Juffs - Feature Focus in Word Learning

PI	Ben Friedline, Alan Juffs
Start date	September 2009
End date	July 2010
Learnlab	English

L2 learning of derived words

Benjamin Friedline and Alan Juffs

Background Inflected and derived words occur frequently in the English language and serve important functions in everyday communication. The term derived word refers to the combination of a base word with a derivational affix. For instance, if the derivational affix –ness is added to the base kind (adjective), the word kindness (noun) is derived. The affix –ness is very productive and can be added to many words to derive novel words such as darkness, awareness, and illness. Other derivational affixes such as –ity are not as productive as –ness and can be used to form a limited number of words, such as ¬purity and scarcity. The term inflected word refers to the combination of a base word and an inflectional affix. For example, if the inflectional affix –ed is added to the base word walk (verb), the resulting word is walked (verb). The addition of the –ed affix is very productive in the formation of the past tense even though it does not apply to a number of irregular past tense forms, such as drove, ate, and sat.

Importantly, the differences in productivity of each type of affix have led some researchers to conclude that there are differences in how they are processed by native speakers. In Words and Rules (WR) theory, Pinker and Ullman (2002) argue that irregular inflected forms (e.g., drove) are stored in the lexicon (or mental dictionary) as whole words, whereas regular forms (e.g., walked) are generated by a regular rule. This theory has also been applied to the processing of derived words in two recent psychological investigations (cf. Alegre & Gordon, 1999; Hagiwara et al., 1999). The main point behind both of these studies is to illustrate that derived words can be either rule-governed (e.g., words with –ness) or stored as whole words in the lexicon (e.g., words with –ity as in purity).

In the area of second language acquisition, adult second language (L2) learners often fail to attain native-like proficiency when producing derived and inflected words in an L2. Lardiere (1998), for instance, showed that second language learners still make errors with inflectional morphology even after many years of exposure to English. In Lardiere’s (1998) study, she recorded and analyzed naturalistic conversations from a Chinese learner of American English. The results of this study indicated that the learner supplied the inflectional affix –ed correctly in only 34% of obligatory contexts even after 18 years of exposure to English. Additionally, in terms of derived words, a recent study by Juffs and Friedline (2010) revealed that intermediate L2 learners often made errors in the production of derived words such as those in examples (1) and (2). (1) We have one different [difference]. (2) I like doing something music [musical]. In example (1) the learner uses the adjective form different instead of the grammatically correct form difference, which is a noun. In example (2) the learner uses the noun form music in a position that requires the adjective form musical.

The preponderance of such errors in L2 speech has led some researchers to conclude that L2 learners are permanently impaired on the production of derived and inflected words because they do not have access to the same rule-based mechanisms that are present during L1 acquisition (Jiang, 2004; Felser & Clahsen, 2009; Silva & Clahsen, 2008). This hypothesis is formally known as the Fundamental Difference Hypothesis (FDH; Bley-Vroman, 1989), and it has received support from a number of recent studies in the field of second language acquisition. Silva and Clahsen (2008), for instance, use evidence from a masked-priming experiment to compare native speakers to adult L2 learners on a series of morphological priming tasks. The results from this study indicated full priming effects (e.g., darkness primes dark) for native speakers on both inflections and derivations, but only partial priming effects for L2 learners on derivations and no priming effects for L2 learners on inflections. Silva and Clahsen (2008) argue that the limited priming effects (or complete lack thereof) indicate that L2 learners lack rule-based mechanisms and do not know that ¬–ness can be affixed to many adjectives to derive nouns such as darkness, awareness, and illness. This lack of rule-based mechanisms may mean that adult L2 learners memorize all words as unanalyzed chunks of language, without realizing that dark and darkness or walk and walked are intimately related in both form and meaning.

The research by Silva and Clahsen (2008) makes an important contribution to a theory of second language acquisition because it provides a possible explanation for why L2 learners have difficulties with derived and inflected words (i.e., they cannot access rule-based mechanisms). However, this research is limited because it assumes that L2 learners are permanently impaired when compared to native-speakers on all types of rule-based inflectional and derivational morphological processes. This assumption is at odds with findings from studies such as the morpheme order studies (e.g., Bailey, Madden, & Krashen, 1974), critical period studies (e.g., Johnson & Newport, 1989), and studies on the role of morpheme salience in L2 acquisition (e.g., Ellis, 2006) in that these past studies indicate that certain morphemes, such as progressive, may be more easily acquired than others. Johnson and Newport (1989), for instance, claim that the progressive morpheme may not be subject to critical period effects.

Additionally, the role of a learner’s first language may also influence the relative difficulty of a particular morpheme. Potential L1 effects have been discussed in a number of recent SLA studies (e.g., Juffs & Friedline, 2010; White, 2003). Before we conclude that L2 learners are equally impaired in all areas of morphological knowledge, further research is needed to identify if certain morphological structures are easier to acquire than others, how exactly L2 morphological knowledge diverges from native-speaker knowledge, and how a learner’s first language might influence L2 morphological knowledge. The goal of the present research is to answer these questions as they relate to derivational morphology.

Research Questions

Why are ESL learners so poor in their knowledge of English morphology? What are the knowledge components that are the most challenging for learning through normal language exposure? Do learners have a representational problem or a processing problem? Specifically, what instructional interventions can be designed to overcome observed processing differences in L1 and L2 morphology?

Research plan

For year 1, the goal of the research is to analyze the knowledge components of ESL learners to lay the groundwork for a hypothesis-based intervention. The research will systematically investigate the components of L2 learners’ knowledge of English derivational morphology to address the following questions:

1) What are the components of L2 derivational knowledge? 2) Are these components different from L1 derivational knowledge? 3) Does L1 matter for the acquisition of derived words in an L2?

Methodology

To answer these questions, Friedline has developed a series of tasks that will be used to assess what native English speakers and second language learners know about derived words. These tasks included lexical decision, semantic relatedness, and morphological decomposition. Each of these tasks contained several conditions that tested different components of morphological knowledge. Studies on the acquisition of L1 morphological knowledge (e.g., Carlisle, 2000; Carlisle & Fleming, 2003) were consulted in order to develop these conditions. Each condition is outlined below.

Lexical decision task

Explanation: In this task, students were asked to rate words from 1 (not a word) or 6 (definitely a word). All words were morphologically complex (e.g., base + affix). Some of the words were real words in English, while other words were not real words in English. The purpose of this task was to assess if native-speakers were sensitive to the effects of semantic blocking and affix ordering. There were four conditions in this task. The conditions are listed below along with an example to illustrate the types of words that were presented in each condition.

Condition 1: Real words Example: The suffix –able is added to verbs to derive adjectives such as workable or comfortable. A response of 4, 5, or 6 would be counted as accurate.

Condition 2: Semantic blocking Example: Even though you can add the affix –able to many verbs to derive adjectives, there are some verbs like arrivable and departable look that do not normally take the suffix –able to form adjectives. A response of 1, 2, or 3 would be counted as accurate.

Condition 3: Correct affix ordering Example: There are some bases that can take two affixes. You can add the affix –able to the verb respect to derive the adjective respectable. Then, you can add the affix –ity to respectable to derive the noun respectability. A response of 4, 5, or 6 would be counted as accurate.

Condition 4: Incorrect affix ordering Example: In a word like respectability, the word is correct because the affixes are added in the correct order. However, if I add the affix -ity before I add the affix –able, I derive a word like respectitiable. This word is not correct because the affixes are not added in the correct order. A response of 1, 2, or 3 would be counted as accurate.

Word relatedness task

Explanation: In this task, students were asked to rate words based on their meaning from 1 (not related) to 6 (definitely related). There were five conditions in this exercise.

Condition 1: No relationship in meaning Some words are not related in meaning in any way. The words cat and bus are not related in meaning in any way. A response of 1, 2, or 3 would be counted as accurate.

Condition 2: Relationship in meaning Other words are related in meaning. For instance, bank and money are related in that a bank is a place where you deposit your money. A response of 4, 5, or 6 would be counted as accurate.

Condition 3: Relationship in meaning with different affixes. This condition contained words with suffixes that were related in meaning. For example, productive (adj.) and production (n.) both share the base produce (v.). A response of 4, 5, or 6 would be counted as accurate.

Condition 4: Relationship in orthography only, not meaning There are some words that may look like they are related in meaning because they share the same initial letters. In this condition, students saw words like permanence and permission. These words share the letters p-e-r-m, but are unrelated in meaning. A response of 1, 2, or 3 would be counted as accurate.

Condition 5: Relationship in affix only, not meaning In the final condition, students were presented with words that shared the same affix, but were unrelated in meaning. For example, the words reality and curiosity are unrelated in meaning, but share the affix –ity. A response of 1, 2, or 3 would be counted as accurate.

Word Analysis Task

Explanation: On the Word Analysis Task, students were asked to provide the base word of the word provided. Some of these words consisted of a base and an affix such as musician, which has music as a base. Other words, however, could not be broken down into a base and a affix. For instance, dollar cannot be broken down into doll + ar because dollar is a base form. Accuracy was computed for decomposable and non-decomposable words.

Native speakers piloted these tassk in the fall of 2009, and preliminary results are reported for each task in the tables below. A pull out from the ELI in Spring 2010 will collect learner data.

Participants

These tasks were administered to native speakers and L2 learners during the fall 2009 and spring 2010 semesters. A total of 23 native-English speakers participated in the study. All of the native speakers were undergraduates at the University of Pittsburgh. Ninety ESL learners participated in this study from three different levels of language proficiency: beginner (n=26), intermediate (n=36), and advanced (n=28). These learners were enrolled in an intensive English program at the University of Pittsburgh.

Descriptive Results

Lexical Decision Task

Condition	NS Accuracy
Condition 1: Real words	93%
Condition 2: Semantic blocking	81%
Condition 3: Correct affix ordering	93%
Condition 4: Incorrect Affix ordering	95%

Condition	L2 Beginner Accuracy
Condition 1: Real words	89%
Condition 2: Semantic blocking	48%
Condition 3: Correct affix ordering	66%
Condition 4: Incorrect Affix ordering	53%

Condition	L2 Intermediate Accuracy
Condition 1: Real words	90%
Condition 2: Semantic blocking	59%
Condition 3: Correct affix ordering	69%
Condition 4: Incorrect Affix ordering	70%

Condition	L2 Advanced Accuracy
Condition 1: Real words	96%
Condition 2: Semantic blocking	53%
Condition 3: Correct affix ordering	77%
Condition 4: Incorrect Affix ordering	81%

Word relatedness task

Condition	NS Accuracy
Condition 1: No relationship in meaning	92%
Condition 2: Relationship in meaning	92%
Condition 3: Relationship in meaning with different affixes	97%
Condition 4: Relationship in orthography only	89%
Condition 5: Relationship in affix only	90%

Condition	L2 Beginner Accuracy
Condition 1: No relationship in meaning	91%
Condition 2: Relationship in meaning	84%
Condition 3: Relationship in meaning with different affixes	89%
Condition 4: Relationship in orthography only	73%
Condition 5: Relationship in affix only	74%

Condition	L2 Intermediate Accuracy
Condition 1: No relationship in meaning	89%
Condition 2: Relationship in meaning	87%
Condition 3: Relationship in meaning with different affixes	89%
Condition 4: Relationship in orthography only	70%
Condition 5: Relationship in affix only	69%

Condition	L2 Advanced Accuracy
Condition 1: No relationship in meaning	95%
Condition 2: Relationship in meaning	79%
Condition 3: Relationship in meaning with different affixes	96%
Condition 4: Relationship in orthography only	81%
Condition 5: Relationship in affix only	89%

Word Analysis Task

Condition	NS Accuracy
Condition 1: Decomposable	85%
Condition 2: Non-decomposable	92%

Condition	L2 Beginner Accuracy
Condition 1: Decomposable	59%
Condition 2: Non-decomposable	73%

Condition	L2 Intermediate Accuracy
Condition 1: Decomposable	61%
Condition 2: Non-decomposable	84%

Condition	L2 Advanced Accuracy
Condition 1: Decomposable	65%
Condition 2: Non-decomposable	81%

Discussion

Beginning L2 Learners (Level 3)

The results of this study indicate that level 3 learners from the ELI at the University of Pittsburgh often have problems when processing derived words. Firstly, the lexical decision task may indicate that beginning learners are not sensitive to constraints on the use of morphemes such as –able and –ness or constraints on the ordering of affixes. For instance, the learners in the present study often judge words such as departable and hopenessful to be real English words in spite of the fact that most native speakers rarely (if ever) consider these words to be real English words. Sixteen of 26 level 3 learners said that smileable was a real word in English, while only 1 of 23 native speakers said that this was a real word in English. Likewise, for hopenessful, 18 of 26 learners said that this was a real word, but only 1 of 23 natives considered hopenessful to be a real English word. Second, the word relatedness task seems to indicate that learners rely heavily on orthographic/phonological overlap when processing the meaning of words. Put another way, the learners in this study said that word pairs that were related in form only were also related in meaning. For instance, 12 out of 26 level 3 learners said that the word pair majority-activity were related in meaning, while native speakers (N=23) never say that these words are related in meaning. Finally, the results from the word analysis task may suggest that learners at this level have significant difficulties with derived words that involve phonological and/or orthographic changes to the base. In the present study, almost all of the level 3 learners (22 of 26) incorrectly provided the base word for extension. Native speakers, on the other hand, provided the incorrect base for extension only 4 times out of 23 subjects.

Intermediate L2 Learners (Level 4)

For Level 4 L2 learners, the lexical decision task may indicate that even intermediate-level learners are not sensitive to constraints on the use of morphemes such as –able and –ness or constraints on the ordering of affixes. For instance, the learners in the present study often judge words such as departable and hopenessful to be real English words in spite of the fact that most native speakers rarely (if ever) consider these words to be real English words. Seventeen of 36 level 4 learners said that smileable was a real word in English, while only 1 of 23 native speakers said that this was a real word in English. Likewise, for hopenessful, 21 of 36 learners said that this was a real word, but only 1 of 23 natives considered hopenessful to be a real English word. Second, the word relatedness task seems to indicate that learners rely heavily on orthographic/phonological overlap when processing the meaning of words. Put another way, the learners in this study said that word pairs that were related in form only were also related in meaning. For instance, 14 out of 36 level 3 learners said that the word pair majority-activity were related in meaning, while native speakers (N=23) never say that these words are related in meaning. Finally, the results from the word analysis task may suggest that learners at this level have significant difficulties with derived words that involve phonological and/or orthographic changes to the base. In the present study, more than two-thirds of the level 4 learners (25 of 36) incorrectly provided the base word for extension. Native speakers, on the other hand, provided the incorrect base for extension only 4 times out of 23 subjects.

Advanced L2 Learners (Level 5)

For level 5 L2 learners, the lexical decision task may indicate that even advanced learners are not sensitive to constraints on the use of morphemes such as –able and –ness. For instance, the learners in the present study often judge words such as smileable and leavable to be real English words in spite of the fact that most native speakers rarely (if ever) consider these words to be real English words. Fourteen of 28 level 5 learners said that smileable was a real word in English, while only 1 of 23 native speakers said that this was a real word in English. Likewise, for leavable, 19 of 28 learners said that this was a real word, but only 4 of 23 natives considered leavable to be a real English word. Second, the word relatedness task seems to indicate that advanced learners still rely to some degree on orthographic/phonological overlap when processing the meaning of words. Put another way, some advanced learners in this study said that word pairs that were related in form only were also related in meaning. For instance, 11 out of 28 level 3 learners said that the word pair constantly-conservative were related in meaning, while native speakers (N=23) never say that these words are related in meaning. Finally, the results from the word analysis task may suggest that learners at level 5 have significant difficulties with derived words that involve phonological and/or orthographic changes to the base. In the present study, many of the errors on the word analysis task were errors on words that involved a significant orthographic and sometimes phonological change to the base. For instance, 19 of 28 level 5 students incorrectly provided the base word for extension. Native speakers, on the other hand, provided the incorrect base for extension only 4 times out of 23 subjects.

General Discussion

• • Forthcoming in the first author's doctoral dissertation**

Selected References

Alegre, M., & Gordon, P. (1999). Rule-based versus associative processes in derivational morphology. Brain and Language, 68, 347-354. Bailey, N., Madden, C., & Krashen, S. (1974). Is there a "natural sequence" in adult second language learning? Language Learning, 24(2), 234-243. Bley-Vroman, R. (1989). The logical problem of second language learning. In S. Gass & J. Schachter (Eds.), Linguistic Perspectives on Second Language Acquisition. Cambridge: Cambridge University Press. Carlisle, J.F. (2000). Awareness of the structure and meaning of morphologically complex words: Impact on reading. Reading and Writing, 12(3-4), 169-190. Carlisle, J. F., & Fleming, J. (2003). Lexical processing of morphologically complex words in the elementary years. Scientific Studies of Reading, 7(3), 239-253. Ellis, N. C. (2006). Selective attention and transfer phenomena in L2 acquisition: Contingency, cue competition, salience, interference, overshadowing, blocking, and perceptual learning. Applied Linguistics, 27(2), 164-194. Felser, C., & Clahsen, H. (2009). Grammatical processing of spoken language in child and adult language learners. Journal of Psycholinguistic Research, 38(3), 305-319. Gonnerman, L. M., Seidenberg, M. S., & Andersen, E. S. (2007). Graded semantic and phonological similarity effect in priming: Evidence for a distributed connectionist approach to morphology. Journal of Experimental Psychology, 136(2), 323-345. Hagiwara, H., Sugioka, Y., Ito, T., Kawamura, M., & Shiota, J.-i. (1999). Neurolinguistic evidence for rule-based nominal suffixation. Language, 75(4), 739-763. Hay, J. (2002). From speech perception to morphology: Affix ordering revisited. Language, 72(3), 527-555. Hay, J. B., & Baayen, R. H. (2005). Shifting paradigms: gradient structure in morphology. TRENDS in Cognitive Science, 9(7), 342-348. Jiang, N. (2004). Morphological insensitivity in second language processing. Applied Psycholinguistics, 25, 603-634. Johnson, J. S., & Newport, E. L. (1989). Critical period effects in secon dlanguage learning: The influence of maturational state on the acquisition of English as a second language. Cognitive Psychology, 21, 60-99. Juffs, A. & Friedline, B. (2010). L1 influence, morphological (in)sensitivity and L2 lexical development: Evidence from production data. Unpublished manuscript, University of Pittsburgh, PA. Lardiere, D. (1998). Dissociating syntax from morphology in a divergent L2 end-state grammar. Second Language Research, 14(4), 359-375. Lardiere, D. (2006). Ultimate attainment in second language acquisition: a case study. New York: Routledge Marslen-Wilson, W. D., Bozic, M., & Randall, B. (2008). Early decomposition in visual word recognition: Dissociating morphology, form, and meaning. Language and Cognitive Processes, 23(3), 394-421. Nation, I. S. P. (2001). Learning vocabulary in another language. New York: Cambridge University Press. Pinker, S., & Ullman, M. T. (2002). The past and future of the past tense. TRENDS in Cognitive Science, 6(11), 456-463. Silva, R., & Clahsen, H. (2008). Morphologically complex words in L1 and L2 processing: Evidence from masked priming experiments in English. Bilingualism: Language and Cognition, 11(2), 245-260. White, L. (2003). Fossilization in steady state L2 grammars: Persistent problems with inflectional morphology. Bilingualism: Language and Cognition, 6(2), 129-141.