Connotative meaning and the challenges of international financial reporting/auditing standards convergence: the case of Taiwan’s Statement of Auditing Standards Number 33

Abstract

Translating international standards into local languages is a big obstacle to the global convergence of financial reporting and auditing practices. This study uses Taiwan’s promulgation of Statements on Auditing Standards No. 33 to examine an important component of this challenge: the need for local standard setters to anticipate how domestic constituents would interpret local words used to translate original English terms. Data collected from samples of Taiwanese auditors, public prosecutors, and investors revealed that as compared to the auditors, financial statement users ascribe different connotative meanings to a Chinese term that was originally proposed, and one that was chosen as its replacement after objections from the accounting profession. Further, the words’ connotative meanings significantly mediated the link between the local term and financial statement users’ judgments about auditor legal liability. Significantly, the connotative meaning that financial statement users ascribed to the replacement term was associated with a higher level of ascribed auditor liability, such that in successfully pushing for a wording change, members of Taiwan’s accounting profession may have increased, rather than reduced, their exposure to liability from audited financial statements. These findings have a number of implications for standard setters, the standard setting process, and accounting firms.

Keywords:

• international harmonization of accounting and auditing standards
• translation
• semantic differential
• Taiwan’s auditing standard setting

Acknowledgments

We are grateful to the anonymous referee and Suresh Radhakrishnan (Editor) for their helpful comments and suggestions. The financial support of National Science Council in Taiwan is also gratefully acknowledged.

Notes

1. For example, in the 2010 World Standard-setters Conference, participants from South Korea and Turkey stressed the difficulty in translating financial reporting standards from English into other languages. See http://www.ifrs.org/NR/rdonlyres/CBEAB38D-3B50-46F6-920F-7EABE90E0850/0/ConferenceDocumentation2010.pdf. Accessed January 15, 2011.

2. Prior studies examining the effect of IFRS adoption on financial reporting quality around the world suggest that merely adopting a common set of accounting standards cannot improve quality and that institutional and firm incentive factors are important factors (e.g. Isidro and Raonic 2012; Houque et al. 2012; Daske et al. 2013). An implicit assumption is that the language and therefore translation issues are not a concern in non-English speaking countries. Yet, Jeanjean, Lesage, and Stolowy (2010) indicate that the language barrier affects the publication of English-language annual reports by firms in non-English speaking countries. Another implicit assumption is the adoption of a common set of auditing standards.

3. The semantic differential method was developed by Osgood, Suci, and Tannenbaum (1957) for use in experimental psychology to measure the connotative meaning of a concept. In the semantic differential method, subjects are required to connect a concept with several adjectival pairs. Bipolar adjectives (e.g. good–bad) are usually presented on a seven-point scale, and subjects are asked to mark one of the spaces to indicate their perception of the concept and the adjective. More details on the scale are provided in the methods section.

4. Houghton and Hronsky (1993) tested the meaning of the following accounting concepts: asset, liability, owners’ equity, revenue, expense, profit, loss, going concern, matching, historic cost, consistency, conservatism, periodic, accrual, and realization.

5. The inconsistencies were in regard to the concepts of owners’ equity, matching, going concern, accrual, and realization.

6. By “three readings,” it is meant that an initial version goes through three rounds of discussion and revision. The first draft for a standard is discussed from the first paragraph to the last, and it is then revised as the second version of the draft. The second version is discussed in the same manner and further revised. Finally, the third version is discussed and revised until consensus is reached.

7. The instrument was first developed in English and then translated into Chinese following Brislin’s (1986) translation-back translation method.

8. The semantic differential scale is commonly used for measuring the connotative meaning of subjects or concepts. The respondent is asked to choose where his or her position lays on a scale between two bipolar adjectives pairs that represented the evaluative dimension of meaning. The person’s perception score would be the sum of the numbers corresponding to the positions checked on the scales (Petty and Cacioppo 1981).

9. For each pair of adjectives, the adjective that has a positive meaning associated with a translated word is on the left of the scale and the antonym of that adjective is on the right. Take “guarantee” as an example: “credible” is positively associated with “guarantee” and is placed on the left. The antonym of “credible” – non-credible – is placed on the right.

10. While our illustration in the instrument covers only three possible responses (“1”, “4”, and “7”) for brevity, our participants were not led to confine their responses to these three answers. Among the 125 responses to the five semantic differential scales by the 25 audit partners in the “guarantee” group, the frequencies of responses “1” , “4”, and “7” were 1, 10, and 34, respectively, whereas the frequencies of responses “2”, “3”, “5”, and “6” were 1, 0, 32, and 47. The responses by audit partners in the “confirmation” group and the prosecutors as well as investors reveal a similar pattern. In total across the 690 responses to the semantic differential scales by the participants, the frequencies of responses along the 1–7 scale were 4, 4, 17, 99, 148, 237, and 181, respectively.

11. Throughout the paper, all tests are two-tailed unless otherwise mentioned.

12. Since the debate on SAS No. 33 occurred about a decade before our experiment, there may be a concern that the connotative meanings of words have changed, so that our results do not map into the substance of the discussion back then. We believe that the chances of such a mismatch are remote. While the meanings of words may change over time, such changes tend to be very gradual. In their examination of the structure and meaning in Chinese, Liu et al. (2010) indicated that Chinese idioms have highly stable structure and meaning integrity. Furthermore, the two words in question are used in much wider circles than SAS alone, and it is extremely unlikely that their ascribed meanings would have been altered just because they had been invoked in SAS No. 33. A further consideration is that in our study, the words’ connotative meanings were solicited in a context-free manner prior to introducing the issue of auditor legal liability. This makes it unlikely that participants’ responses were influenced by the auditing standard in question.

13. Splitting the sample by the average score of connotative meaning yielded qualitatively equivalent results, as did an ANCOVA using the raw score of connotative meaning as the covariate (p < 0.001).

14. In the regression analysis for the translation-connotative meaning path, the independent variables are translation (represented by one dummy variable) and profession (represented by two dummy variables), and the dependent variable is connotative meaning (using raw scores). In the regression analysis for the connotative meaning-judgment of auditor legal liability path, the independent variables are connotative meaning and profession, and the dependent variable is the judgment of auditor legal liability. The Sobel test statistic =  where a = regression coefficient for the translation-connotative meaning path; b = regression coefficient for the connotative meaning-judgment of auditor legal liability path; s_a=standard error of a; s_b=standard error of b.

15. We are indebted to an anonymous reviewer for this suggestion.