Class: How important is pronunciation?

Semester : WS 2004/2005

Type: Englisch S II

Hauptstudium

Linguistik

A Closer Look at
Similarity and Dissimilarity

Analysis of the Speech Learning Model and the
Similarity Differential Rate Hypothesis

Carolin Kohlmeier

Contents

1 Introduction 1
2 Problem areas of native German learners of English 2
2.1 Terminology--`area of difficulty′ vs. `error′ 2
2.2 The collection of data 3
2.2.1 Consonantal sounds 3
2.2.2 Vowel sounds 5
2.2.3 Assessment and selection of data 7
3 Theoretical approaches in second language phonology 8
3.1 The Speech Learning Model and equivalence classification 9
3.1.1 Theory 9
3.1.2 Application 12
3.2 The Similarity Differential Rate Hypothesis and markedness 14
3.2.1 Theory 14
3.2.2 Application 16
4 Critical evaluation of the SLM and the SDRH 17
4.1 The quality of hypotheses 18
4.3 Similarity and dissimilarity reconsidered 19
5 Conclusion 20
6 References 22

Figures

Figure 1: a spectrogram analysis of a male native German learner′s realization of the English word `gathered′ 4

Figure 2: a spectrogram analysis of a male native German learner′s realization of the English word `healthy′ 5

Figure 3: a spectrogram analysis of a male native German learner′s realization of the English word `teapot′ 6

Figure 4: a spectrogram analysis of a male native German learner′s realization of the English word `snapper′ 7

Tables
Table 1: Postulates of Flege′s Speech Learning Model 10
Table 2: Hypotheses of Flege′s Speech Learning Model 11

1 Introduction

Research on language learning has been and still is dominated by studies focusing on grammatical and lexical issues, whereas pronunciation "by common concession ... is seldom discussed" (Eckman, Elreyes&Iverson 2003:1). This is not to say that Second Lan-guage Phonology had not come up with fertile approaches to account for learners′ difficulties to acquire certain pronunciation patterns or that little would be known about the underlying phonological processes of language learning--on the contrary, researchers have proposed fascinating models that illuminate the process of language acquisition and that identify factors governing the pronunciation of learners.

There are basically three branches that can be distinguished among the approaches offered to explain the characteristics of learners′ pronunciation: the influence that languages acquired previously have on the learner language, the influence of universal principles and the concept of interlanguage, i.e. learner language as a language in its own right having its own pronunciation pattern. Theories based on the influences of a learner′s native language and those based on the influences of universal principles are not to be seen as rival views, but the two kinds of constraints can be considered complementary. The concept of `interlanguage′ is closely linked to both fields of research so that no sharp boundaries are to be drawn between the three directions of research.

This paper will mainly focus on the influence of a learner′s native language on his or her performance in the target language, an aspect which is also, somewhat misleadingly, referred to as `language transfer′. The theoretical approaches to be examined will be Flege′s (1995) Speech Learning Model (SLM) and the Similarity Differential Rate Hypothesis (SDRH) proposed by Major & Kim (1996). As the perspective of the SDRH also includes markedness as a governing factor, one aspect from the field of universal principles will be discussed in this paper as well. Insofar, the SDRH is a good example of the interrelation of language transfer and universal principles.

In order to test the accuracy of these models they will be applied to pronunciation difficulties of native German learners of English on the segmental level to see whether they can account for the problems identified. The evaluation of this investigation will then be supplemented by a general examination of the theories′ hypotheses, which will be based on methodological remarks from Eckman (2004) on the quality of hypotheses and explanatory models as a whole.

One crucial point in the context of these two approaches is the discussion of the concepts of `similarity′ and `dissimilarity′, since they are essential to both the SLM and

1

the SDRH but still lack precise definition. The final aim of this paper will therefore be to attempt a definition of the two terms which is adequate for the role they play in the models of Flege and Major & Kim.

2 Problem areas of native German learners of English

In order to examine the SLM and the SDRH, a collection of areas of difficulty with respect to native German learners of English will be compiled, which the two explanatory models will then be applied to. The identification of problem areas will be based on both theoretical as well as empirical findings taking as a starting point a comparative analysis by Barry & Gutknecht (1971). But before their results and those of other researchers will be compared and analyzed, a brief discussion of the general view on learners′ deviations from standard pronunciation in second language phonology seems appropriate.

2.1 Terminology--`area of difficulty′ vs. `error′

When dealing with differences between learners′ and native speakers′ realizations of words in the target language (TL), different authors have different ways of referring to these phenomena. The influences on a learner′s pronunciation mentioned above are often discussed as sources of `pronunciation mistakes′ or `errors′, but these terms will be used rarely and only attentively in this paper; instead, the phenomena resulting from these influ-ences will be referred to as, for example, deviations or inappropriate realizations. The reason for this apparent hair-splitting is not only the awareness of learner language as a language in its own right, but what should also be kept in mind is that there is not one objectively correct or incorrect manner of pronunciation.

In fact there is a multitude of models which a speaker′s performance can be compared with to assess its appropriateness: if the understanding of `correctness′ is based on the authenticity of a speaker′s pronunciation, in a language like English, which has a great number of varieties, the appropriateness of an utterance can either be judged with respect to one certain accent or it can be measured against average native speakers′ performances measured in terms of, for example, vowel frequency or voice onset time. If, on the other hand, intelligibility is taken as the aim of the learning process instead of authenticity, things will look completely different and many phonetic features which are crucial to authenticity will be only of minor importance.

2

This discussion cannot be fully carried out in the context of this paper, but as a basis for the following analysis one could suggest the standpoint that it is finally the learner who chooses the aim of his or her language learning process--and this aim may vary from mere intelligibility to an authentic Northern Irish accent. The major concern in second language phonology can be seen as the search for the factors that keep learners from reaching their individual aim, whatever it might be.

2.2 The collection of data

In their study from 1971, Barry and Gutknecht compare the sound systems of English and German and derive areas of difficulty of German learners of English. The investigation focuses on language learning in a class-room environment, either at school or at university, where the instructor sets the pronunciation standard that the students are expected to conform to. That is why Barry and Gutknecht can limit their perspective to one specific accent and base their study on Received Pronunciation as the target pronunciation exclusively (1971:1).

Basically, the study is a contrastive analysis of the two sound systems, but Barry and Gutknecht also give anecdotal evidence for the severity of difficulties and thus go slightly beyond a mere comparison. As the aim of this section is to gather a collection of problem areas of German learners which are not only predicted but given evidence for, additional material is required. In the following sections, selected findings from Barry & Gutknecht′s study will therefore be supplemented with results from empirical research in order to obtain more reliable data which the explanatory models can then be applied to.

2.2.1 Consonantal sounds

The first problem identified is the production of voiced consonants in final position, which does not occur in German (Barry & Gutknecht 1971:35). The final devoicing of Standard German leads to the effect that there is no difference in the production of, for example, final / p / and / b /, / t / and / d / or / s / and / z /, and Barry and Gutknecht remark that "persistent transfer of this speech habit to English proves to be an extremely stubborn pronunciation problem" (1971:35). This diagnosis is also supported by the findings of [...]

3