Sound perception and production in a foreign language: does orthography matter?

Learning a foreign language is hard. Many who have learnt German as a Foreign Language (GFL) may have felt challenged by the position of the finite verb in main versus subordinate clauses or the many different plural forms German nouns exhibit. But learning the sounds of German, a language with an unusually high number of 15 vowel phonemes, is just as hard as learning its grammatical structures - possibly even harder (Hirschfeld, 2003: 873). This is especially true for speakers of a language with considerably fewer contrastive vowels, such as Polish. The following utterance by a Polish learner of L2 German exemplifies this issue: Kann ich den Schrank in Ratten zahlen? (“Can I pay for this closet in rats?”) The Polish learner shared this anecdote in a linguistics seminar at Humboldt-University, Berlin, in order to illustrate the persistent challenge Polish speakers face with German vowels. By producing the troublesome segment in the word Raten [ʁa:tən] (“instalments”) too short, the Polish learner found herself in an undesired communication situation, talking about rats while in fact wishing to discuss instalments. Cutler (2015) points out that difficult L2 segmental contrasts are not only problematic for minimal pairs, as was the case in the Raten-Ratten example. Another serious problem is that indistinguishable L2 segments can lead to temporary ambiguity among L2 words. This is caused by (spurious) initial overlap of words, which lead to additional processing costs for L2 learners. For example, Dutch learners have difficulties with English /ɛ/ in pencil versus /æ/ in panda. Cutler et al. (2006) found that when these learners heard panda, they were likely to look at the competitor pencil too and consequently were slowed down in their language processing. The investigation of L2 speech learning is therefore not only of theoretical interest, but can potentially help learners with serious problems in the acquisition of a foreign language. This study focuses on the perception and production of German vowels by Polish GFL learners. It draws on different areas of research, most prominently on GFL research and L2 phonetics and phonology. While some “intellectual imperialists” (Pierrehumbert, 1990: 375) have taken the view that phonetics and phonology cannot be integrated, the field of L2 phonetics/phonology does not seem to be concerned with this traditional divide. The reason why the distinction is not as fundamental in L2 research may be that much of the experimental work in the field makes use of phonetic methodologies, such as acoustic analyses, but at the same time takes interest in the sounds that are often also those sounds which are contrastive in the L2 phonological system. The field of L2 sound acquisition can therefore be understood as an area of research where Ohala's call for the integration of phonetics and phonology is put into practice (Ohala, 1990). As far as GFL research is concerned, Grotjahn (1998) points out the need to incorporate findings from L2 phonetics and phonology into the applied field of GFL pronunciation teaching. He criticizes practices that promote various didactic recipes without taking latest empirical findings into account. Most studies in L2 speech acquisition emphasize the perceptual side of the learning process (Bohn and Munro, 2007: 9), while publications in the field of GFL tend to base their findings on production data (e.g. Morciniec, 1990, Slembek, 1995, Müller 2005). It seems therefore worthwhile bringing the findings and hypotheses from both fields together. Most research in the field of experimental phonetics and L2 phonology has been done on English as an L2 (Hayes-Harb, 2012). Even though some findings on English may be transferable to other languages, experimental research which specifically addresses topics in German L2 sound acquisition is still rare (but see Dieling, 1983, Richter, 2008, Darcy and Krüger, 2012, and Darcy et al., 2013). This seems unfortunate since English may not always be the best candidate to investigate current issues in the field. For example, the acquisition of vowel length in L2 English has attracted much attention (e.g. Bohn, 1995, Bogacka, 2004, Cebrian, 2006, Rojczyk, 2011) even though this feature of the vowel system plays only a secondary role in English (Hillenbrand et al., 2000). This is different for German as an L2, especially in the case of lower vowels (Sendlmeier, 1981). For example, in the [ʁatən] (“rats”) versus [ʁa:tən] (“instalments”) minimal pair, it is primarily the difference in vowel length which differentiates the German phonemes /a/ and /a:/. Flege (1999: 1275) has pointed out the need to focus on more than one acoustic dimension when investigating L2 sound acquisition. In the case of German vowels, it is not only the length of vowels which is of interest, but also the quality of vowels (i.e. tense versus lax). Hence, both dimensions will be addressed in this study. In the case of the perception experiments, this implies that stimuli will be manipulated both in their length and in their quality. For the production experiment, this means that both vowel duration and formant values are of interest in the acoustic analysis of the present data. In the reprint of Ellis' (2008) monumental work The Study of Second Language Acquisition only about 20 pages (out of over 1100 pages) are dedicated to the topic of L2 phonetics and phonology. Yet he states that there has been a considerable growth of interest in L2 phonology in recent years (Ellis, 2008: 103) and various publications of anthologies dedicated to the field of L2 speech learning reflect this trend (e.g. Strange, 1995, James and Leather, 1997, Bohn and Munro, 2007, Edwards and Zampini, 2008). One of the most recent and conspicuous factors of interest in the field is the influence of orthography in the acquisition of an L2 phonology. As early as in 2002, Young-Scholten (2002: 264) noted that “we know very little about the influence of written input on the development of a second phonology. Studies which compare L2 children and adults or L2 adult groups neither control for this input variable nor do they treat it as an independent one”. About a decade later, research does begin to take orthography into account. For example, in 2015, Applied Psycholinguistics published a special issue on this topic, but findings are not straightforward. While some researchers find orthography to have a positive influence on L2 sound acquisition (Rafat, 2015), others find no results at all (Showalter and Hayes-Harb, 2015), and still others report on negative influences of orthographic input (Bassetti and Atkinson, 2015). The latter's research looks at the influence of orthography in acquiring L2 English, i.e. in a writing system known for its idiosyncrasy. For example, the marking of vowel length is relatively opaque, and both phonetically short and long vowels may be marked by double vowel graphemes, as in hood [hʊd] and food [fu:d]. In German, vowel length is marked more systematically, for example, by the lengthening h in the word Sahne (“cream”). However, not every long vowel is explicitly marked, as demonstrated in the above mentioned example Rate /ʁa:tə/. This optional but coherent marking of vowel length in German allows for a set-up of an experimental design that would not be possible in English. Most current theories in L2 sound learning would propose that the quoted Polish GFL learner produced Raten as Ratten because she perceived it as such (see the following chapter). This study asks the question whether orthography may also play a role in this process, as the absence of the lengthening h suggests that the vowel might also be short. In order to investigate this question experimentally, two groups of speakers - 20 Polish GFL learners and 20 native speakers of German - took part both in a perception and a production task involving orthographic marking as an experimental variable. A third discrimination experiment was conducted to investigate the perception of German vowels without any possible influence of orthography. The study is organized as follows. First, prominent theories of L2 speech learning are reviewed, and current hypotheses regarding L2 vowel length acquisition are discussed. Studies which relate L2 sound acquisition and L2 lexical representation are also reviewed, since the current study uses real German words to investigate the phenomenon under discussion (Chapter 2). Following the general overview, extant studies addressing the role of orthography in L2 phonological acquisition will be presented. In light of this, the orthographic systems of German and Polish will be described in detail (Chapter 3). The next chapter introduces the German and Polish vowel systems. The section on Polish includes two exploratory studies, which investigate Polish vowel duration before singleton versus geminate consonants, as well as Polish vowel quality and average Polish vowel duration (Chapter 4). The following three chapters report the main experiments of the study: a discrimination experiment with manipulated nonce words (Chapter 5), a production experiment with real German words which are either explicitly marked or unmarked for their vowel length (Chapter 6), and an identification experiment which includes the same words as the production experiment (Chapter 7). Chapter 8 summarizes the results, discusses how the three studies relate to each other, and illustrates their relevance for foreign language classroom. The thesis closes with an overall conclusion.