Development in L2 phonology: Let’s take a long look at the age issue Martha Young-Scholten Fundamental issues in L2 acquisition are in part problems about time; claims about development are best interpreted within a full longitudinal perspective where light can also be shed on causes and effects regarding various phenomena (Ortega & Iberri-Shea 2005:26;30). Fundamental issues in L2 phonology ‘Traditional’ issues The role of L1 knowledge The contribution of universals/developmental processes The learner’s age upon initial exposure (and peripheral issues such as aptitude, identity, anxiety) Emerging issues Input The relationship of phonological development to other components of language Longitudinal studies In L2A, the aim is the tracking of L1 influence and universal processes during development Longitudinal studies allow observation of simultaneous development of phonology, morphology, syntax and the lexicon to look for causal relationships Longitudinal data on the development of phonology by child L1ers, child L2ers and adult L2ers provides a developmental perspective on age (and follows current trends in the study of the acquisition of morphosyntax) But few longitudinal studies of the phonological development of child L2 learners exist Even fewer such studies exist on adult L2 learners. The rest of the talk 1. A bit more background 2. The modern history of the study of L2 morphosyntax 3. Development in phonology 4. Development in L2 phonology 5. Conclusion 1. A bit more background From the mid-1980s (e.g. Clahsen & Muysken 1986) through the 1990s (e.g. Vainikka & Young-Scholten 1994) and up to now (e.g. Meisel 2003) L2 morphosyntax researchers have paid (roughly) equal attention to the kind of knowledge acquired (mental representations) how this knowledge is acquired (its emergence, in stage-like development) If the same mechanisms are employed across the lifespan there should be no fundamental differences in learner’s mental representations (e.g. no UG-violating grammars) there should be no differences in the developmental pathways taken (apart from L1 influence) Much of this debate revolves around the production of verbal morphology (non-finite vs. inflected forms) and the position in declarative and embedded clauses of these (main) verbs Longitudinal studies of morphosyntax Typically developing L1 children (including sign language learners) Atypically developing L1 children Simultaneous bilingual children Child L2 learners (= the least number of studies) Adult L2 learners Studies have typically been of only one of these populations; multi-age studies such as Snow & HoefnagelHöhle (1982) and more recently Unsworth (2005) – both coincidentally on L2 Dutch – still represent a small minority Longitudinal studies of phonology Typically developing L1 children Bilingual children (simultaneous) Atypically developing L1 children Child L2 learners Adult L2 learners In 39 years, 17 longitudinal L2 phonology have been carried out, mostly on segmental phonology, and mostly on English (Gut 2009) L2 phonology models could be tested with such data (Best’s Perceptual Assimilation; Flege’s Speech Learning; Eckman’s Markedness/Structural Conformity; Major’s Ontogeny/Phylogeny; Archibald’s parameter resetting, Dziubalska-Kołaczyk’s Natural Phonology; Escudero & Boersma’s OT-based Functional Phonology) IL phonology and age L2 phonology is natural (Eckman 1981) L2 adults’ phonologies are constrained by the same principles as younger learners’ phonologies (Young-Scholten 1996) Many (all?) current L2 phonology models assume access to innate language-specific/phonology-specific mechanisms across the lifespan E.g. Optimality Theory (Prince & Smolenksy 1993) assumes lifelong availability of a set of universal constraints; ranking differs across languages Note: This puts L2 phonology at odds with L2 morphosyntax where debate regarding access to UG/use of same language-specific mechanisms across the lifespan is unrelenting (e.g. Bley-Vroman 2009). IL phonology and age To examine the critical/sensitive period for L2 phonology Look at end-state L2ers with differing ages of initial exposure Include extralinguistic factors, e.g. length of residence Less less attention has been paid to factors such as input quality (e.g. from non-native speakers), input type (orthographic exposure/literacy), identity, aptitude and motivation (this is changing; perhaps presumably peripheral factors are important) If there is a CP for L2 phonology, when does it end? Pre-15 and post-15 year olds constitute two separate populations (=the CP closes around age 15; Patkowski 1990) The CP closes earlier, starting at age 6 (Long 1990) IL phonology and age On the one hand, we have plenty of evidence of use of the same mechanisms by younger and older learners On the other hand, we have plenty of evidence of non-convergence on the target language phonology by older L2 learners This paradox (if it is one) can best be resolved by more work in L2 phonology from a developmental perspective: Without longitudinal data ‘claims about acquisition within an OT model are weak’ (Hancin-Bhatt 2008:142). Errors (as we’ve known for awhile) aren’t the only measure of difficulty ‘rate of acquisition [...] is a more insightful measure of learning’ (Eckman 2008:101) – and best measured in a longitudinal study. 2. The modern history of the study of L2 morphosyntax Studies of L2 learners in a naturalistic acquisition context initially dominated research on L2 morphosyntax (e.g. the ZISA project of adult Italian, Portuguese and Spanish learners of German) and through the 1990s (ESF project on 8 NLs/5 TLs) and currently continue to do so in the study of the L2A of information structure Studying naturalistic learners removes one potentially crucial difference between older and younger L2 learners: instruction L2 morphosyntax Note: The sub-group of L2 syntax researchers currently approaching the UG access question by looking at poverty of the stimulus/PoS effects (where underlying structure is underdetermined by the input and not available from the L1), instructed L2 learners needn’t be excluded because PoS phenomena are not taught, because Forms may not exist for teachers to focus on/correct when inaccurate Lack of awareness by teachers of the latest generative analyses of syntax means that learners aren’t taught the underlying structure Representative longitudinal studies of L2 morphosyntax by naturalistic (TL-immersed) adult learners (numerous such studies exist on L2 children, usually by PhD students) Study L1 and L2 Subjects Duration Cazden et al. 1970s L1 Spanish/L2 English 2 children, 2 teenagers, 2 adults 10 mnths ZISA 1970s/80s L1 Spanish, Portuguese, Italian/L2 German 12 adults 2 years ESF 1980s 6 L1s/5 European L2s 40 adults 2½ years VYSA 1990s L1 English/L2 German Three teenagers 1 year What we know and don’t know What we know Older learners demonstrate UG access, too (White 1989 and work since then on PoS effects up to the present) Learners’ ILs develop in stages where both younger and older learners display similar non-target patterns What we are unsure of The strength of influence of L1 knowledge (Schwartz & Sprouse 1996 vs. Vainikka & Young-Scholten 1996; to appear) The relationship between verbal morphology and syntax The source of non-target production in inflectional morphology Debated: The interpretation of variable production of inflectional morphology During acquisition, there is tight coupling of morphology + syntax: acquisition of a given morpheme represents related syntactic structure Vainikka & Young-Scholten (1994): this holds in L1 and L2 acquisition Haznedar & Schwartz (1997): morphology + syntax are tightly coupled in L1 acquisition but neither in child nor L2 acquisition Missing inflection is due to surface-level mapping problems Prévost & White (2000): morphology + syntax are tightly coupled for L1 and L2 children, but not for L2 adults Goad, White & Steele (2003): variable production of functional morphology is related to L1 transfer of prosodic licensing (past/agreement/plural suffixes in L1 Chinese/L2 English) Oldenkamp (2010): variable production (in L2 Dutch) involves interaction of morphology and prosodic structure What is the role of phonology in the L2 acquisition of morphosyntax? How does the learner go from reception of a continous speech stream to construction of a mental grammar? This has been ‘one of the most under-researched and under-theorized aspects of second language acquisition’ (Carroll 2001:1) This is changing with experimental studies of the phonological realization of grammatical morphemes in L2 acquisition (FrenckMestre, Foucart, Carrasco & Herschensohn 2009; see also Arteaga, Herschensohn & Gess 2003 for pedagogical applications thereof) Hancin-Bhatt (2008:119): substantial empirical evidence for developmental effects in L2 phonology notwithstanding, there are still few theoretical analyses that provide an account of how these effects interact over time with each other how they interact with other components of language Naturalistic and uneducated adults: phonology + morphosyntax + phonological awareness + reading (cross-sectional study, Young-Scholten & Strom 2006) Adults w/no school Som;VN Phung Target-like Morphosyntax Phonological awareness reading segmental and stage % correct level prosodic 1= bare VP; 4= onset; rime phoneme phonology 4=CP decoding 29% 1ii 51% 0% 1 Nien 3% 1i 34% 17% 1 Keif 69% 1ii 61% 8% 1 Abba 56% 1ii 56% 17% 1 Aliya 63% 1ii 37% 0% 1 Shamey 54% 1i 20% 16% 1 Asia 81% 1ii 36% 0% 2 Sharif 71% 4 68% 42% 4 3. Development in L1 phonology A time-line of studies Diary studies Preyer 1889 Large sample studies Longitudinal sampling Late 1920s1950s: groups Stern 1907,1924 of between 72 Leopold 1939-1949 (Fisher 1934) Grégoire 1937,1947 and 430 (Templin Gvozdev 1949 1957) children from 1;6 to 8;0 on their Zarębina 1965 articulation, vocabulary, sentence length Braine 1963: 3 children Miller & Ervin 1964: 5 children Brown 1973: 3 children Bloom 1970 3 children Others (usually one child) e.g. Velten 1943, Waterson 1971, Menn 1973, Smith 1973, Ingram 1974 Shvachkin 1973 study of 19 Russian children’s perception of phonemic distinctions from 0;10 to 1;6 Solely on phonology: Smith (1973)…Fikkert (1994) ... Lleó & Prinz (1997)... Rose (2000) 4. Development in L2 phonology Morphosyntax researchers’ recruitment of phonology to address the intractable problem of variation in production (and more; see below) This points to the need to conduct studies along the lines of Snow & Hoefnagel-Höhle (but see below) Do we know as much about L2 phonological development as we do about morphosyntactic development? Do we know as much about L2 phonological development as we know about L1 phonological development? A time-line of longitudinal L2 phonology studies Study Description Features investigated Dickerson 1974 1 yr: 5 Japanese students’ English /r/ and /l/ Wode 1978 ½ yr: 4 German children’s English Hecht & Mulford 1979 2/3 yr: 1 Icelandic child’s English fricatives Snow&HoefnagelHöhle 1982 1 yr: 33 children’s, teenagers’, adults’ Dutch phonology, morphology, vocabulary, syntax Sato 1984 5/6 yr: 2 Vietnamese boys’ English syllable structure Rankin 1985 5 yrs: 32 Spanish university students’ English segmental perception, production Winitz et al. 1995 6 2/3 yrs: 1 Polish boy’s English . accent Carlisle 1998 4 yrs: 4 Spanish students’ English syllable structure Abrahamsson 2003 1 2/3 yrs: 1 Chinese adult learner’s Swedish syllable structure Edwards 2006 5/6 yr: 2 Vietnamese adults’ English syllable structure phonology; morphology Derwing et al. 2008 2 yrs: 32 Mandarin, Ukrainian, Russian adults’ English fluency; comprehensibility Oldenkamp 2010 phonology; morphosyntax 1 -2 yrs: 24 Turkish, Arabic, Chinese adults’ Dutch Gut (2009): a corpus study that can provide a foundation for longitudinal, comparison data Large-sample synchronic corpus L2 German (n= 55) and L2 English (n=46) focusing on rhythm, stress and intonation German L2 learners’ initial exposure ranged from age three to 33 and included both instruction and immersion Sample includes spontaneous speech, a reading passage, nonsense word reading and interview Qs on exposure, motivation, attitude, musical and acting ability Gut examined relationships between prosody, morphology and extra-linguistic factors Findings The variation found is highly systematic Acquisition involves a set of mutually dependent factors Age is no barrier to acquisition of phonology Phonology and vocabulary In L1 acquisition, a vocabulary spurt coincides with acceleration of phonological development (Ingram 1976). This spurt may relate to variability in syllable structure simplification processes: typically developing L1 children delete rather than epenthesize, but early talkers exhibit considerable epenthesis along with their larger vocabularies In L2 acquisition epenthesis is the preferred syllable simplification strategy for older learners (Abrahamsson 2003; Weinberger 1987 ) Is there a comparable vocabulary spurt in (naturalistic) second language acquisition that coincides with phonological development? Do older L2 learners resemble atypical L1 children? One line of inquiry in L2 research on morphosyntax has involved asking whether L2 acquisition by older learners resembles that of atypically developing L1 children (most recently see Marinis to appear). In L2 phonology, the question is whether developmental processes resemble that of the atypically developing children discussed for example in Ingram (1976) who either exhibited prolonged use of typically developing children’s processes or who followed unique processes. The role of phonology in triggering morphosyntactic development The VYSA longitudinal corpus LEARNER EXPOSURE to foreign languages AGE at arrival in Germany Joan 1 month of Spanish; no German 16 Paul 1 semester of French; no German 17 George 1 year of French; no German 15 Input and data collection in the VYSA study Input For the first four weeks From a German teacher and other American students during language + culture course From learners’ initial host family During 11 subsequent months From (native speaking) students in German secondary schools From host families Data collection 11 times (monthly), with individual learners in Germany Broad and narrow morphosyntax and phonology tasks, including grammaticality judgment, comprehension and translation Informal conversation with the researcher about the learner’s unfolding exchange experience Joan, Paul and George’s acquisition of German syntax, from 3 wks after arrival (file 1) to ca. 12 months later (file 11) Stage: highest syntactic projection Joan Paul George 1i: VP (verb phrase) – no function words produced head-initial : Der Mann trinken der Kaffee. 1 1-2 1 1ii: VP switches to head-final 3 3 3 2-3 3-4 2-3 5 5-6 4 6-7 7-8 6-8 9 11 never 11 never never Die Madchen immer die Buch lesen. 2: FP/TP emerges and is head-initial Ich gegessen der Apfel. 3: AgrP emerges and is head-initial Ich wohnst there. 4i: CP emerges and is head-initial Ich habe gewunst, dass du hast gefragt. 4ii: AgrP switches to head-final Willst du es wirklich wissen, was wir gemacht haben? 4iii: AgrP head-final throughout Individual variation Despite being the youngest, George made the least progress in his acquisition of syntax: we attribute this to his conscious learning of inflectional morphology (cf. Felix 1985 on competing cognitive structures) George also made the least progress in acquiring final devoicing (see Young-Scholten 2004) While Joan was best at morphosyntax, she Joan did not make the most progress in final devoicing (due to more orthographic exposure? (see Young-Scholten 2002) How does this relate to the acquisition of morphosyntax? All (naturalistic) learners (subconsciously) scan the input for cues or triggers that reveal how a given language operates Phonological and morphological triggers for the acquisition of syntactic structure in German Projection Trigger in child German Trigger in adult L2 German VP Stress pattern L1 bootstrapping FP 3rd sg. –t Modals TP Participle affixes (ge/-t) Auxiliary haben ‘have’ AgrP Agreement (2nd sg. –st) Copula sein ‘be’ CP Object clitics (-s/-n) Complementizers George consciously learned these two paradigms. Detecting triggers For us, triggers are among the morphological (verbal) elements which start to become productive in a learner’s speech before s/he posits new functional projections They are robust (but does this equate to salience?) in the input Longitudinal data required for examining such causal effects: morphosyntactic and phonological data The VYSA have been orthographically transcribed They have not yet been phonetically transcribed When the data are phonetically transcribed, they can be donated to PHON PHON: MacWhinney & Rose’s CHILDES-inspired data management tools for contribution of data to PhonBank. PhonBank: overwhelmingly covers L1 phonology What’s the ideal longitudinal L2A study for PhonBank? Ortega & Iberri-Shea (2005) emphasize the need to gain a sufficiently lengthy picture in order to observe development Expected time-scale for emergence of investigated phenomenon/a determines study length/session frequency Although the boy in Winitz et al.’s study was tracked for 62/3 years, native production was attained after 1 year… Emergence varies: re-examine cross-sectional data, including the details of participants’ input quantity and quality Data from longitudinal studies not on phonology Data collected for morphosyntactic analysis have been only orthographically transcribed PHON only requires phonetically transcribed data, so recordings needn’t be of quality required for acoustic analysis If state-of-the-art recording equipment (including analogue) has been used to enable transcription of TL- and non-TL-like inflectional morphology And the usual care has been taken when recording And if storage of recordings has been careful --- and it has been for the VYSA data --- then such data can be phonetically transcribed and analyzed to shed light on phonological development from the initial state onwards and on the role of phonology in triggering morphosyntactic development.