Phonemic and Phonotactic Inference in Early Interlanguage: Americans Learning German Fricatives in L2 Acquisition

Abstract

Second language (L2) phonological acquisition involves learning novel target-language sounds, variable forms of sounds that arise in different phonological contexts, and any phonotactic constraints that govern their appearance. Interlanguage (IL) grammars must adapt to represent sounds and constraints that are novel to the native language (L1) grammar. Research conducted in the perceptual assimilation (PA) paradigm addresses naïve pre-learners' and L2 listeners' perception of foreign sounds but typically controls for phonological context rather than manipulating it. This design priority limits phonological insights to position-dependent perceptual knowledge, not addressing high-order phonological knowledge across multiple contexts. Conversely, most investigations of acquisition of L2 phonological distributions neglect to investigate category acquisition. The German so-called dorsal fricatives, voiceless palatal [ç] and voiceless velar [x] (with uvular variant [χ]), orthographically represented by the letters , provide a valuable test case to examine development of novel categories and novel distribution patterns simultaneously. Both novel to L1 American English speakers, the dorsal fricatives exist as allophones (i.e, positional variants) of the same phoneme, their distribution determined largely according to the rule of Dorsal Fricative Assimilation (DFA), by which the dorsal fricative must agree in frontness/backness with the segment that precedes it. Furthermore, both languages' phoneme inventories include acoustically similar /h/ but allow it only in simple onsets of syllables, whereas dorsal fricatives most commonly appear in syllable codas.

To investigate acquisition of the dorsal fricatives in early L2 exposure, this dissertation undertakes four auditory perception experiments. Chapters 3 and 4 follow the typical PA paradigm, in which participants listen to a particular sound in a static phonological frame, select which category of their L1 (American English) the sound fits best, and then rate its goodness of fit. These two experiments investigate naïve pre-learners and early L2 learners' mappings of the German sounds [[esh] ç x k h pf], presented auditorily beside either [a] or [ε] vowels, to L1 English categories denoted orthographically. Innovatively, these experiments investigate three syllable contexts (syllable-initial in a simple onset, penultimate as first consonant in a coda cluster, and syllable-final in a simple coda). These experiments reveal different mappings and perceptual overlaps tat vary according to sound, participant group, and syllable context.

Investigation of how learners acquire novel distributions and phonotactic constraints in L2 is undertaken primarily by means of two phoneme detection (PD) experiments, in which participants listen for a specific sound in each trial and respond as fast as possible by keyboard when they perceive it, measuring reaction time (RT) as an indication of phonological processing. These experiments utilize only the penultimate (C)CV_[t] context, which is a common phonological context for the [ç x [esh] k] in German. To avoid the problems inherent with labeling a phonologically unfamiliar listening target for L2 learners, Chapter 5 uses German native speakers (NS) to calibrate a novel version of the PD task that robustly replicates previous findings for PD with regressive nasal assimilation; however, the experiment does not replicate previous results with German DFA, which are mixed and conflicting in the literature. Chapter 6 expands the investigation of DFA and expands to include investigation of the ban again [h] in syllable codas that is common to both German and English. This experiment investigates RT shifts of NSs and L2 learners of German (L1 American English) in response to licit coda sequences (axt]_σ, εçt]_σ) and illicit sequences (*açt]_σ, *εxt]_σ, *aht]_σ, *εht]_σ). Surprisingly, early L2 learners exhibit greater (albeit variable) sensitivity to violations of DFA and the ban on *Coda-[h] than NSs, who show reliable RT effects only for the *εht]_σ context. Thus for German NSs, neither PD experiment detects previously reported RT effects associated with violation of DFA in auditory stimuli. In contrast, early L2 learners exhibit a small facilitation effect (faster RT) in response to DFA violations. A second finding is that L2 learners exhibit robust inhibition (slower RT) in response to violation of the phonotactic ban barring [h] in syllable codas, but German NSs mitigate this effect specifically following [a], suggesting that Germans exhibit phonotactic assimilation of [h] in the same context where the acoustically similar axt]_σ sequence is permitted.

Taken together, these results indicate that IL phonological development is context-sensitive and encompasses both novel sound categories and novel phonotactic knowledge. Implications for a unified model of L2 phonological acquisition are discussed, including the representation of early stage phonologization, in the sense of novel phoneme assembly across contexts. To inform the advancement of L2 phonological theory, future research should investigate remaining empirical and theoretical gaps in the research on (German) mid and back continuants in the domains of (socio-)phonetics, dialectology, formal phonology, and the effects of orthographic literacy in L2.