The acquisition of probabilistic patterns in Spanish phonology by adult second language learners: the case of diphthongization

Open Access
Carlson, Matthew Thomas
Graduate Program:
Doctor of Philosophy
Document Type:
Date of Defense:
April 30, 2007
Committee Members:
  • Dr Henry Gerfen, Committee Chair
  • Paola Eulalia Dussias, Committee Member
  • Judith Fran Kroll, Committee Member
  • Daniel J Weiss, Committee Member
  • linguistics
  • psycholinguistics
  • second language acquisition
  • Spanish
  • phonology
  • diphthongization
  • emergent grammar
  • usage-based grammar
  • working memory
The increasingly global nature of our society bears witness more and more to the importance of learning additional languages. Human language is, however, an extremely complex phenomenon and decades of research have raised at least as many questions as they have answered. In particular, the learning of additional languages has fueled a wealth of research, especially regarding the cognitive underpinnings of adult second language acquisition. This dissertation seeks to shed light on the question of adult language acquisition by exploring learners’ knowledge of fine-grained statistical patterns in Spanish morphophonology. Crucially, it adds to this an exploration of the role of certain kinds of memory and attentional control in modulating adults’ sensitivity to these subtle patterns in their second language grammar. Two areas of recent research are especially relevant to the approach taken here. First, recent findings point to the ability of both adults and children to segment words and extrapolate grammatical rules from statistical patterns in (generally artificial) language input. Second, a large body of psycholinguistic literature shows that adults are sensitive to frequency at a variety of levels of structure, including the positional probability of segments and syllables in words. These findings suggest that at least some grammatical information is resonant in individuals’ experience of language, and that both adults and children have access to statistical learning mechanisms that allow them to capitalize on frequency information in their first language. Findings such as these have led to the development of emergentist and usage-based theories of language that address phenomena from child language acquisition to historical linguistics, and which develop a view of grammar as instantiated in statistical patterns in actual language, dynamically evolving in response to subtle changes in the frequency of particular structures. Importantly, such a view of language structure, learning, and representation suggests both a continuity in the mechanisms of language acquisition throughout the lifespan, and also specific reasons for the obvious differences in the course and outcome of language acquisition in children and adults. This opens the door to investigating the contribution of these mechanisms to adult second language learning, and to exploring the differences between child and adult second language acquisition based on the cognitive structures that underlie and modulate frequency-based learning processes. In line with these findings, the present research investigates the sensitivity of adult second language learners to a subtle probabilistic subpattern of Spanish diphthongization in derived words, and compares their performance with that of native Spanish speakers. The Spanish diphthong/mid-vowel alternation may be explained through a highly consistent rule in Spanish verbs, but in suffixed forms such as those examined here, the probability that a given form will conform to the consistent rule varies according to the suffix used. The present experiments measure learners’ processing of neologisms created using real Spanish stems with alternating diphthongs and real derivational suffixes as a way of assessing their sensitivity to the probability of a diphthongized stem in specific morphological contexts. The consequences of this probabilistic pattern in diphthongization are measured in participants’ perception and production on a lexical decision task and a conditional naming task in which participants say palabra ‘word’ if they hear a real Spanish word, and repeat the nonword items, including the neologisms. Learners are also compared based on Spanish proficiency and on three measures of cognitive capacity accuracy in repeating words from an unfamiliar foreign language (Korean), inhibitory control, as measured on the Simon task, and working memory, based on reading span scores. Native speakers performing the same tasks are shown to be sensitive to this subtle probabilistic variability in their processing of neologisms, although the results vary in strength between the lexical decision and conditional naming tasks. Strikingly, learners also appear to be sensitive to this variability, despite the complex morpholexical and paradigmatic knowledge required to detect these patterns. However, learner behavior differs from that of native speakers in interesting ways. The lexical biases of some suffixes appear to be more easily acquired than others, and learner performance differs more dramatically between reaction time and error rate measures, and between lexical decision and conditional naming. This suggests that the morphophonological pattern in question may not affect all levels of lexical processing, and that its effects are also contingent to some extent on task demands. Finally, there is also evidence that phonological memory, and to a lesser extent working memory also modulate learners’ sensitivity to probabilistic variation in Spanish diphthongization, but these effects also vary from one dependent measure to another. The results have important implications for theories of adult second language acquisition, suggesting that adults are sensitive to similar kinds of statistical information in language experience to that exploited in emergentist theories of first language acquisition. This indicates that adult language acquisition may rely on many of the same cognitive processes as does child language acquisition, although adults’ more developed cognitive resources also lead to important differences.