<<

Proceedings from FONETIK 2019 , June 10–12, 2019

A Lingua Franca view on Swedish Pronunciation – A review and Pedagogical Implications Bosse Thorén Department of Social and Behavioral Studies, University West, Sweden. [email protected]

Abstract mild foreign accent, as long as intelligi- Learners of Swedish as a second lan- ble, does not seem to hamper communi- guage need a meaningful pronunciation cation and integration into a new com- training. To achieve this, teachers need munity. However, an unclear pronuncia- guidelines how to plan and perform tion may do, possibly because it could meaningful teaching. This paper argues impede communication considerably. for a priority ranking among phonologi- For second speakers using cal and phonetic features for pronuncia- English as a lingua franca, Jenkins tion teaching. The arguments are in- (2000, 2002) has suggested “Lingua spired by the concept of Lingua Franca Franca Phonetic Core”, i.e. phonetic and Phonetic Core for English and is based phonological features that are thought to on studies and experience concerning in- be most crucial for intelligibility. In the telligibility. It highlights Swedish pro- same vein, Brown (1991) and Catford sodic and segmental features on phono- (1987) promote the idea of “Functional logical, acoustical and strategic levels Load” for phoneme contrasts, which is and ventures to outline an agenda for calculated based on the number of possi- Swedish pronunciation teaching, incor- ble minimal pairs as well as how fre- porating the fact that Swedish is used in quent the sounds occur in speech. The all constellations of L1 and L2 speakers central point in the Lingua Franca Pho- from different language backgrounds, netic Core and Functional Load is that plus the fact that adult L2 learners are realizing certain features properly is suf- unlikely to achieve a nativelike L2 pro- ficient for intelligible pronunciation. nunciation. The suggested Swedish pho- This point could be applied to Swe- netic core gives priority to temporal pro- dish as a second language and some re- sodic features over tonal, and conso- search has been carried out in this field, nants over . by e.g. Bannert (1984) and Abelin & Thorén (2015, 2017). This paper dis- Introduction cusses what promotes intelligibility in Language proficiency includes skills in Swedish L2 pronunciation. The reader is listening, speaking, , writing and referred to Table 1 for an overview of . Competence in pronuncia- Swedish . tion can involve both intelligible speech Swedish as a Lingua Franca and listening with understanding. A for- Swedish does not have the role of a eign accent can affect how other compe- global Lingua Franca like English, but tencies as well as the person’s credibility since inhabitants in Sweden nowadays are judged (Boyd & Bredänge 2013; have more than 150 different mother Lev-Ari & Keysar 2010). According to tongues (Parkvall 2016), Swedish actu- Munro & Derwing (1995), perceived de- ally serves as the means of communica- gree of foreign accent does not correlate tion between L2 and L1 speakers as well strongly with degree of intelligibility. A as between L2 speakers of different lan- guage backgrounds. The Lingua Franca

43 Proceedings from FONETIK 2019 Stockholm, June 10–12, 2019 situation for Swedish in Sweden calls for found by Derwing & Munro (2005) and educational goals that promote commu- Murphy (2014). Teachers lack training nication efficiency, rather than native- for teaching pronunciation and therefore likeness. Moreover, research by Piske, avoid doing it. MacKay & Flege (2001) and Abra- The situation in Sweden can be de- hamsson & Hyltenstam (2009) shows scribed as slowly realizing how L2 pro- that a nativelike pronunciation in an L2 nunciation instruction can be more pri- can be a realistic goal for young learners oritized, more realistic and more focused but a utopia for most adult learners. on intelligibility. Here is not the place to Before the present multilingual situ- grade teaching materials with respect to ation had arisen and before we knew how “good” they are in this respect, but what we know today about realistic ex- it should be pointed out that they often pectations for learners of Swedish as a lack clear guidelines concerning im- second language, at least implicit curric- portant and less important phonetic fea- ula used to have as a main goal that L2 tures. Some features can be presented as speakers should acquire a native-like important because they are “hard to Swedish pronunciation, irrespective of learn”, but not necessarily because they age or L1. If they could not, it was seen are important for intelligibility. For ex- as a failure for teachers and for learners. ample, the Swedish word accents (acute A Lingua Franca view, on the other and grave) and the [ɧ] sound are hard to hand, focuses on mutual intelligibility learn but not important for intelligibility. and does not see the foreign accent per The Swedish phonologic features’ re- se as a problem. spective influence on intelligibility Today’s situation with respect to pro- Earlier, Bannert (1980) suggested that nunciation teaching for learners of some contrasts and other phonological Swedish as an L2. properties in Swedish should be given My impression gained from being more higher priority than others, based on as- than 40 years in the field is that pronun- sumed importance for intelligibility. Alt- ciation teaching gets lower priority than hough this was suggested on an intuitive grammar and vocabulary training. More basis, it was anyhow an important step than 10 years of teacher training in three into promoting intelligibility rather than Swedish universities has given the im- nativelikeness as an acceptable and de- pression that both in-service and pre-ser- sirable learning outcome. vice teacher students know more about vs segments basic grammar concepts than they do about basic and phonology. In various teaching materials, e.g. Kjel- Zetterholm (2018) found that among lin (1978), prosodic features in general 92 teachers of Swedish as a second lan- are assumed to be more important for in- guage, prioritized learning goals were telligibility than segmental features. “communication, reading, writing, Prosody is compared to syntax as a grammar and vocabulary” (p. 81), and macro level and segmental properties are pronunciation was seldom taught explic- compared to and defined as itly. The teachers in the study generally micro level. The macro level is assumed thought that a listener-friendly and intel- to be generally more important than the ligible pronunciation is important, and micro level, for intelligible speech. It the reason given for not teaching pro- may be true, but to my knowledge, there nunciation explicitly was mainly lacking are no studies supporting that view. knowledge of phonetics and competence There are a couple of studies (Abelin & in pronunciation teaching. Her result Thorén 2015, 2017) that show signifi- agrees with the situation for English, as cantly higher perceptual weight for the

44 Proceedings from FONETIK 2019 Stockholm, June 10–12, 2019

Figure 1. Overview of Swedish phonology prosodic contrasts of and quantity, Vowels vs consonants compared to the tonal word accent con- By looking at the numbers of and trast. Furthermore, the latter has at least consonant phonemes; 9 and 18 respec- five different tonal patterns in different tively, plus the possibility for conso- regions (Gårding 1977), plus non-exist- nants to appear in clusters, we can as- ent in a couple of regional varieties. This sume that consonants carry more cues to means that we know something about in- meaning than vowels by virtue of their telligibility within the prosodic field but number. Some anecdotal evidence for not between prosody and segments.Fur- this assumption is that you can replace thermore, a comparison between pros- vowel letters in a written text by hyphens ody and segments with respect to per- or asterisks and still read it with a fairly ceptual importance would be compli- good understanding, while doing the cated since the three prosodic phonemic same thing with consonant letters causes contrasts can be naturally dichotomized, more struggling and guessing. Further- allowing experimental distortions in a more, there are “secret ” often consistent way to test intelligibility: Tro- used by children, that distort the lan- chaic stress pattern as opposed to iam- guage in different but regular ways. One bic, /VːC/ as opposed to /VCː/ and ac- of those is “I-sprikit” (the I-language), cent 1 as opposed to accent 2 (Abelin & where all vowels are replaced by /i/ and Thorén 2017). The phonemes on the still rendering intelligible speech to at other hand cannot be divided that way. least a native Swedish listener who has More correctly, every vowel can be con- had some training. The latter example is trasted to every other vowel phoneme assumed to be more relevant to spoken and every consonant phoneme can, with language than the former, but the former a few exceptions, be contrasted to every illustrates the fact that there are over all other consonant phoneme. more consonant than vowel sounds in Swedish speech.

45 Proceedings from FONETIK 2019 Stockholm, June 10–12, 2019

Phonotactics closed in others. In the province of Väs- Swedish phonotactics allow word initial tergötland, where I currently live, I hear consonant clusters of three consonants lots of [ʝøːrɐ] and [ɕøːrɐ]. and up to five consonants in word final In a similar way, much effort has position. Allowing that heavy consonant been put into teaching and learning the clusters is unusual in a universal per- “very Swedish and exotic” [ɧ] allophone spective. In medial position, in com- as the only acceptable variant of a /ɧ/ʃ/ʂ/ pounds, there can be at least up to six phoneme. One allophone of the same consonants in a sequence, e.g. phoneme is the retroflex [ʂ], which is textstruktur [ˈtɛkːststrɵkˌtʉːr]. similar to English ‘sh-sound’ and Ger- man ‘sch-sound’ and has counterparts in Phonemes, allophones and a host of languages. This means that [ʂ] Although all phonemic contrasts are not should be an easier and fully acceptable equally important for intelligibility, we target allophone. Furthermore, the may agree that in general, phonematic learner who aims for the special Swedish variation, e.g. in L2 speech, is more det- [ɧ] allophone, to sound “standard Swe- rimental to intelligibility than allophonic dish”, anyhow has to use the [ʂ] allo- variation. Nevertheless, in educational phone in -final positions in contexts, allophonic rules are often pre- words like garage [̟ɡɐˈrɑːʂ] and dusch sented as equally important as phonemes [dɵʂː] ‘shower’. A third allophone of /ɧ/ and phonemic variation. A typical exam- is a velar [x] or a uvular [χ], which are ple is that the long allophone of /a/ often associated with or Persian should have the back quality of [ɑ] or the accents, but they are also found in native back plus rounded [ɒ]. Back [ɑ]/ [ɒ] al- Swedish speech in Västergötland, e.g. lophone is often equivalently classified the local pronunciation of Skövde as as /o/ by L2 learners whose L1 has fewer [ˈxœvdɛ] or even [ˈxʌvdɛ]. It does not vowel phonemes, which should result in sound beautiful in most native Swedish higher priority to discern /a/ and /o/ than ears, but it does not cause ambiguity. to achieve the “dark quality” of the long Phonology and acoustic correlates in a /a/. Articulating long and short vowels didactic perspective with different spectral quality has been suggested, e.g. by Bannert (1980), as a The quantity contrast seems to rely on highly ranked feature although many in- two acoustic perceptual cues: vowel du- Sweden regional varieties realize these ration related to subsequent consonant spectral differences in many different duration as well as to the duration of big- ways and sometimes not. Behne, Czigler ger units (Traunmüller & Bigestans & Sullivan (1997) and Thorén (2003) 1988) and to some degree on vowel found that he spectral differences in long spectrum (Hadding-Koch & Abramson and short vowel allophones contribute 1964; Thorén 2003). less to quantity category perception than In an educational context, where relative duration. In a similar way, artic- neither teachers nor learners can be ex- ulating /ɛ/ and /ø/ as the more open /æ/ pected to be trained phoneticians, and and /Œ/ respectively, when followed by learners are unlikely to achieve native- /r/, has been presented as a rule equal to like pronunciation of the target lan- discerning different phonemes, although guage, it is necessary to aim for a limited the mentioned allophonic variation is ab- number of simplified but efficient de- sent in some regional varieties, meaning scriptions and rules, which put intelligi- that the /ø/ and /ɛ/ phonemes are consist- bility as the central aim. In the case of ently open in all phonetic contexts in the Swedish quantity contrast, it means some regional varieties and consistently that the complementary durational reali- zation /VːC/ - /VCː/ is recommended as

46 Proceedings from FONETIK 2019 Stockholm, June 10–12, 2019 the overall safe way, while spectral dif- and quantity are more important for in- ferences between long and short vowel telligibility than the tonal word accent allophone can be optional. contrast. We also have some structural Another area where it is fruitful to and anecdotal indications that conso- look at the phonological and the acoustic nants are more important for intelligibil- levels together, is the relationship be- ity than vowels. tween word stress and the quantity dis- tinctions. Many languages use word Implications and suggestions for stress either to discern meaning (Eng- future language instruction lish, Swedish, Spanish) or to signal word We hope for a situation where teacher boundaries (Polish, Finnish, Persian), trainers and teachers share the ambition but most of them do not have mandatory to prioritize among phonetic and phono- signaling of quantity category in stressed logical features to optimize pronuncia- , and hence no obligation to tion teaching to promote intelligibility. lengthen stressed syllables as much as is Otherwise every single (known) detail in required in Swedish. When looking at the pronunciation becomes equally im- the Swedish stress and quantity together, portant to learn and the task becomes we learn from Fant & Kruckenberg overwhelming for most teachers and (1994) and from Traunmüller & Biges- learners. tans (1988), that duration is the main Like Bannert (1980) and Jenkins acoustic perceptual cue to both stress (2000, 2002), I would like to suggest a and quantity. This is well expressed by set of Lingua Franca Phonetic Core Fea- Engstrand (2004): “Stressed syllables tures for Swedish. They should include are long syllables; but in Swedish, sylla- 9 vowel phonemes and 18 consonant bles can be long in two different ways. phonemes. In prosody, priority should The difference between these two ways be given to stress and quantity over the form a distinctive contrast: The quantity tonal word accent and temporal realiza- contrast” (p. 183 My translation, italics tion of quantity should have priority by original author). Kjellin (1978) cap- over spectral. Consonants in general tured the same relationship in a self-il- should be given priority over vowels in lustrating slogan: “All-la starr-ka staaa- general and consonant clusters should be velser måss-te va långng-nga”, ‘All highly ranked. strong syllables must be long’ (p. 30), Finally, we can hope for more re- showing the mandatory lengthening of search contributing to deepening the stressed syllables as well as the comple- knowledge of what phonological and mentary vowel-consonant length. If we phonetic properties are more or less cru- look at the stress and the quantity con- cial to make Swedish intelligible for all trasts separately, we may not see this users of Swedish in all parts of the Swe- possibility to promote both contrasts dish speaking community. with one acoustic means, which is dura- tion. This finesse has not been widely References acknowledged or understood, and many Abelin, Å., & Thorén, B. (2015). The teachers and academics seem to think it relative perceptual weight of two is safer to stick to long and short vowel. Swedish prosodic contrasts. Proceedings of the International To conclude, we still don’t know if Symposium on Monolingual and prosody is more crucial for intelligible Bilingual Speech 2015 speech than segmental features, but we Abelin, Å., & Thorén, B. (2017). The have findings indicating that the tem- perceptual weight of word stress, poral phonemic contrasts of word stress quantity and tonal word accent in Swedish. Phonology in Protolanguage

47 Proceedings from FONETIK 2019 Stockholm, June 10–12, 2019

and Interlanguage. Equinox Jenkins, J. (2000). The phonology of English Publishing, Sheffield, UK. as an international language. Oxford, Abrahamsson, N., & Hyltenstam, K. (2009). UK: Oxford University Press. Age of Onset and Nativelikeness in a Jenkins, J. (2002). A sociolinguistically Second Language: Listener Perception based, empirically researched Versus Linguistic Scrutiny. Language pronunciation syllabus for english as an Learning 59, 249-306. international language. Applied Bannert, R. (1980). Svårigheter med svenskt Linguistics, 23 (1), 83-103. uttal: Inventering och prioritering. doi:10.1093/applin/23.1.83 Praktisk Lingvistik nr 5. Institutionen Kjellin, O. (1978). Svensk prosodi i för lingvistik. Lunds universitet. praktiken. Uppsala: Hallgren & Bannert, R. (1984). Prosody and Fallgren studieförlag. intelligibility of Swedish spoken with a Lev-Ari, S., & Keysar, B. (2010). Why don't foreign accent. Nordic Prosody III (pp. we believe non-native speakers? The 7-18). Umeå, Sweden. influence of accent on credibility. Behne, D., Czigler, P., & Sullivan, K. Journal of Experimental Social (1997. Swedish Quantity and Quality: Psychology, 46. 1093–1096 A Traditional Issue Revisited. In Munro, M. J., & Derwing, T. (1995). Phonum 4, Department of Linguistics, Foreign Accent, Comprehensibility, Umeå University. and Intelligibility in the Speech of Boyd, S., & Bredänge, G. (2013). Attityder Second Language Learners. Language till brytning - exemplet utländska lärare Learning 45: 1 i svenska skolor. In K. Hyltenstam and Murphy, J. (2014). Myth 7: Teacher training I. Lindberg, (eds.) Svenska som programs provide adequate preparation andraspråk : i forskning, undervisning in how to teach pronunciation. In L. och samhälle. Lund: Studentlitteratur. Grant (ed.) Pronunciation Myths: 437-459. Applying to Brown, A. (1991). Functional load and the Classroom Teaching. The University of teaching of pronunciation. In A. Brown Michigan Press. 188-224. (Ed.), Teaching English pronunciation: Parkvall, M. (2016). Sveriges språk i siffror: A book of , 221–224. London: vilka språk talas och av hur många? Routledge. Stockholm: Morfem. Catford, J. C. (1987). Phonetics and the Piske, T., MacKay, I. R. A., & Flege, J. E. teaching of pronunciation: A systemic (2001). Factors affecting degree of description of . In J. foreign accent in an L2: A review. Morley (Ed.), Current perspectives on , 29 (2), 191-215. pronunciation: Practices anchored in doi:https://doi.org/10.1006/jpho.2001.0 theory (pp. 87–100). Washington, DC: 1340 TESOL. Thorén, B. (2003). Can V/C-ratio alone be Derwing, T., & Munro, M. J. (2005). sufficient for discrimination of Second Language Accent and V:C/VC: in Swedish? A perception test Pronunciation Teaching: A Research- with manipulated durations. Based Approach. TESOL Quarterly, Proceedings of FONETIK 2003. 49-52. 39:3. 379-397) Umeå University. Engstrand, O. (2004). Fonetikens grunder. Traunmüller, H., & Bigestans, A. (1988). Lund: Studentlitteratur Perception of the quantity distinction in Fant, G., & Kruckenberg, A. (1994). Notes Swedish /VC/-sequences. Papers from on stress and word accent in Swedish the Second Swedish Phonetics STLQPSR 35, 125-144. Conference held at Lund, May 5-6 Gårding, E. (1977). The Scandinavian Word 1988. 124-127 Accents. Lund: Gleerup. Zetterholm, E. (2018). Teachers’ opinion on Hadding-Koch, K., & Abramson, A. (1964). the teaching of Swedish pronunciation. Duration versus spectrum in Swedish Proceedings of FONETIK 2018. 81-82. vowels: Some perceptual experiments. University of . In Studia Linguistica 18. 94-107.

48