in the male group (20.8% for the female group; 31.1% for the male group), with pretty being overwhelmingly preferred by the female L2 speakers (24.5% for the female group; 6.6% for the male group).

4.3.2. Comparison Between English Proficiency Groups

We compared the distribution patterns between two groups of Chinese L2 speakers with different English language proficiency (LP) levels (lower and upper; see Section 3.1) with the AusE speakers for the most popular amplifiers (Figure 2).

Really is found to be the dominant variant in the amplifier system of both the upper LP group and native speakers, which accounts for more than 30% of all amplifiers, followed by very and pretty. However, the proportion of really is much lower among the lower LP group, at just 12.9%. Instead, the lower LP group uses more very (37.1%) than really. Notable differences are also observable for so, which is the second most popular amplifier in the lower LP group after very with a higher rate of use (32.9%) than in AusE (6.7%) and the upper LP group (12.4%). Conversely, the upper LP group and AusE speakers use more pretty as compared with the lower LP group.

4.3.3. Comparison Between Length of Residence Groups

Figure 3 shows the distributions of the four major variants across corpora, incorporating different length of residence (LoR) levels (short and long) for the L2 group. The figure shows that while both the long LoR group and the native AusE speaking group use more really than very, the opposite situation is found among the Chinese L2 learners with short length of residence, who use more very than really.

Moreover, we find that the short LoR group has a high proportion of so (38.8%), which is in sharp contrast to the figures for the AusE group (6.7%) and the long LoR group (9%). It is also noteworthy that pretty is absent in the data of the short LoR group, while it represents the third most frequent variant in both the long LoR group and AusE, accounting for 30% and 20.3% of all amplifiers, respectively.

After reviewing the distribution patterns across various LP groups and LoR groups, we observed that the lower LP and short LoR groups behave similarly to one another, as do the longer LP and the long LoR groups. The only meaningful difference we found was that pretty was absent in the short LoR group but present in the low LP group. Overall, this suggests that LP and LoR are related, and indeed present some overlap. Two out of five lower proficiency speakers are in the short LoR group (four participants in total), while there are four (out of six) higher proficiency speakers in the long LoR group (seven participants in total).

5. Discussion

First of all, we examine the overall distribution of amplification to answer the first research question—Do Chinese L2 English learners in Australia follow the distribution pattern of AusE in the use of English amplifiers? The results show that the Chinese L2 group employs more English adjective amplifiers than the AusE speakers, which is consistent with the findings of Dunn (2012), Liang (2004), and Davydova (2023). One interpretation for the greater degree of amplification is that L2 English learners’ vocabulary in adjectives may be limited (Dunn, 2012, p. 23). They may want to accomplish the same interactional work as L1 speakers but have a smaller lexical range of adjectives at their disposal (Dunn, 2012, p. 23). Therefore, they have to rely on amplification to achieve the level of expressivity comparable to that of L1 speakers, while native speakers are able to use more specialized and specific adjectives rather than combinations of an amplifier and adjective (Dunn, 2012, p. 23). In other words, the “overuse” of amplifiers becomes Chinese L2 learners’ strategy for effective verbal communication in English. Moreover, while we collected the Chinese L2 data in 2023, the AusE spoken data were recorded before 2016. According to Fuchs (2017), rates of intensification have been expanding over time in L1 varieties. Therefore, this is another possible explanation for the higher amplification rate found in our Chinese L2 group and we may have found more comparable rates of amplification had we compared with more recent AusE data.

Our initial (descriptive) findings also show that the L2 group uses very more than really, which echoes previous studies on L2 English speakers (e.g., Liang, 2004; Dunn, 2012). It is also in contrast to the distribution found among the AusE group in the current study as well as in work across different native varieties and/or other corpora, e.g., AusE (Schweinberger, 2021), Canadian English (Tagliamonte, 2008), and NZE (D’Arcy, 2015), suggesting that the L2 groups are either (a) not mirroring the change-in-progress in the L1 community in which they live or (b) partaking in the change-in-progress, but lagging behind.

We also note that in recent work on outer circle varieties, very is found to be the absolute dominant variant among young speakers of Indian English and Hong Kong English (over 60% in both varieties), whereas so and really have fully replaced very in Philippine English (Schweinberger, 2024, pp. 13–14). In German L2 English, really is also leading the way (Davydova, 2023, p. 11). In our sample, the L2 group differs from these findings in the sense that they do have some variability in their amplification strategies (very at 29.3%, so at 20.9%, and really at 24.6%), which is more similar to the patterns in our AusE group. Furthermore, our findings for LoR in particular, whereby the L2 group with longer residence in Australia tend to have more Australian-like patterns of amplification, suggest that the L2 group cleaves more closely to the L1 community in which they are immersed, rather than following any particular general “L2-like” pattern. As a result, we suggest that they are lagging behind, rather than not participating at all, in this change-in-progress. More data would be needed to further lend weight to this proposal.

To answer the second research question—How do Chinese L2 English learners compare with AusE speakers in their patterns of use of so, very and really?—we investigate the amplifier–adjective collocations of the popular amplifiers in both corpora. First of all, the LD scores of really (0.68) and very (0.67) are comparable in the Chinese L2 corpus. Conversely, in the AusE data, we find that really has a relatively low LD score (0.61) compared with that of very (0.78). This trend echoes, but does not exactly replicate, the findings of Schweinberger (2021) for a young AusE group (aged 17–25) with even lower LD scores (0.3 for really; 0.4 for very) in ICE-Aus (p. 15). However, according to Schweinberger (2021), the replacement of very by really in AusE is accompanied by the broadening of very and the specialization of really (p. 21), which explains the difference in really and very LD scores in our AusE data, and contrasts with our Chinese L2 speakers, who show little difference in LD between very and really. One plausible explanation for this is that, as discussed above, the amplifier system of the Chinese L2 English learners has not yet reached the later stage of the change-in-progress, a benchmark we assume to be set by our AusE data. We also found a much higher proportion of really good found among the young AusE speakers, reflecting a specialization of the variant, and echoed by the findings of Schweinberger (2021, p. 16) in a covarying co-lexeme analysis. It is possible that with a larger dataset, our LD scores would be as low as those found in Schweinberger (2021), who, as we mentioned earlier, notes that LD scores can be skewed by low frequencies.

Moreover, the repetitive use of really and very is found in both our Chinese L2 corpus and AusE corpus at identical (and low) rates (3% for the Chinese L2 group; 3.1% for the AusE group), which is similar to the figure (3%) reported in Davydova (2023)’s variationist study on German learners of English. This indicates a parallel in the use of double intensification as a widespread cross-linguistic phenomenon in L1 and L2 English varieties. Further, in addition to the explanations by Liang (2004) and Davydova (2023), L1 (Chinese) influence is another potential reason. According to Yang (2022), in the Center for Chinese Linguistics PKU (CCL) Corpus, repetitive use of amplifiers has been found among eight popular Chinese degree adverbs, including two Chinese amplifiers “fēi cháng” (非常)5 “very/too/pretty” and “shí fēn” (十分) “very/really/pretty” (p. 73). Therefore, it is possible that Chinese equivalents of really and very may have a transfer effect on these amplifiers’ collocations in English. It is also worth mentioning that a smaller lexical range of amplifiers (N = 8) is identified among our native AusE speakers compared with the Chinese L2 group (N = 10), suggesting that the L2 group is not struggling to retrieve appropriate amplifiers. However, our sample sizes are much smaller compared to other corpora (e.g., ICE), so these figures should be interpreted with caution.

Returning to the finding regarding the overall higher amplification rate by our Chinese L1 group, we suggest that an L1 (Chinese) transfer effect could also play a reinforcing role, particularly regarding the higher rates of so and very in their amplifier system. Emphasis, exaggeration, and overstatement often appear in Chinese, which are fulfilled by the use of “chéng dù fù cí” (程度副词) “degree adverbs” (including amplifiers) (Qi, 2006, p. 50). Meanwhile, a lot of frequently used adjectives, including around 21% of “new” Chinese adjectives that started to be prevalent in the 1990s (Cai, 2013, p. 8), are monosyllabic, such as “kù” (酷) “cool”, “huǒ” (火) “hot”, “měi” (美) “beautiful”, and “dà” (大) “big/large”, but two-syllable constructions are generally preferred in modern Chinese (Pan, 2003). As a result, Chinese speakers tend to employ combinations of monosyllabic amplifiers (e.g., “hěn” 很 “very/so”) and monosyllabic adjectives, leading to, e.g., “hěn dà” (很大) “very/so large”, “hěn tián” (很甜) “very/so sweet” (Qi, 2006, p. 50), reflecting the syllable-timed isochrony of Mandarin Chinese. As a result, it could be argued that with the influence of these frequent collocations in their L1, the Chinese L2 speakers with lower proficiency and short LoR in the present study prefer to use similar constructions, using the translation of “hěn” (很)—“very/so”—into English.

When it comes to the effect of linguistic-external factors, the analysis of language proficiency and LoR shows that higher LP and longer LoR in Australia might lead to a preference for really (the dominant amplifier in AusE) and pretty, as well as a more native-like, or Australian-like, distribution pattern of amplifier usage. Thus, it can be cautiously suggested that LP and LoR in Australia have an effect on Chinese L2 English speakers’ use of amplifiers. According to Hendrikx et al. (2019), from a usage-based perspective, the amount of input an L2 learner receives is an essential factor in second language acquisition. Intensifiers (including amplifiers), and language-specific intensifying patterns such as collocations and recurrent phrases, tend to not feature prominently in explicit second language instruction (Meunier, 2012). Thus, the acquisition of amplifiers likely relies on implicit learning, and “the amount of target language exposure may be key to a target-like acquisition” of amplifiers (Hendrikx et al., 2019, p. 64). Our initial findings support this view, and a larger dataset with a greater time depth for length of residence could further confirm this, and pinpoint more clearly what stage of residence/exposure is pivotal for the acquisition of such features.

With respect to the effect of gender, the female Chinese L2 English speakers tend to use more amplifiers compared with the male speakers. However, more robust evidence is still needed. We also find that really is more frequent among the male L2 speakers, which is in contrast to Schweinberger (2021), who reports that women are more likely to use the variant compared with male speakers in AusE (p. 18). Further, it is noteworthy that pretty was found to be considerably preferred by the female speakers in the Chinese L2 corpus. This does not align with Davydova (2023) and Tagliamonte (2016), who find that pretty is more frequently used by the male speakers in German L2 English and Canadian English, respectively. In brief, some of the patterns of gender differentiation found in our Chinese L2 sample differ from those reported in other L1 and L2 varieties, which compels us to explore this variable in future studies with a larger dataset.

6. Conclusions

This variationist study investigates the English adjective amplifier system of young Chinese L2 English learners residing in Australia and compares the results with AusE data from the AusTalk corpus. The findings uncover differences in the distribution and collocation patterns between the L1 and L2 groups, indicating that the Chinese L2 group, an immigrant group in Australia, appears to be lagging behind, but still participating in, the ongoing change observed in AusE and other L1 varieties. Our work opens up new opportunities in variationist sociolinguistics for examining the mechanisms of language change in immigrant groups from different linguistic backgrounds and with varying degrees of participation in the L1 community. In future work, we will incorporate a more advanced language exposure metric with a larger set of participants and further examine psychosocial factors such as attitudes towards the target language and towards specific amplifiers.