The Effect of Orthography on Global Foreign Accent: Zhuyin vs' Pinyin

1
The Effect of Orthography on Global Foreign
Accent Zhuyin vs. Pinyin

• John Archibald Wei Cai Shu-ning Sciban
• Language Research Centre
• University of Calgary
• (Baltimore 2005 CLTA Conference)

2
The Effect of Orthography on Global Foreign
Accent Zhuyin vs. Pinyin

• 1.0 Introduction/Background
• 2.0 Data Collection
• 3.0 Data Analysis
• 4.0 Executive Summary
• 5.0 Conclusion

3
1.0 Introduction/Background

• The Language Research Centre at the University of
  Calgary was contracted by the School Board to
  investigate a research question of relevance to
  their Mandarin bilingual program.
• Some parents in City 1 were concerned that if
  their children were taught Mandarin using the
  Pinyin writing system that their childrens
  Mandarin would have more of an English accent
  because the children would be confusing the
  English alphabet with the Chinese Pinyin
  alphabet.

4
1.0 Introduction/Background

• 1.1 The Research Question
• 1.2 Zhuyin versus Pinyin
• 1.3 Global Accent Ratings
• 1.4 Experimental Design

5
1.1 The Research Question

• Will the method of instructionZhuyin or
  Pinyinaffect the childrens Mandarin
  pronunciation

6
1.2 Zhuyin versus Pinyin

• Similarity Both Pinyin and Zhuyin are alphabetic
  writing systems.
• Difference They differ only in the nature of the
  symbols which make up the alphabet.

7
1.3 Global Accent Ratings

• Global accent studies ask native speaker judges
  to assess the accent of speaker.
• Usually the accent is rated along some scale
• (No foreign accent -gt Very heavy foreign
  accent).
• Caution Global accent is one small part of
  language proficiency and does not tell us
  anything about syntax or vocabulary or literary
  or any of the myriad of properties that are
  involved in being bilingual.

8
1.4 Experimental Design

• Questionnaires
• Global accent rating
• The statistical procedure known as the ANOVA
• Post hoc statistical tests (i.e. tests run after
  the ANOVA)

9
2.0 Data Collection

• 2.1 Subjects
• 2.2 Tasks
• 2.3 Questionnaires
• 2.4 Data Preparation/Editing
• 2.5 Judging

10
2.1 Subjects

• Students who were taught via the Zhuyin
  methodology and students who were taught via the
  Pinyin methodology
• Students in Grade 1 4 and 6
• Three cities
• Ethics clearance
• Parental Consent Form made available in English
  and Mandarin (both simplified and traditional
  characters)

11
2.1.1 Schools Part 1

• Interview period March 2004 January 2005
• City 1 had five schools (Zhuyin) City 3 had two
  schools (Pinyin) and City 2 had one school
  (Zhuyin).

12
2.1.1 Schools Part 2

• We gathered data from 209 children broken down
  in the following way

13
2.2 Tasks

• Prior to our data collection we met with the
  teachers in the schools and circulated our
  conversational protocols to see whether the
  topics were considered age appropriate.
• A childrens picture book was available for very
  low level children so that they could say the
  names of colours or numbers or animals.
• Children were recorded for approximately five
  minutes each.

14
2.3 Questionnaire

• Before the recording day parents had completed a
  questionnaire.
• The questionnaire was available in both English
  and Mandarin (simplified and traditional
  characters).

15
2.4 Data Preparation/Editing

• All English was removed.
• Only the childs voice was retained (i.e. the
  interviewer was not heard).
• Any content material which could identify the
  child as a non-native speaker was removed .
• Any content which would allow the childs
  identity to be determined was removed.
• Files rang from 4 to 94 seconds in length.

16
2.5 Judging Part 1

• Four judges were native speakers of Mandarin and
  had familiarity with other Chinese languages.
• We wanted to avoid a situation where a judge
  would give someone a heavy-accent rating because
  they sounded like they were speaking with a
  low-prestige accent from some other region of
  China.

17
2.5 Judging Part 2

• We wanted to be confident that their ratings were
  actually rating the English accent of the
  speakers.
• They were first asked to complete the task and
  make judgments based on the degree of English
  accent the subjects were perceived to have. Then
  they listened to the tapes again and were asked
  to make a judgment on the non-native accent of
  the subjects.

18
3.0 Data Analysis

• 3.1 Inter-rater Reliability
• 3.2 ANOVA
• 3.3 Instructional Time
• 3.4 Demographic Patterns

19
3.0 Data Analysis

• The data from the questionnaires were coded to
  represent the information provided.
• The following categories were pulled out as most
  relevant
• School Attended City of Residence Grade
  Chinese Heritage (Yes/No) Parents know Chinese
  language (Yes/No) Born in China (Yes/No)Years
  of Mandarin Instruction Age of Onset of Mandarin
  Instruction Age Sex (M/F) Tutor (Yes/No)
  Saturday School (Yes/No)

20
3.1 Inter-rater Reliability

• We ran a Cronbachs Alpha test to determine the
  inter-rater reliability and this came back at
  .955 which indicates a very high degree of
  consistency between the judges.

21
3.2 ANOVA Part 1

• An Analysis of Variance run on the global accent
  ratings revealed a significant difference between
  the three cities (p.000).
• City 1 (n66) 1.54
• City 2 (n47) 3.36
• City 3 (n91) 5.06

22
3.2 ANOVA Part 2

• A breakdown of results by school raises doubts
  about the validity of combining the scores of the
  two City 3 schools (School 7 and School 8)

23
3.2 ANOVA Part 3

• All four groups (City 1 City 2 School 7 and 8)
  were behaving significantly differently (p.000).
  
• City 1 schools were not significantly different
  from School 7 (p.656) but were significantly
  different from City 2 (p.007) and School 8
  (p.000).
• School 7 was significantly different from School
  8 (p.000) but not from either City 1 (p.656)
  or City 2 (p.267) schools.
• City 2 is significantly different from City 1
  (p .007) and School 8 (p.000) but not from
  School 7 (p.267).
• School 8 is significantly different from the
  other three (p.000).

24
3.2 ANOVA Part 4

• There is clearly variation that cannot be caused
  by the zhuyin versus pinyin instructional
  methodology because
• School 7 (Pinyin) was NOT significantly
  different from either City 1 (Schools 1-5)
  (Zhuyin) or City 2 (School 6) (Zhuyin) but WAS
  significantly different from School 8 (Pinyin).

25
3.3 Instructional Time

• What this shows is that instructional time is a
  better predictor of global accent than writing
  system is. The score of the School 7 school
  (Pinyin) and the School 6 school (Zhuyin) are not
  significantly different. They have similar hours
  of instruction but different writing systems.

26
3.4 Demographic Patterns Part 1

• The schools with the highest numbers of
  non-Chinese students are scoring the lowest on
  global accent.
• The mean score of children whose parents knew
  Mandarin was 1.18 while the mean of children
  whose parents did not know Mandarin was 1.98. The
  difference was significant (p.000).

27
3.4 Demographic Patterns Part 2
28
4.0 Executive Summary Part 1

• Students with a Mandarin-speaking parent had
  significantly better accent ratings than those
  without.
• Students with more hours of instruction performed
  better than students with fewer hours of
  instruction.
• The students in the three different cities were
  significantly different from one another.

29
4.0 Executive Summary Part 2

• There was no significant difference in the
  English accents of students from School 7
  (Pinyin) or City 1 and City 2 schools (Zhuyin).
• There was a significant difference between School
  7 (Pinyin) and School 8 (Pinyin).

30
5.0 Conclusion

• Ultimately we feel that it is NOT the fact that
  students received instruction in either the
  Zhuyin or Pinyin that caused the differences in
  their performance.