Variation in the use of sentence final particles in Macau Cantonese

1

Variation in the use of sentence final particles in Macau Cantonese

澳门粤语句末助词使用的变异研究

Werner Botha and Lawrie Barnes

Abstract

摘要

This paper considers how meaningful social information is conveyed with the use of sentence final

particles in Macau Cantonese. The purpose in this research is to provide a general sociolinguistic

account of sentence final particles in Macau Cantonese, and specifically to illustrate that social meanings

of SFPs are variable, and do not constitute rigid or fixed meanings and interpretations. These social

meanings, this paper argues, are a potential for indicating speaker identity at the individual level, and

constitute a rich resource for communicating speaker identity in Macau Cantonese. This study uses an

eclectic sociolinguistic approach, and combines elements of distributionist analyses, social network

theory and constructionist approaches with a view to accounting for the dynamics underlying sentence

final particle variation. Finally, this research considers constraints such as conversation topic, the

affective relations between interlocutors, and gender as impinging on the distribution and use of SFPs in

Macau Cantonese.

本文从社会语言学角度出发，考察澳门粤语句末助词的使用情况，并揭示句末助词如何帮助传达

有意义的社会信息，旨在对澳门粤语句末助词的大体使用情况进行分析，并明确提出句末助词所

传达的社会意义并非一成不变的，并无固定意义及解释。研究发现，句末气助词所传达的社会意

义不仅能为指明单个说话者身份提供可能性，而且为识别澳门粤语说话者身份提供丰富的资源。

本研究理论上采用折衷主义的做法，从社会语言学角度出发，结合概率分析、社会网络理论及构

式主义方法等因素，旨在解释说明句末助词变化的个中原因。研究结果表明，众多因素如对话话

题，对话者之间的情感性关系以及性别等，都对澳门粤语句末助词的分布及使用产生影响并有制

约作用。

Keywords: Social Network Theory; Sentence Final Particles; Language Variation; External Linguistic

Constraints; Macau Cantonese

关键词：社会网络理论；句子尾音；语言变异；外在语言制约因素；澳门粤语

2

Variation in the use of sentence final particles in Macau Cantonese1

INTRODUCTION

Over the last decade or so, language variation research has increasingly parted with simply correlating

sets of social variables with linguistic features (e.g. phonemes, grammatical features, etc). Specifically,

much recent attention in the field considers linguistic variables as resources that speakers draw on in

order to convey certain social information (see Schilling-Estes 2004). This paper considers how social

information is conveyed through the use of sentence final particles (or SFPs) in Cantonese, as the

language is spoken in the Special Administrative Region of Macau (i.e. Macau Cantonese). The purpose

of this study is to provide a sociolinguistic account of SFPs in Macau Cantonese, and to argue that social

meanings of SFPs are variable, and do not constitute rigid or fixed interpretations. These social meanings,

it is argued, are a potential for indexing speaker identity at the individual level and constitute a rich

resource for communicating speaker identity in Macau Cantonese.

This sociolinguistic study was conducted in Macau Special Administrative Region (hereafter Macau),

as Cantonese is the most widely used language in the territory. Macau consists of a small territory of

about 29 square kilometers and is home to a local population of some 552,500 people (DSEC 2010).

Macau‟s sovereignty officially reverted to the People‟s Republic of China in 1999, two years after the

much-reported „handover‟ of Hong Kong to the Chinese government. The territory is also one of the two

Special Administrative Regions (SARs) belonging to the People‟s Republic of China (with the other

SAR being Hong Kong), and is located on the western side of the Pearl River Delta, and borders

Guangdong Province to the north and faces the South China Sea to the east and south. The territory

currently is noted for its tourism and gaming (casino) industry.

SOCIOLINGUISTIC CONTEXT OF THE STUDY

According to the Basic Law of Macau, which provides the mini-constitution of the territory after 1999,

Chinese and Portuguese are specified as the official languages of the territory (Chinese Government

1993). According to the DSEC (2011) the „usual language‟ of the respondents refers to „the language an

individual mostly used at home‟. From the 2011 government census, Cantonese was specified as the

„usual language‟ of some 83% of the population (DSEC 2011). From the government census statistics on

languages in the territory, it would appear that the linguistic environment in Macau is dominated by

Cantonese, with only small numbers of citizens claiming to use other languages, such as Putonghua and

English, as a „usual language‟. However, as mentioned earlier, there are two designated co-official

languages, „Chinese‟ (undefined, as a general designation) and Portuguese, which is very much a

minority language in usage mainly in government and in law.

3

The purpose of this study is to specifically investigate the sociolinguistic dynamics of certain particles

in Cantonese, especially in how they convey different kinds of social information of speakers. These

particles are referred to Cantonese sentence particles (or utterance particles). Together with the tone

system of Cantonese, these sentence particles are considered to be a characteristic feature of Cantonese

and they can occur in the initial, medial and final position in sentences (Luke 1990; Matthews and Yip

1994; Leung 2005; Wong 2009; Leung 2011). However, this paper is only concerned with sentence final

particles (hereafter SFPs), as they show the most varied forms of pragmatic functions (Wee 2004).

According to Kwok (1984: 5), a sentence particle is a „linguistic form which does not occur as an

independent unit‟. Traditional views in the literature on Cantonese are that these SFPs only contain a

grammatical function to assist in the structuring of sentences (see Kwok 1984; Luke 1990). However,

recent studies have illustrated the range of discourse functions these SFPs can have in pragmatics (e.g.

Matthews and Yip 1994; Leung 2005; Wakefield 2011). For instance, Matthews and Yip (1994) have

suggested that the SFPs are related to „affective and emotional colouring‟, indicating that these particles

also function as speech act markers (see also Leung 2011). Generally, most studies on SFPs appear to be

concerned with the semantic functions of SFPs, and typical data analyses comprise of the corpora studies

(e.g. Leung 2005; Leung 2011), yet very little appears to have been said on how social variables

influence these pragmatic functions in discourse (see Kwok 1984; Luke 1990; Mathews and Yip 1994;

Wee 2004). Another shortfall on SFP studies appears to be the lack of sociolinguistic fieldwork which

explicitly investigates how social factors influence the distribution and use of these particles. Another

shortfall appears to be the focus of the semantic and/or pragmatic functions of these particles, and

accounting for these functions by ignoring the social variables which may impinge on the use of these

particles. One of the few exceptions, which include a discussion of social functions on the use of SFPs, is

Wong (2009), who studied the use of SFPs in computer mediated communication (i.e. online chats) to

signal specific social information between interlocutors (when using SFPs in English discourse). Wong‟s

(2009) study, however, does not indicate how certain variables (such as the use of English by her

subjects) impinge on the use of the SFPs. The study also tried to account for the functions of the use of

the SFPs without duly considering the influence of more general social variables of age, gender, and

social class, among others.

Although around 100 different SFPs and SFP clusters have been identified (Luke 1990; Yip and

Matthews 2000; Leung 2005, Leung 2011), only 30 or so are in basic form (Kwok 1984). Due to the

large number of SFPs, this study considers only a handful of some of the prominent particles, which is a

limitation of this study. It needs to be noted that studies on SFPs have most often considered some of the

pragmatic functions of these particles, and a summary of these functions are presented in Table 1.

4

Table 1 Classification of some SFP functions in the literature

SFP Functions Attitude Statement Question Command

laa1

啦

Statement

lack of forcefulness and definiteness

tentativeness

agreement sequence

eliciting positive response

non-committal answer

pre-closing

Command

persuasive function

solidarity marker

tentativeness

persuasiveness

X

X

lo1

咯

reluctance to talk

obviousness

signals a severe attitude

reluctance

obviousness

severe attitude

X

gaa4

嘎 confirmation marker assumptive X X

wo2

喔

to contrast an expectation

to emphasize „noteworthy‟ information noteworthiness X

aa4

呀

a replacement question marker

indicates a more severe message

to soften an utterance to make it more

natural

softening

X X X

(Adapted from: Wong 2009)

Despite the list of pragmatic functions that are proposed, it would appear there are more factors that

constrain the use of SFPs in contrast to the largely semantic functions summarized by Wong (2009). In

essence, variation and the sociolinguistic dynamics of the SFPs are ignored, and this study addresses this

shortfall on SFP studies by investigating the effect of social variables on the use of these particles. This

particular study only reports on 啦 [laa1]; 咯 [lo1]; 啊 [aa3]; 喔 [wo2]; 呀 [aa4]; and 嘎 [gaa4], and

how social variables influence the use of these particles. It also needs to be pointed out that many SFPs

have tonal variations, which are not discussed in this paper, but which may also indicate variations in the

use of the particles (see Leung 2005). Unfortunately, these tonal variations fall outside of the scope of

this paper. Bearing these considerations in mind, this paper proceeds to discuss the theoretical

framework of the study.

SOCIAL NETWORKS, CONSTRUCTIONIST APPROACHES, AND INDEXICALITY

Milroy (1980) has argued that a social network approach should be considered when the purpose of

research is to capture the dynamics underlying speakers‟ interactional behaviours, as opposed to

stratifying these behaviours as fixed social categories. Social networks, according to Milroy (1980) are

essentially a way to account for the way an individual speaker varies his/her linguistic behaviour in a

particular society or area, which is not possible with a large-scale analysis, such as the study by Labov

(1972) in New York. The study of social networks and the linguistic behaviour of people within specific

5

networks offers the opportunity to investigate more detailed information regarding variation, with

particular reference to the more „natural‟ language use by speakers.

Another perspective which is considered here is that of so-called „constructionist approaches‟. These

approaches are concerned with how social actors use linguistic and other cultural resources in ongoing

construction and re-construction of personal and group identity (see Rampton 1999; Schilling-Estes

2004). Schilling-Estes (2004) notes that constructionist approaches view (a) the notion of „identity‟ as

dynamic; (b) identity as multifaceted and multilayered; and (c) the notion of „agency‟ as important (i.e.

what people do with language).

Another concept used is that of „indexicality‟, which according to Silverstein (2003), is the semiotic

association between a linguistic form and a contextualized meaning. In other words, an „index‟ is when,

for example, a linguistic element „points to‟ a social variable. Silverstein‟s (2003) discussion of Labov‟s

(1972) taxonomy of linguistic variables („indicators‟, „markers‟, and „stereotypes‟) reveals how speakers

gradually become aware of how various linguistic forms are used by different people in different

contexts. For instance, first-order indexicality presupposes the existence of an identity or pragmatic

function, but shows no pattern of stylistic variation in a user‟s speech (see also Liao 2010). This relates

to Labov‟s notion of „indicators‟. Second-order indexicality shows that speakers come to notice that

specific variables are used by different people in different contexts „and their understanding of the

linguistic form may be reflected through their linguistic behaviour such as style-shifting or

hypercorrection‟ (Liao 2010: 49). According to Silverstein, this is what Labov (1972) refers to as

„markers‟. „Stereotypes‟ are a situation when a community is aware of a particular variable and share a

common knowledge of its (social) meaning. Located in contemporary Macau, this study draws on the

concepts of indexicality to explain how SFPs come to index social variables such as gender,

conversation topic and the affective relations between members in a particular social network. This

research also used the constructionist approach as well as social network theory with a view to

interpreting the meaning potential of second order indexes, such as SFPs. This research investigates

conversations between various speakers to explain how the linguistic variables index social variables,

both directly and indirectly.

METHODS

This paper is primarily concerned with aligning certain social variables with a selection of prominent

SFPs introduced in Section 2 above. Essentially, both qualitative and quantitative methods are employed

in the research with a view to account for the social and individual factors that may provide new ways of

understand the use of SFP use in Cantonese. This study employed two research approaches: one being a

survey which considers the impact of age and gender on the selected SFPs presented above; and the

other aspect of the study employed a social network approach, whereby social variables, as well as SFPs

6

are considered from within the research data itself and were not predetermined. The results of both

research methods were used to indicate correlations between SFPs over certain social categories.

In terms of the sampling of the survey, a stratified judgment sample was used. The main reason for this

is that the research is based on the choice of SFP for a specific group of people in Macau (Cantonese

speaking Macanese). To that end, only local residents (with at least one parent born in Macau), and who

speak Cantonese as a first language were allowed to be surveyed. In terms of stratifying the sample,

respondents were specifically chosen according to the following criteria:

i. Gender: A relatively equal number of men and women were interviewed

ii. Age: A range of people were chosen to fit the following age-groups: 18-29 age-group, 30-39 age-

group, 40-49 age-group, and the 50+ age-group

ii. Social Class: Social class was defined in terms of „education‟, „income‟ and „area of residence‟

In total, 58 respondents took part for this part of the study, and there were between 6 and 9 speakers in

each of the control categories of gender, age, Cantonese proficiency, and social class. The respondents

were first asked to provide their demographic details, and then the interviewees were given two cartoons

which show a conversation between various cartoon characters (of different genders). The respondents

were then asked to do a short discourse completion exercise, in which they had to verbally finish part of

a given dialogue for the cartoons presented. The given parts of the dialogues sought to elicit the particles

that are introduced in Section 2, and the particles that were used were noted down. In this elicitation, a

total of 116 particles were recorded, of which some 70% (92) belonged to the predetermined ones

introduced in Section 2.

In the social network study a person was selected who was well known to the researcher, and who

belongs to the typical „Cantonese speaking Macanese‟ designation. This person was used as the ego of a

network and all her associations were investigated. This person was interviewed on a number of

occasions and a list was made of all the people she had contact with over a period of two months. The

concern in the research, however, was only with obtaining her so-called „core‟ networks, or people with

whom she has had regular contact with. The consideration in the research was that her regular, close

associations could be used as a springboard for investigating how she uses Cantonese, and how her

language use changes according to certain contexts. Considering this, a slightly modified version of

Milroy‟s (1980) method of measuring network density was used. The „core subject‟ (hereafter Lia)

reported to have many friends and so it could be stated she has a „dense‟ social network. In fact, she

named at least 112 people she had had contact with over a two-month period. At the time of the research

she lived with her whole immediate family in the same flat, and she claimed to have strong ties of

kinship in her neighbourhood. Lia reported to work, study and socialize with at least two others of the

7

same gender. And finally, she also stated that she had voluntary associations with workmates, classmates,

and church members during her leisure time. However, the researchers felt that Milroy‟s (1980) method

of measuring network density did not reveal precisely the affective quality of Lia‟s friendships, as this

showed to be an important variable in an earlier pilot study. For example, a person may socialize with

one colleague, but he/she would not ask that person for any advice regarding serious matters. To solve

this problem, the affective quality of the relationships Lia reported to have with the people in her social

network was also considered in this research (for a discussion on affective quality in social networks see

Stoessel 2002). To that end, Lia was interviewed again and another list of questions was used which

helped establishing the affective quality of Lia‟s relationships.

Speech recordings of Lia‟s conversations with most of the people in her social network were

conducted, but people who were not part of the „dense‟ section of her network were not included. A total

of ten conversations were recorded for the social network part of the study, and just over four hours of

dialogue were collected for this research. Finally, it needs to be pointed out that contact frequency

(„multiplexity‟) was not considered as a social variable in this study. Multiplexity was only used to

identify the dense section of Lia‟s social network, as the researchers were only interested in gaining

access to the vernacular, as SFPs are generally used less frequently in formal situations and when

speakers do not know one another well. As already mentioned, for the social network study a number of

recording sessions were conducted with certain members of Lia‟s social network, and Lia was

accompanied to a number of social events and occasions where she regularly met her family and friends.

This was done over a period of a month, and she always introduced the primary researcher as a friend

who was interested in understanding more about the Cantonese language in Macau, and that they may at

times record their language use. As the researcher was introduced as a friend of the subject, people were

at ease with his presence, and some were initially interested in teaching him and talking about Cantonese.

Many of Lia‟s friends and family were also not able to speak much English. The researcher pretended to

know very little Cantonese, and in most situations, people ended up talking with each other in Cantonese

and left him to his own devices. It was during these times that a small recording device was produced for

the purpose of recording the conversations. After a few occasions no-one seemed to notice the recording

device anymore and the conversations normally continued without much attention paid to the researcher.

In total, 991 speaking turns were transcribed from the recorded data. A modified version of the discourse

transcription (DT) system was used when transcribing the recorded data (Du Bois 1991). Finally, the

social networks were analyzed using Ucinet (v. 6.282), which is software designed for social network

analysis, developed by Borgatti et al. (2002), and relevant sections of the dialogues captured were

analysed using the latest version of Praat (v. 5.1.34), which is a specialist speech analysis software

developed by Boersma and Weenink (2010). A limitation of this study concerns its lack of detailed

8

comparison with corpora data so prevalent in studies on SFPs. However, the reader is reminded that the

specific focus is on the social dynamics of SFPs.

SURVEY RESULTS

The survey used 116 counts of SFPs in the two different discourse completion tests that were given to

the respondents. Table 2 presents how the respondents used these particles in the discourse test, and it is

evident that the aa3 particle was used most frequently, with this particle being used 18.1% of the time.

This is followed by la and ga/gaa4 , with these particles used 14.6% and 13.7% respectively. In contrast,

the least frequently used particle was aa4 which was used only 8 (6.8%) times, and other SFPs were

used 20.6% of the time.

Table 2. Number of SFPs in the survey

Particle % and no.

[laa1]

[lo1]

[aa3]

[wo2]

[aa4

[gaa4]

Others

14.6 % (17)

12.9% (15)

18.1% (21)

12.9% (15)

6.8% (8)

13.7% (16)

20.6% (24)

No.= 116

Table 3 presents the use of the particles according to the variable „gender‟, and it is evident that the

female respondents appeared to use the lo1 and laa1 particles the most, with 80% and 64.7%

respectively using these particles. In contrast, men appeared to prefer the aa3 and gaa4, with 62.5% and

57% respectively using these SFPs. Men also appeared to show a greater variety in their choice of

particles, with 75% using another particle that was not part of the predetermined selection. None of the

other social variable presented any interesting findings.

Table 3. Number of particles used, according to gender

Particle Men Women

The reported use of SFPs according to gender, in % and no.

[laa1] (No.=17) 35.2% (6) 64.7% (11)

[lo1] (No.=15) 20% (3) 80% (12)

[aa3] (No.= 21) 57.1% (12) 42.8% (9)

[wo2] (No.=15) 40% (6) 60% (9)

[aa4] (No.=8) 12.5% (1) 87.5% (7)

[gaa4] (No.=16) 62.5% (10) 37.5% (6)

[other] (No.=24) 75% (18) 25% (6)

9

RESULTS OF THE SOCIAL NETWORK STUDY

For this part of the study a participant (i.e. Lia) and her associations were organized in terms of a social

network, which is based on „affective quality‟. As can be seen from Figure 1, Lia has seven people in her

first-order network zone, and two in her second-order network zone. All the people in Lia‟s first-order

network zone know one another, while only Lia knows those in her second-order network zone. After

the core network and the zones had been established, the affective quality of the ties was evaluated. The

affective quality of the network is also presented in Figure 1, and from the „ties‟ it can be seen that Lia

has the „strongest‟ relationships with Veronica, her father, and her mother. These relationships all have

value of „5‟. Lia also has strong relationships with Betty, Tinki and her sister, with a value of „4‟. The

value of Lia‟s relationships with her aunt, Rita and with Sharon measure 3 – the lowest of the network

ties. All the people in this network were interviewed and the affective quality of the networks between

all the members is also presented in Figure 2. From these results it is interesting to note that Lia

considers Rita a „closer‟ friend than Rita considers herself to be with Lia. In other words, Lia has a tie

strength of „3‟ for Rita, but Rita has „connection‟ of „2‟ for Lia. It is also noted that Rita and Sharon

know each other, but do not consider themselves „friends‟, and they have a mutual connection valued at

„1‟

Figure 1. Lia’s social network (with Lia as the ego of the network)

Second order zone First order zone

10

After Lia‟s „core‟ social network was established were many of her conversations with members in this

network recorded and then later transcribed. The transcribed sections of the recorded data include 15

minutes of the first recording, 13 minutes of the second recording, and 10 minutes for the other two

recordings. In total, around 50 minutes of speech was transcribed, and these transcriptions were done in

both Chinese script and Yale romanization with tone diacritics. A consideration for using this form of

romanization and not the popular Jyutping romanization (developed by the Linguistic society of Hong

Kong) is that this study considers Macau Cantonese a variety of standard Cantonese. Considering this, a

total of 991 speaking turns were analyzed for this part of the study. The conditions where the recordings

took place which were social and relaxed and these environments were not different from the ordinary

lives of the people in Lia‟s network.

The frequency distribution for the number of SFPs used is presented in Tables 4 and 5. From this it is

observed that the particles laa1, gaa4, and lo1 were the most frequently used. In addition, it appears that

some of the speakers had a preference for certain particles depending on context and whom they were

speaking with. For example, in the first recoded (sample), Lia has a high frequency of use for wo2, but in

the second sample she only uses the same particle once. However, in Samples 2, 3 and 4 Lia uses lo1

more frequently. Rita and Veronica both seem to prefer the use of laa1 and gaa4 in their conversations

with Lia. The SFP cluster „嘎(gaa4)啦(laa1)‟ was also detected, but not used very frequently in the

samples. This cluster was not included in the frequency counts displayed in this paper.

Table 4. Frequency distribution of the selected SFPs

Table 5. Frequency distribution of the selected SFPs among the speakers

Sample 1 Sample 2 Sample 3 Sample 4

Lia Tinki Betty Lia Veronica Lia Father Mother Sister Aunt Lia Rita

呀 (aa4) 4 17 2 9 14 8 1 2 2 1 6 8

嘎 (gaa4) 14 16 7 26 28 20 1 2 2 0 14 36

啦 (laa1) 19 14 7 37 26 30 6 4 8 0 25 50

喔 (wo2) 20 8 3 1 1 0 0 0 0 0 0 0

咯 (lo1) 12 16 7 23 17 20 0 1 5 1 12 10

Sample 1 Sample 2 Sample 3 Sample 4

呀 (aa4) 23 14 14 14

嘎 (gaa4) 37 54 25 50

啦 (laa1) 40 63 48 75

喔 (wo2) 31 2 0 0

咯 (lo1) 35 40 27 22

11

The perceived influence of context and relations between the speakers on the use of SFPs is further

confirmed by the fact that the topics discussed between the members of the network. The results reveal

that a greater affective quality of the relations between the speakers may lead to greater frequency in

linguistic variation. For example, in the first sample the conversation topics were of a very general

nature, as shown below:

L1-50 Lia's three-year-old tutorial student

L52-74 Nursery school issues in Macau

L76-98 Claims of luggage damages paid by airlines

L99-125 Travel experiences

L127-161 Finding employment

L255-321 Lia's school days – „sitting at the back of the classroom‟

L353-361 Discussion about S.H.E. (a Taiwanese pop group)

Variation of the SFPs is spread almost evenly over the topics in the first sample. However, in the

fourth sample, Rita discusses some issues related to „dating‟ and „men‟, and Rita seems to have a very

high count for some of the variables, in particular with the use of particles (laa1 and ga) when discussing

these topics. It is also Rita who leads the conversation when these topics are discussed. Consider the

comparison in Table 6 when Rita discusses the non-personal topic „books‟ and a more personal topic „a

date with an Italian guy‟. From these results it appears that conversation topic causes a sort of

ideological positioning on the part of the speaker which is reflected in the frequency use of some of the

SFPs.

Table 6. Frequecy distribution of Rita’s use of SFPs between two topics

Further to this, in the second sample, the following topics were discussed between Lia and Veronica:

L1-87 What food to order at the restaurant

L88-94 Places of interest / attractions

L97-108 Cathay Pacific airlines (sleeping on the plane / food on the plane)

L109-125 Water dispensers

L127-240 Lia's family (her sister and mother)

Topic: books (L34-52) Topic: dating (L53-80)

嘎 (gaa4) 0 嘎 (gaa4) 6

啦 (laa1) 3 啦 (laa1) 11

12

In the second sample, Lia and Veronica began by discussing non-personal matters. However, from

lines 127 to 240 (almost half of the discussion) Lia talked about personal issues relating to her family. In

this section of the sample, there was also a greater frequency in Lia‟s use of SFPs, as well as lengthening

of these particles (see below for more on vowel lengthening). For example, Lia used laa1 26 times when

talking about her family, compared with a total frequency count of 11 for all the other topics discussed in

this conversation. It needs to be noted that for the comparison mentioned above involving the more

personal „family‟ topic, all other topics had to be combined so that an almost equal number of speaking

turns could be evaluated. If this had not been done the frequency counts would have been

disproportionate. In none of the other samples does Lia talk about very personal issues.

It was also interesting to find that the vowels in the SFPs were often lengthened to further intensify

specific meanings, especially when a personal topic was discussed. To cite an example of this, consider a

selected part of the transcription from Sample 2 below:

160 Lia: yi1！"(ng)o5" a3 ba1 jung6 cheui1！

161 Veronica: sik1 saai3 "l(n)ei5" a3 mui6？

162

Lia: m4 hai6 lo1== "(ng)o5" a3 ba1？gan1 jyu6 "(ng)o5" wa2，“ "l(n)ei5" a1 "l(n)aau6"

ha6 keui6 la1= "l(n)ei5" tai2 ha6 keui6 gei2 cheui1 a1=！” gan1 jyu6 keui6 jau6 wa2

laa1，“hoi1！"l(n)ei5" jau6 bei2 go3 wai2 bei2 "l(n)ei5" ga1 je2 cho5=laa1==！”

gam1 wo4，gan1 jyu6 jung6 bei2 "(ng)o5" a3 mui6 "l(n)aau6" faan2 jyun2 tau4

wo4！

163 Veronica: ha1 ha1，"l(n)ei5" a3 ba1 a1？！

164 Lia: hai6 a1！

In the example above, the length for the first lo1 in 162 is 1.088 seconds. For the first laa1 in 162 the

particle length is 0.55 seconds. Vowel length for the third laa1 above in 162 is 0.67 seconds. Compare

this with the word (ng)o5 (I/me) in 162, which is on average 0.3 seconds in length. Pitch range and

intensity of the SFPs were not measured and may indicate specific meanings, but this falls outside of the

scope of this paper. Li (2006) has proposed that tonal variations indicate a change in the meaning

potential of SFPs, but as an internal constraint, this was not investigated in this study.

DISCUSSION

The social network study indicates that the affective relationships between the speakers (i.e. the affective

quality of the relationships), as well as the conversation topics may promote SFP variation. These

findings are in line with Milroy‟s (1980) assertion that the stronger the network ties, the more a person

will conform to the most natural speech (vernacular). Generally, it was found that when Lia was talking

to a person who scores high on the affective quality chart and about personal topics (e.g. her troubles

with her family), she used the lo1 particle more frequently. Further, there is a frequency count of 23 and

13

20 for Lia‟s use of lo1 in samples two and three respectively, compared with 12 in each of samples one

and four. The social network analysis indicates that variation is largely dependent on the individual

speaker concerned, and frequency of distribution of a particular variant or SFP increases when the

relations between the speakers are strong, and the topics under discussion are more personal.

In terms of the social network analysis, it was found to be relatively easy to gain access to ordinary and

natural spoken Cantonese, as it is normally used in Macau. Once the recordings were transcribed it was

also convenient to generate specific SFPs which could be scrutinized in specific contexts. This latter

advantage is important as many more external linguistic variables can be generated and considered as

well, besides any predetermined ones. An example in this research is the notion of „affective quality‟,

which considers the strength of the relations between the interlocutors, and how this may lead to

variation. In previous studies and discussions on discourse it has long been suggested that the

relationship of the interlocutors influences language style (see Schiffrin 1994; Cook 1989; and Bell 1984;

Eckert 2000), but here further evidence in the network analysis illustrates that this can be measured

quantitatively for SFP variation. For example, it was shown that Lia considers Veronica to be very close

friend (with an affective quality score of 5), and in their conversations linguistic variation is common,

compared with less variation in Lia‟s conversations with her good friends Tinki and Betty (with lower

affective scores). Other categories such as conversation topic may also be carefully scrutinized using a

social network approach and the frequency of the linguistic variations seems to increase when the topic

is more personal for the person. An example of this was briefly mentioned above where Lia was talking

to Veronica about her family issues.

In terms of Labov‟s (1972) taxonomy and the SFPs that were studied in this research, it is necessary to

first consider whether the selected SFPs conform to Wong‟s classification introduced earlier, and to offer

further explanations that are not covered in this classification (cf. Table 1). From this study there appear

to be more factors that constrain (or play a role in) the use of SFPs than the largely semantic functions

summarized by Wong (2009). In essence, the semantic and general pragmatic functions presented in

Table 1 ignore variation and the sociolinguistic dynamics of the SFPs. Specifically, the significant use of

lo1 (p < 0.05) and aa4 (p < 0.05) by females in the distributionist data presented above suggests that

gender is a key variable that constrains the use of this variable. From the results shown in Table 2 it is

evident that men used a greater repertoire of SFPs in the discourse completion exercise besides the ones

that were preselected for investigation. In addition, the social network study indicates that there may be a

correlation between conversation topic and the frequency of distribution of a particular SFP in a

discourse context. More personal topics, such as Rita‟s discussion of books and dating indicate that her

use of SFPs (such as that of gaa4 and la1) increases with the discussion of more personal topics, or

topics which indicate a sort of ideological positioning. In addition to conversation topic, it would appear

that „affect‟ constrains the frequency of SFP distributions as well, with greater affect appearing to

14

promote greater SFP distribution. An example of this is Lia‟s use of ga, laa1 and lo1 in the four samples

(see Table 7), where she uses these SFPs with greater frequency in samples 2 and 3.

Table 7. Lia’s use of ga, laa1 and lo1 in the four recordings

Sample 1 Sample 2 Sample 3 Sample 4

嘎 (gaa4) 14 26 20 14

啦 (laa1) 19 37 30 25

咯 (lo1) 12 23 20 12

The functions presented in Table 1 are evidenced in many of the conversational uses of the said

particles, as when Lia uses laa1 as a „solidarity marker‟ in her conversation with Veronica, indicated in

Example 1 below. However, a constructionist interpretation of the use of this particle in the context of

the conversation indicates that this particle indexes more social functions than only the „solidarity‟

function proposed in Table 1.

Example 1: Discussion between Lia (A), Veronica (B). lines 184-186 of sample 2

A: gan1 jyu6 le2，hou2 laa1.. gan1 jyu6 (ng)o5 mou4 l(n)aau6 keui6 a1 ma4 gwo2 chi3

185 B: keui6 goi2 bin3

A: gan1 jyu6 yin4 ji1 hau6 (ng)o5 jau6 tung4 (ng)o5 .. lou5 dau2 tung4 (ng)o5 a3 ma1

gong2，“aai1 a1，m4 goi1 l(n)ei5 a1 tung4 l(n)ei5 go3 l(n)eui5 gong2 ha6 laa1，

hou2 faan4=a1=keui6==mat1 mat1 mat1 a1 gam1！ gan1 jyu6 le2 (ng)o5 lou5 dau2

jau6 wa2 laak3，aai1！gam1 l(n)ei5 mai1 (ng)ok3 D= lo1==！” gam1 wo4，gan1

jyu6 (ng)o5 wa2，“ l(n)ei5 m4 tung4 keui6.. l(n)aau6 ha6 keui6？” gan1 jyu6 keui6，

“aai1！dak1 laa1 dak1 laa1！aai1！jan6 gaan1 tung4 h(k)eui6 gong2=ha6=laa1=！”

gan1 jyu6..jau6.. jung2 ji1 jau6 chiu1=chiu1 wo4 sin6，“aai1！l(n)ei5 jau6 m4 hou2

de1 saai3 l(n)ei5 ga1 je2 laa1==” gam1，gan1 jyu6 (ng)o5 a3 mui6 jau6 l(n)aau6

faan2 keui6，

B: [da1 keui6]

Specifically, in Example 1 above, Lia talks about her family issues and uses laa1 frequently to signal

solidarity with Veronica, as Lia is complaining about her sister and appears to want Veronica‟s

sympathy and agreement regarding her attitude towards her sister. The solidarity marker in this instance

may be seen as a realization of the affective quality of the relationship between the speakers, and as a

result the use of laa1 here can be seen as instance of „markers‟ and „stereotypes‟. When considered in

this light, as a stereotype a SFP typifies a specific utterance as a characteristic of a social category (or a

set of social categories), but as a marker it conveys additional social information that indicates salient

features of speaker identity, such as gender and affect2.

The use of lo1, on the other hand, may indicate a different function from the ones proposed by Wong

(2009) in Table 1, but before this function is discussed some theoretical issues that were presented above

15

need to be repeated here. Some linguistic variables are produced with a speaker‟s knowledge (conscious

or unconscious) that these variables will be interpreted in a certain way. For example, if a speaker is

talking about a personal issue to a close friend, it is would be the norm to use the most natural language

for that particular speaker. Thus, the relevance for this interpretation and production of varied language

use may be the result of larger social processes that can be said to have a sort of normative function. The

opposite may also be true, that a speaker may seek to engage in an ideological move to either distance or

connect herself with another speaker by using a particular SFP, in a particular way. To illustrate these

latter points further consider Example 2 from the social network recordings:

Example 2: Discussion between Lia and Veronica. lines 160-162, of sample 2

160 Lia: yi1！(ng)o5 a3 ba1 jung6 cheui1！

161 Veronica: sik1 saai3 l(n)ei5 a3 mui6？

162

Lia: m4 hai6 lo1== (ng)o5 a3 ba1？gan1 jyu6 (ng)o5 wa2，“l(n)ei5 a1 l(n)aau6 ha6

keui6 laa1= l(n)ei5 tai2 ha6 keui6 gei2 cheui1 a1=！” gan1 jyu6 keui6 jau6 wa2

laa1，hoi1！“l(n)ei5 jau bei2 go3 wai2 bei2 l(n)ei5 ga1 je2 cho5=laa1==！”

gam1 wo4，gan1 jyu6 jung6 bei2 (ng)o5 a3 mui6 l(n)aau6 faan2 jyun2 tau4

wo4！

160 Lia: Hey, my dad‟s even worse!

161 Veronica: (He) only likes your younger sister?

162

Lia: Not really, my dad? And I said (to him), „you need to try to discipline her, look at her,

and see how bad she is‟. Then my dad said, „give a seat to your older sister‟, and just

like that, afterwards, my sister was rude towards my dad.

(Own translation)

Lia‟s response (in 162) to Veronica‟s question is „no‟, yet she intensifies her expression with the use of

a lengthened lo1 (1.088secs). According to Wong‟s classification in Table 1, this particle colloquially

signals reluctance, obviousness or a severe attitude, and in her study on computer mediated discourse it

further signals reason and a reluctance of a speaker to connect with another speaker in addition to

obviousness. However, when the meaning potential of lo1 is considered in the particular context of Lia‟s

conversation – that is, with the topic under discussion (Lia‟s family), the conversation context (i.e.

setting, relation between interlocutors, etc.) – it can be concluded that there is no pre-conditioned social

agreement to the meaning of the use of this lo1. Wong (2009) and others would claim that „reluctance‟,

„obviousness‟ or a „severe attitude‟ is the pragmatic function of this particle, but Lia uses the particle not

to distance herself from Veronica. On the contrary, Lia uses this particle as a resource to change the

meaning of „no‟ to „not really‟, so as to soften the disagreement and to keep her social bond with

Veronica. This is in contrast to the established functions that have been reported on the use of this

particular SFP in Cantonese. In other words, both the social variables of topic and the affective quality of

the relations, as well as the particle lo1 can thus be said to be a specific resource for Lia (and perhaps for

16

Veronica) for identity (re)construction. The particle lo1 specifically does not contain meaning in the

classified sense yet it‟s very clear that it has a social function which is directed by the independent

variables of topic and affective quality. It is our view here that the pragmatic classification in Table 1

would not be able to account for this variation. Finally, it is difficult to state from the results of this

research whether Cantonese-speaking Macanese are socially drawn together, or socially distanced by

their use of SFPs. From the social network study it would appear that this is the case within a high-

density social network where the affective quality between the interlocutors is high3.

CONCLUSION

The results of this research indicate that additional information is conveyed through the use of SFPs in

Macau Cantonese. Specifically, social information pertaining to individual identity seems to be realized

by the use of SFPs in specific speech contexts. The contexts which may impinge on the use of SFPs

include:

i. the affective quality between speakers,

ii. conversation topic (personal vs. non-personal),

iii. gender, and

iv. possibly the network density of speakers in a social network

Results of this research suggest that SFPs in Macau Cantonese are a resource for communicating

speaker identity. From the data gathered in this research it is shown that the frequency of the use of SFPs

is greatest when the affective quality between interlocutors is high and the conversation topic is more

personal. It would be the purpose of a future study to further investigate the interplay between personal

identity and interpersonal connections and the way in which SFPs realize the variations that occur in

establishing and maintaining interpersonal relationships in various contexts.

END NOTES

1. The paper is based on some of the key findings that were reported in Botha‟s (2012) Masters

Dissertation on language variation in Macau Cantonese.

2. This raises the issue of how „affect‟ is related to so-called „persona creation‟ in discourse. Future

studies on correlating affect and identity in discourse may want to explore explicit links between affect

and identity further, with the view to establishing exactly to what extent affect could be related to

speaker identity and persona creation.

3. Future research may also want to compare SFP distribution between high and low-density networks. It

is possible that only people in high-density networks (and possibly also between first and second order

17

network zones in such networks) use these linguistic variables with greater frequency, compared with

less-frequent variation in low-density social networks.

REFERENCES (ADD NEW REFS)

Bell, A. 1984. “Language Style as Audience Design.” Language in Society (13): 145–204.

Boersma, P., and D. Weenink. 2010. Praat: doing phonetics by computer. http://www.praat.org/.

Borgatti, S.P., M.G. Everett, and L.C. Freeman. 2002. Ucinet 6 for Windows: Software for Social

Network Analysis. Harvard: Analytic Technologies.

Botha, W. 2012. “Dimensions in Variationist Sociolinguistics: a Sociolinguistic Investigation of

Language Variation in Macau”. MA dissertation, Pretoria: University of South Africa.

Cook, G. 1989. Discourse. Oxford: Oxford University Press.

Chinese Government. 1993. The basic law of the Macao Special Administrative Region of the People’s

Republic of China. Macao: Legal Affairs Bureau.

Du Bois, J.W. 1991. “Transcription Design Principles for Spoken Discourse Research.” Pragmatics 1 (1):

71–106.

Eckert, P. 2000. Linguistic Variation as Social Practice. Oxford: Blackwell.

Kwok, H. 1984. Sentence Particles in Cantonese. Hong Kong: Centre of Asian Studies: The University

of Hong Kong.

Labov, W. 1972. Sociolinguistic Patterns. Philadelphia: University of Pennsylvania Press.

Leung, C.S. 2005. A Study of the Utterance Particles in Cantonese as Spoken in Hong Kong. Hong Kong:

Language Information Sciences Research Centre, City University of Hong Kong.

Leung, H.L. 2011. “The Semantics of the Cantonese Utterance Particle laa1”. Paper presented at the

42nd Australian Linguistic Society Conference, Australia.

Li, B.Y. 2006. “Chinese Final Particles and the Syntax of the Periphery”. PhD Thesis, Utrecht: Leiden

University.

Liao, S.W. 2010. “Identity, Ideology, and Language Variation: A Sociolinguistic Study of Mandarin in

Central Taiwan”. PhD Thesis, Davis: University of California.

Luke, K.K. 1990. Utterance Particles in Cantonese Conversation. Philadelphia: John Benjamins.

Macao Special Administrative Region Statistics and Census Service (DSEC) (2011) Usual languages in

Macao. http://www.dsec.gov.mo/Statistic.aspx?NodeGuid=8d4d5779-c0d3-42f0-ae71-8b747bdc8d88

Macao Special Administrative Region Statistics and Census Service (DSEC) (2010) Yearbook of

Statistics, 2009. http://www.dsec.gov.mo/Statistic.aspx?lang=en-US.

Matthews, S, and V Yip. 1994. Cantonese: A Comprehensive Grammar. London: Routledge.

Milroy, L. 1980. Language and Social Networks. 2nd ed. Oxford: Blackwell.

Rampton, B. 1999. “Styling the Other: Introduction.” Journal of Sociolinguistics 3 (4): 421–427.

Schiffrin, D. 1994. Approaches to Discourse. Oxford: Blackwell.

Schilling-Estes, N. 2004. “Constructing Ethnicity in Interaction.” Journal of Sociolinguistics 8 (2): 163–

195.

Silverstein, M. 2003. “Indexical Order and the Dialectics of Sociolinguistic Life.” Language and

Communication 23: 193–229.

Stoessel, S. 2002. “Investigating the Role of Social Networks in Language Maintenance and Shift.”

International Journal of the Sociology of Language 153: 93–131.

Wakefield, J.C. 2011. “The English Equivalents of Cantonese Sentence-final Particles”. PhD Thesis,

Hong Kong: The Hong Kong Polytechnic University.

Wong, Y.T. 2009. “The Linguistic Function of Cantonese Discourse Particles in the English Medium

Online Chat of Cantonese Speakers.”. MA dissertation, New South Wales: University of Wollongong.

Wee, L. 2004. “Singapore English: A Grammatical Description.” In Reduplication and Discourse

Particles, edited by L Lim, 105–126. Philadelphia: John Benjamins.

APPENDIX

Transcription Key

18

:

[ ]

.

?

\

/

^

`

=

…(N)

…

..

(0)

( )

(H)

@

<Q Q>

/ /

<X X>

1-9

Speaker turn

Speech overlap

Final

Appeal

Falling tone

Rising tone

Primary accent/stress

Secondary accent/stress

Lengthening

Long pause

Medium pause

Short pause

Latching

Linguistic variables

Audible inhalation

Laughter

Quotation quality

Phonetic transcription

Uncertain hearing

Tones (Adapted from: Du Bois, 1991)