The Prosodic Transcription of a Corpus of Hong Kong English: 
Collection Criteria, Transcription System and Preliminary Findings 
 
Shiao Ying Sharon Chu 
Department of English 
The Hong Kong Polytechnic University 
egchusy@inet.polyu.edu.hk 
 
 
 
 
 
Abstract 
 
This paper describes the prosodic 
transcription of a corpus of Hong Kong 
English and some preliminary findings on 
the communicative role of intonation in 
Hong Kong English. 
 
1 Introduction 
 
As an important feature of the communication of 
meaning in English, intonation of other non-native 
English varieties have also begun to receive 
interest and attention. At the Hong Kong 
Polytechnic University, we are attempting to 
describe the communicative role of intonation in 
Hong Kong English. By uncovering the similarities 
and differences of intonation patterns between 
native and Hong Kong speakers of English, we 
will be in a position to describe how Hong Kong 
speakers make use of intonation to communicate 
meanings, and find out whether or not there is a 
systematic pattern which we can name “The 
Intonation system of Hong Kong English”. The 
findings will hopefully better inform our 
understanding of Hong Kong English and help the 
development of English language learning and 
teaching materials directed at discourse intonation. 
 
2 Our Data 
 
Spoken discourses between native speakers of 
English and Hong Kong Chinese have been 
recorded as a first step to build up the Hong Kong 
Corpus of Spoken English (HKCSE). The HKCSE 
is divided into four sub-corpora: conversations, 
business discourses, academic discourses and 
public discourses. Each sub-corpora consists of 50 
hours of naturally-occurring discourses. The 
participants are monitored in terms of age, gender, 
nationality, occupation, educational backgrounds 
and residence overseas. 
 
3 The Transcription System 
 
Data collected are first orthographically 
transcribed. The intonation system proposed by 
David Brazil (1997) known as the Brazil System is 
being used for the prosodic transcription of data 
manually. This is a discourse-based system that 
suits our need of describing Hong Kong intonation 
along with discourse features. Below are the labels 
adopted from the Brazil System of intonation: 
 
 
Prominence: Capitalised letters 
Tone: p (fall), r (fall-rise), p+ (rise-fall), r+ 
(rise), o (level), u (unfinished utterances) 
Key and termination: hl and ll, stands for high 
key /termination and 
low key/termina-tion 
respectively  
Tone unit boundaries: // 
 
 
Prominence refers to the stressed syllables. 
Tone is the tonic or nucleus in a tone unit (i.e. the 
last prominent syllable of a tone unit), and either 
the letters are capitalised or underlined, the latter is 
for cases in which more stressed syllables follow 
the tonic. Key and termination refer to the key of 
the first stressed syllable and the tonic respectively. 
       Philadelphia, July 2002, pp. 11-14.  Association for Computational Linguistics.
                  Proceedings of the Third SIGdial Workshop on Discourse and Dialogue,
A system for computer labelling has also been 
developed: 
 
 
 
Fig 1. Computer speech file of sample 
conversation between a male native speaker and 
male non-native speaker 
 
The program used for the purpose of computer 
transcription is ‘xwaves’, a package developed as a 
research tool for all kinds of acoustic analysis. The 
recording process is followed by this computer 
labelling using Brazil’s paradigm along with the 
orthographic transcription. As can be seen from the 
above speech file, orthographic transcriptions are 
put in alternate columns, with the intonation 
notations above. Only the capitalised onset syllable 
(the first prominent syllable of the tone group) and 
the tonic (indicated by p, p+, r, r+ or o) of each 
tone unit are inserted. 
Statistical analysis has also been carried out for 
part of the conversation sub-corpus of 
conversations: 
 
Table 1: Crosstabulation of tones used by native 
speakers (NS) and Hong Kong speakers (NNS) 
 
Tone p r p+ r+ o u* Total 
NS 308 90 92 142 208 35 875 
NNS 559 25 24 106 796 43 1543 
Total 867 105 116 248 1004 78 2418 
 
* unfinished utterances 
 
It can be seen that there is a dominant use of 
the level tone o and proclaiming tone p among 
Hong Kong speakers of English. Contrast is also 
seen in the use of the referring tone r between the 
two parties. Native-speaker data show more 
instances of the referring tone. Another referring 
tone r+ has more or less the same frequency of 
occurrence in the two different sets of data. All 
these will be viewed along with representative 
examples extracted from the data concerned in the 
sections that follow. 
 
4 Description of Tone Choices 
 
In Brazil’s (1997) system of native-English 
intonation, R and P stand for the referring and the 
proclaiming tone respectively. They are further 
subdivided into r and r+ (for R), and p and p+ (for 
P).  
A: What will you do on your day off? 
 
B: // r WHEN i’ve prepared my LECture // r 
if there’s any TIME left // p i shall GO into 
TOWN // r and AFter THAT // p it will 
dePEND on the WEAther // p perHAPS i 
shall play TENnis// r if its FINE // r and if 
there’s anyone aROUND // r OTHerwise // 
p i’ll WRITE some LETters // (Brazil 
1997:73) 
 
This example shows the use of a fall-rise r to 
indicate negotiated information/assumed common 
ground, and p to signal new information. This 
stands for the P/R opposition proposed by Brazil 
(Brazil, 1997). 
 
5. Preliminary Findings 
 
Preliminary results from the conversation sub-
corpora show that the level tone is dominant and is 
used in almost half of the tone units. Below is a 
typical example: 
 
A: // p WHAT’s the REAson // p for reCORding 
us// 
 
b: // p YEA // p SHE is DOing a phD PROject // 
o ON ER // u ENglish // o ER// p TEAching // 
o in FACT ER // u IT IS ON THE aNAlysis 
OF THE // p for eXAMple // o converSAtion // 
o HOW USually ER er CHINESE // p ER the 
USE of WORDS // o FOR the ORdinary 
converSAtion // u and THEN THE // o er i i 
think it’s MAINly THE // o ER // p ENglish 
Usage // u MAYBE SHE is GOing TO 
anaLYSE MY ER // o in ONE SENTENCE // 
p ER how i STRUCture MY WORDs // 
 
At points where the speaker is obviously 
signaling shared information, or information 
already present in the consciousness of the two 
parties (e.g. conversation, ordinary conversation, 
in one sentence), the speaker uses the level tone o 
instead of the referring fall-rise r. 
  
The use of proclaiming tone is also dominant. 
The next example presents us with a series of falls 
(p): 
 
a: // o beCAUSE // r IN PRINciple // p we should 
ASK the conTRACtor // p PROject 
conTRACtor // p the MAIN conTRACtor // o 
to DO it// p to TOUCH up ALL THESE 
FAULTs // 
 
The “contractor” mentioned in the above 
chunks should be understood as already present in 
the consciousness of the hearer, for the 
conversation parties have been discussing 
construction matters. Subsequent “contractors” 
would become old information which should be 
used with a referring tone according to Brazil. But 
here they are all said with a proclaiming tone p. 
 
5.1 The Intonation of Questions 
 
It is claimed by Brazil that when native English 
speakers echo what they have just heard or ask to 
have an assumed common ground confirmed or 
refuted, they use either a fall p or a rise r+. Below 
are examples from the Hong Kong English data: (a: 
Hong Kong male speaker, b: Hong Kong female 
speaker) 
  
E (echoing questions) 
 
b: // r+ CHANging techNOlogy // 
a: // p JUST ONCE or TWICE TIMEs // 
a: // r+ caSIno GROUP // 
a: // p two DAUGHters // p in UK // 
b: //r+ not BAD // 
a: // p you WANT to GET an ANSwer // 
 
A (assumed common ground) 
 
b: // r+ your GOOD friend // 
a: // p so you you you will GO as WELL // 
a: // p OH // r+ you LIKE THIS er // p you LIKE 
this // 
a: // r+ say LIKE er RUSsel STREET // 
b: // p YOU’re NOT REALly INteresTED in 
LEAving HONG KONG // 
a: // r+ IT’S BETter // 
 
Preliminary results show that Hong Kong 
speakers use both p and r+ in asking questions as 
native speakers do. 
Further, we have examined examples of yes/no 
questions and information questions. This time the 
P/R opposition in terms of information status 
comes into play. According to Brazil (1997), p is 
used when the question content is not yet the 
negotiated common ground. On the other hand, r 
and r+ are used when the question is put forward 
as a presumed assumption. Some examples of the 
transcribed data are as follows (Q stands for 
“genuine questions”, meaning the speaker is asking 
for information unknown to him/her: 
 
Q (Information questions including Wh-questions): 
 
a: // p HOW to SPEND // r+ toNIGHT // o HOW 
to SPEND // r+ to SPEND toNIGHT // p HOW 
to SPEND toNIGHT // (Q) 
 
a: // p YEA // o WHAT // p what KIND of 
PROblem // (Q) 
 
b: // o and you’ll be LEAVing TO // o UK // p as 
WHEN // 
 
b: // o HOW about the overALL // o UM // p 
PASSing RATE // p for THIS YEAR // 
 
Q (Yes/no questions) 
 
a: // r+ DO you // r+ DO YOU // o LIKE //r to 
EAT the INside BODY // p I MEAN the // p 
Animal INside BOdy // r+ DO you GET the 
MEAning // 
 
b: // r CAN you suGGEST a CLEver WAY TO // 
p to FIgure OUT // 
 
b: // o BUT UM // o do you MIND if // p we 
JUST SPEND FIVE or TEN MInutes // 
 
b: // o SO // o JACK // p have YOU SEEN THIS 
ONE // 
 
b: // o HOW aBOUT THE // u ORgaNI- // r+ AS 
a COMmiTEE MEMber // 
 
P: (including declarative mood questions) 
 
b: // o is IT BE BEcuase DUE TO the // r+ ALL 
THE CRACKS // 
 
a: // p and HOW’S GOing with the ONE that 
YOU you TALKED about LAST TIME // r+ 
that is the (.) LIKE er INStitutional BUILdings 
is it // 
 
a: // r+ WORK DOing IN // r+ DOing FOR // o 
you MEAN //  
 
b: // r+ your GOOD friend // 
a: // r+ IT’S BETTER // 
 
One point worth noting is the fact that speakers 
tend to use a fall p more often in Wh-questions. 
This is in line with Brazil’s (1997) proposal that 
this type of question is asked using a fall because 
the speaker has no idea about the answer. A rise r+ 
is associated with a considerable number of 
questions that carry presumed assumptions. 
However, as a general picture, it is not yet clear 
whether Hong Kong speakers actually manipulate 
the P/R opposition in asking questions as native 
speakers do, since we do have instances of both the 
P and R tone within the same category of questions. 
This matter is further complicated by the 
overwhelming use of level tones (Please refer to 
table 1 above) which are frequent in questions as 
well: 
 
a: // o DO you LIKE CHInese GOOD // 
 
a: // o ARE you also WORKING FIVE DAY 
PER WEEK // 
 
b: // u WILL IT AUTO // o AUTOMAtically // o 
ER CAN GO TO THE // o TO THE SOUTH 
AfriCA CONferENCE // 
  
Use of the fall rise r is emphasised particularly 
by Brazil (1997) as a means to indicate a presumed 
common ground in asking questions. But r rarely 
occurs in our Hong-Kong-speaker data. The 
proclaiming tone p and the rising tone r+ are the 
dominant tones for questions. 
 
6 Conclusion 
 
Preliminary results suggest that Hong Kong 
speakers do use the full variety of tones in 
speaking English. However this pattern of usage 
seems to differ from that of NS (e.g. r is much 
rarer and the level and proclaiming tones are 
dominant). They seem to manipulate the choice in 
the same way as native speakers in some of the 
sub-categories (e.g. Wh-questions). However, we 
have yet to establish whether Hong Kong speakers 
make the same conscious distinction between p 
and r. NNS’s frequent use of r+ in questions, for 
example, may be due to a non-native speaker 
intuition that a rise is the normal “question tone” in 
English. Also, if future results reveal that there is 
an overwhelming use of a fall p and a level o in 
whatever situation, then we have no reason to 
believe that there exists a native-like distinction.  
On top of the main concern regarding whether 
Hong Kong Chinese speakers deviate from native 
speakers in terms of the use of tones (and if so, 
how), it will be interesting to investigate also 
whether Hong Kong Chinese speakers have 
developed a Hong Kong English intonation system. 
If so, it will better inform us the way to design 
teaching materials geared towards discourse 
intonation in English. 
 
Acknowledgements 
 
The work described in this paper was substantially 
supported by a grant from the Research Grants 
Council of the Hong Kong Special Administration 
Region (Project No. BQ 396 PolyU5270/00H). 
 
References
Brazil David (1997). The Communicative Value of 
Intonation in English. Cambridge University 
Press: New York. 
