3rd International Symposium on Teaching English at Tertiary Level Hong Kong, 9-10 June 2007 Jointly organised by: Department of English, The Hong Kong.

3rd International Symposium on Teaching English at Tertiary LevelHong Kong, 9-10 June 2007

Jointly organised by: Department of English, The Hong Kong Polytechnic University

Department of Foreign Languages, Tsinghua University, Beijing, China

Chris Greaves and Martin WarrenEnglish Department

The Hong Kong Polytechnic University

A corpus-driven approach to learning and teaching the

communicative role of discourse intonation

The Hong Kong Corpus of Spoken English (HKCSE)

The HKCSE consists approx. 2 million words made up of four sub-corpora:

Conversations Academic discourses Business discourses Public discourses

Each sub-corpus consists of 50 hrs of naturally occurring data (i.e. a total of 2 million words). Half of all of the sub-corpora have also been prosodically transcribed – approx. 1 million words.

Discourse intonation choices available to speakers

System Choice

Prominence: prominent/non-prominent syllables

Tone: rise-fall, fall, rise, fall-rise, level

Key: high, mid, low

Termination: high, mid, low

(Adapted from Hewings and Cauldwell 1997: vii, in Brazil 1997)

Example imPRESsive a: // SO much MORE // // than it REALly IS //

Example (computer readable)

a: { \ [ SO ] much MORE < ^ imPRESsive > } { \ than it [ REALly ] < IS > } (HKCSE, prosodic)

The prosodic notation system Tone group boundaries are marked with ‘{ }’ brackets. The referring and proclaiming tones are shown using combinations of

forward and back slashes: rise ‘/’, fall-rise ‘\/’, fall ‘\’, and rise-fall ‘/\’.

Level tones are marked ‘=’ and unclassifiable tones ‘?’.

Prominence is shown by means of UPPER CASE letters.

Key is marked with ‘[ ]’ brackets, high key and low key are indicated with ‘^’ and ‘_’ respectively, while mid key is not marked (i.e. it is the default).

Termination is marked with ‘< >’ brackets with high, mid, and low termination using the same forms of notation used for key choices.

What the textbooks say about intonation

Often the textbooks conflate intonation with pronunciation.

Intonation is portrayed as having fixed attitudinal meanings.

Intonation is usually portrayed as simply a means to sound ‘lively’ as opposed to speaking with a ‘flat and monotonous’ intonation that sounds ‘boring’.

Intonation patterns, if described at all, tend to oversimplify or are simply incorrect – for example, yes/no questions are spoken with a rise tone and wh-questions are spoken with a fall tone.

Discourse Intonation

Intonation does not convey fixed attitudinal meanings.

Intonation choices are not determined with reference to grammar.

Intonation choices are situation-specific decisions to add additional meaning to what is being said.

(Brazil, 1997)

Searching the corpus

Example of searches for discourse intonation using the iConc search engine written by Chris Greaves.

Topic Development

One way that a speaker can indicate to the

hearer(s) that she/he is about to develop the

topic in a new direction is by means of

discourse intonation.

The HKCSE (prosodic) can be used to

illustrate this with real world examples, along

with other functions of discourse intonation.

Implications

Need for a shared understanding of intonation and its communicative role.

Need to adopt a standard format for representing and interpreting intonation.

Students and teachers should be exposed to naturally occurring discourse to learn how English intonation functions.

The HKCSE (prosodic) provides a useful resource to explore the full contribution of intonation.

References

Brazil, D. 1997. The Communicative Value of

Intonation in English. Cambridge:

Cambridge University Press.

Sinclair, J. McH. 2004. Trust the Text.

London: Routledge.

Tones in wh-questions

Total

Academic Discourse

87(88.8%)

1(1%)

0 1(1%)

9(9.2%)

98(33.7%)

Business Discourse

40(66.7%)

0 2(3.2%)

8(13.3%)

10(16.8%)

60(20.6%)

Conversation 49(66.2%)

3(4.1%)

2(2.7%)

8(10.8%)

12(16.2%)

74(25.4%)

Public Discourse

51(86.3%)

0 1(1.7%)

0 7(12%)

59(20.3%)

Total 227(78%)

4(1.4%)

5(1.7%)

17(5.8%)

38(13.1%)

291(100%)

Tones in yes/no-questions

Discourse type

Total

Academic Discourse

31(38.8%)

0 1(1.3%)

31(38.8%)

17(22%)

80(20.1%)

Business Discourse

34(25.7%)

1(0.8%)

7(5.3%)

54(41%)

36(27.1%)

132(35.6%)

Conversation 23(28%)

0 2(2.4%)

36(44%)

21(25.6%)

82(21.5%)

Public Discourse

29(33.3%)

0 1(1.1%)

14(16.1%)

43(49.5%)

87(22.8%)

Total 117(30.7%)

1(0.26%)

11(2.9%)

135(35.4%)

117(30.7%)

381(100%)

NOUNS

0

100

200

300

400

500

600

700

800

people sector system person

Occ

urr

ence

sProminences

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

people sector system person

Per

cent

age

CONJUNCTIONS

0500

10001500200025003000350040004500500055006000

and but though

Occ

urre

nces

Prominences

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

and but though

Per

cen

tag

e

VERBS

0

100

200

300

400

500

600

700

think see try learn

Occ

urr

ence

sProminences

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

think see try learn

Per

cen

tag

e

PREPOSITIONS

0

1000

2000

3000

4000

5000

6000

in with through

Occ

urr

ence

sProminences

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

in with through

Per

cen

tag

e

MODAL VERBS

0

100

200

300

400

500

600

700

800

can may might

Occ

urr

ence

sProminences

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

can may might

Per

cen

tag

e

MODALS

0

200

400

600

800

1000

1200

will can would should must may could might shall

Oc

cu

rre

nc

es

Prominences

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

will can would should must may could might shall

Per

cen

tag

e

Possible explanations for the inverse relationship between word frequency and prominence

The more frequent the word, the more functions it typically performs and so the more likely it is that its meaning is determined from the wider context of interaction rather than by the word in isolation.

The notion of ‘phraseology’ (Sinclair, 2004) - that language is typified by extended collocations representing units of meaning rather than by individual words - further supports the inverse relationship we have found.

Many of the grammatical words are the sole occupant of the tone unit and are thus prominent. In this role they appear to have an organisational function linking propositional content.

Nouns that are modified are less likely to be made prominent.

Nouns such as people typically collocate with other items such as many, more, most and it is the latter which tend to be made prominent.

Other nouns such as proposal(s) are less likely to be modified and so more likely to be made prominent by the speaker.

CONJUNCTIONS

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

5500

6000

and but also because however

Occ

urr

ence

s

Prominences

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

and but also because however

Per

cen

tag

e