The relationship between the morphological knowledge and L2 online processing of derivational words

Abstract

Objectives/Research Questions:

Two experiments were conducted to investigate the effect of morphological knowledge on second language (L2) online processing of derivational words by Chinese first language (L1)–English L2 learners.

Methodology:

Experiment 1 was performed at the single word level with a masked priming paradigm using the lexical decision task. Experiment 2 was conducted at the sentence level with a self-paced reading paradigm. According to their performance on the morphological knowledge test, the Chinese (L1)–English (L2) learners were grouped into two groups: a high morphological knowledge group and a low morphological knowledge group; and these were matched for proficiency.

Data and Analysis:

Data were analyzed using a repeated-measures analysis of variance. In Experiment 1, mean response times and accuracy of the derived suffixed primes were analyzed between the two groups. In Experiment 2, mean response times in the regions of interest were analyzed between the two groups.

Findings/Conclusions:

In Experiment 1, results showed that a full priming effect was evident in the group with high morphological knowledge, indicating application of rule-based decompositions, whereas this effect was not observed in the group with low morphological knowledge. In Experiment 2, the group with high morphological knowledge had a significant processing cost in processing the derivational words, suggesting the employment of rule-governed decomposition. These results suggest that morphological knowledge plays an important role in online L2 morphological processing.

Originality:

In L2 morphology literature, the current study empirically explored the role of morphological knowledge in derivational processing.

Significance/Implications:

The study suggested that improvements in morphological knowledge can facilitate the rule-based processing of L2.

Keywords

Morphological knowledge derivational morpheme second language (L2)decomposition whole-word representations

Introduction

Morphologically complex words can be used to investigate whether people automatically employ rule-based decompositions when processing such words. For example, the word darkness is composed of the morphemes dark and -ness. Do people decompose morphologically complex words into their individual morphemes, or are they processed as whole units?

In first language (L1) research, many studies have investigated the role of morphology in visual word recognition using a masked priming paradigm. A word can be recognized faster when it is preceded by a morphologically related word (e.g., darkness-dark), compared to when it is preceded by an unrelated word (e.g., book-dark), which provided direct support for the dual-mechanism model (Clahsen, 1999; Pinker, 1999; Pinker & Ullman, 2002). The dual-mechanism model posits two distinct representational systems and corresponding processing mechanisms for morphologically complex words. For example, regular past-tense forms in English (e.g., walk+ed) are regarded to have morphologically structured representations, making them suitable for application of the rule-governed decomposition in online processing. Irregular past-tense forms (e.g., ran) are claimed to have whole-word representations stored in the memory and to be directly accessed from the separate lexical entries during processing (Clahsen, 1999; Pinker, 1999; Pinker & Ullman, 2002). A number of studies on English inflectional words appear to support the dual-mechanism model (Longworth, Marslen-Wilson, Randall, & Tyler, 2005; Miozzo, 2003; Ullman et al., 2005). For the derived words, most of the evidences for the morphologically structured representations also come from priming studies (Clahsen, Sonnenstuhl, & Blevins, 2003; Feldman & Prostko, 2002; Marslen-Wilson, Tyler, Waksler, & Older, 1994); for example, Clahsen et al. (2003) investigated three fully productive and semantically transparent German derived words, marked by the suffixes -un, -chen and -lein. With a cross-modal immediate repetition priming task, the results showed the full priming effect¹ of derivational words, indicating that the recognition of such forms involves the activation of the stem form in addition to the associated suffix. The results gave support to the view that the full priming effect for the derived words can be explained based on the shared stem entry of the prime and the target. In short, much evidence from both inflected and derived words in L1 provides support that native speakers apply rule-based decompositions in online processing of morphologically complex words.

Recently, a number of researchers have focused on the processing of morphologically complex words in a second language (L2), aiming to investigate whether L2 learners can also apply rule-based decomposition, and to explore the factors that may influence the processing performance of the L2 learners (Lehtonen & Laine, 2003; Pliatsikas & Marinis, 2013; Pliatsikas, Johnstone, & Marinis, 2014; Silva & Clahsen, 2008). Due to a limited number of studies, however, the picture of L2 morphological processing is far from clear. Among these studies, most have focused on the processing of inflected words (Lehtonen & Laine, 2003; Neubauer & Clahsen, 2009; Pliatsikas & Marinis, 2013; Pliatsikas, Johnstone, & Marinis, 2014; Silva & Clahsen, 2008). For instance, in the priming experiment of Neubauer and Clahsen (2009), the researchers investigated processing of German regular and irregular participles by highly proficient Polish (L1)–German (L2) learners. Results showed that, while facilitative priming effects were found for irregular and regular participles in the L1 group, no priming effects for regular processing and only partial priming effects for the irregular participles were found in the L2 learners. This suggests that L2 learners do not decompose inflectional affixes from their stems as native speakers do, which was consistent with the findings of Silva and Clahsen (2008) that L2 learners rely more heavily on whole-word processing for inflected words. In the study of Pliatsikas and Marinis (2013), the target words (inflected words or pseudo-inflected forms) were arranged into a plausible sentence, which was divided into six segments. The reading time spent on the key segment, including target words, was analyzed. According to the researchers, the evidence in favor of decomposition is provided by the processing cost, which will be caused by relatively longer reading time in processing the regularly inflected forms than in the irregularly inflected forms, which are directly accessed as the whole-forms. The longer reading time for the regular verbs may be explained by the activation of the regular rule, which leads to the decomposition of the inflected form (Kirkici, 2005). The results revealed a greater processing cost for the regularly inflected words than irregularly inflected ones, indicating that highly proficient L2 learners (Greek–English bilinguals) applied rule-based decompositions.

Although regularly and irregularly inflected words have been at the center of debate regarding decomposition or whole-word storage in L2, derived words may offer an additional, and perhaps better controlled testing ground for the rule-application. Inflection does not create new words with new meanings, and is sensitive to the grammatical environment in which it occurs. The grammatical information carried by inflection is not only about the stem itself, but also about the phrasal and sentential interpretation to which the stem relates (Bickel & Nichols, 2007; Bozic, Tyler, Su, Wingfield, & Marslen-Wilson, 2013). Therefore, the processing of inflection may not reflect the complete lexical processing of the inflected word itself. However, derivation creates new words that may have new meanings. Processing of derivation morphology is a lexical one, which is much more independent of grammatical context (Bickel & Nichols, 2007; Bozic et al., 2013). Therefore, the processing of derivational morphology may reflect a much purer processing of decomposition of the derived words in natural language processing.

Studies on the processing of derived words in L2 using online techniques have provided inconclusive results (Diependaele, Duñabeitia, Morris, & Keuleers, 2011; Heyer & Clahsen, 2015; Silva & Clahsen, 2008). Silva and Clahsen (2008) investigated deadjectival nominalizations with -ness and -ity in a masked priming experiment. A reduced morphological priming effect was observed in L2 learners with suffix-derived primes, which made the authors conclude that L2 learners employ morphologically structured representations for derived word forms in a less effective way. In the study of Diependaele et al. (2011), three experiments were conducted to investigate whether Spanish–English and Dutch–English bilinguals could engage the combinatorial processing of derived words in a masked morphological priming lexical decision task. The linguistic properties of the suffixes were manipulated by including the transparent suffixed primes, opaque suffixed or pseudo-suffixed primes, and form control primes. A graded pattern of priming was observed across these three conditions, with the largest priming effect in the transparent condition, smallest in the form condition, and intermediate in the opaque condition, indicating that L2 learners can fully engage rule-based decomposition in real time. Heyer and Clahsen (2015) directly compared orthographically related and derived prime-target pairs with English native speakers and German (L1)–English (L2) bilinguals. While native speakers indicated morphological but not formal overlap priming, the L2 learners showed the same magnitudes of facilitation for the two prime types, indicating that these learners were more influenced by surface-form properties in L2 than in L1. Obviously, the results of processing L2 morphologically complex words are inconsistent. Further study is required before stronger conclusions can be drawn.

A number of factors have been suggested to affect the processing of morphologically complex words in L2. The factors such as L1 transfer, cognitive resource limitations (Clahsen, Felser, Neubauer, Sato, & Silva, 2010), L2 proficiency, or level of L2 exposure are all believed to influence the application of rule-based decomposition in L2 words processing (Liang & Chen, 2014; Ullman, 2001, 2004). For example, in the event-related potential study of Liang and Chen (2014), the role of proficiency was supported by the evidence that highly proficient L2 learners manifested rule-governed decomposition in processing morphologically complex words, while less proficient L2 learners relied much more on lexical storage, in accord with the predictions of the declarative/procedural model proposed by Ullman (2004, 2012). According to this model, proficiency is one of the important factors that may affect whether L2 learners can apply rule-governed decomposition. More specifically, at the low proficiency level, the morphological processing of the L2 learners may depend on the declarative memory system. Then with increases in proficiency, they may come to rely more on the procedural memory. At the same time, this theory is also open in principle to the possibility that other factors besides proﬁciency might affect the attainment of rule-based decomposition (Ullman, 2005, 2012). As far as we know, few studies have explored the relationship between L2 morphological knowledge and online morphological processing. It is not yet known whether the morphological knowledge that L2 learners have accumulated will affect the online processing of L2 morphologically complex words. Since in L1, the acquisition of morphological knowledge can develop into the automatic morphological decomposition process (Kraut, 2015; Murrell & Morton, 1974; Taft, 1979; Taft & Forster, 1975), research is needed to investigate its relationship to the development of automatic morphological decomposition processed in L2 learners of English. Therefore, the current study aims to directly explore the relationship between morphological knowledge and L2 morphological decomposition by focusing on derivational morphology.

Morphological knowledge, is a metacognitive skill (Clark, Gilbert, & Anderson, 2011), including morphological awareness (Bowers, Kirby, & Deacon, 2010; Nagy, Carlisle, & Goodwin, 2014), and refers to a more general or superordinate term (Nagy et al., 2014). It refers to the understanding and mastery of the word-formation rules of morphologically complex words. It concerns the ability of the L2 learners to parse and manipulate the morphemes of the morphologically complex words.

Reading is the “process of simultaneously extracting and constructing meaning through interaction and involvement with written language” (RAND Reading Study Group, 2002), which may be affected by many variables such as word recognition (Perfetti & Hogaboam, 1975), decoding (Hoover & Gough, 1990), reading fluency (Padeliadu & Antoniou, 2014), vocabulary (Quellette, 2006), and morphological knowledge (Tong, Deacon, Kirby, Cain, & Parrila, 2011). Morphological knowledge is especially crucial for many different reading skills (Kieffer & Lesaux, 2012) and can facilitate word recognition and reading (Goodwin, Huggins, Carlo, August, & Calderon, 2013).

Some studies have indicated the robust reciprocal relationship between morphological knowledge and reading in L1 (Carlisle, 2000; Deacon, Benere, & Pasquarella, 2013; Ku & Anderson, 2003; Mahony, Singson, & Mann, 2000; Nagy, Berninger, & Abbott, 2006; Tong et al., 2011). Research has consistently found that morphological knowledge, especially the knowledge about derivational suffixes, explains the unique variance in reading after controlling for other related variables (Carlisle & Nomanbhoy, 1993; Singson, Mahony, & Mann, 2000). For example, Deacon et al. (2013) have empirically demonstrated a reciprocal relationship between children’s morphological awareness and their reading accuracy, by the evidence that children’s morphological awareness was associated with their growth in reading to the same extent that their reading was with their growth in morphological awareness (Deacon et al., 2013). As pointed out by Deacon et al. (2013), children might learn about the morphological structure of the words when they encounter them. For example, when they read the word “darkness,” they will see the subcomponents “dark” and “-ness” and thus learn to decompose words. Therefore, a number of morphologically complex words in print that children encounter accumulatively may increase the likelihood of improvements in their morphological awareness, giving support for the prediction that reading supports the development of morphological awareness.

In brief, evidence from L1 demonstrated the bidirectional relationship between reading and morphological knowledge. Reading may help improve children’s morphological knowledge. Thus, it may be possible that this relationship can be extended to adult L2 learners as well. As yet, there are few studies that directly investigate the relationship between L2 morphological knowledge and automatic morphological decomposition. For adult L2 learners, there are many input sources, such as reading print materials like textbooks, novels, and magazines, along with watching TV or films. Increased exposure to the L2 affords increased possibilities for encountering both familiar and new morphologically complex words. Each encounter of a morphologically complex word allows for the possibility that the learner may perceive the subcomponents of the words, thus strengthening their understanding of the word formation rules of these complex words. With increased understanding of the word formation rules, the learners who accumulate higher morphological knowledge through reading or other sources may apply more rule-governed decomposition in online morphological processing. Therefore, it would be possible to predict that the processing mechanism may be different between the bilinguals with higher morphological knowledge and those with lower morphological knowledge, with the former utilizing much more rule-based decompositions.

The present study focused on investigating the relationship between the morphological knowledge and L2 online processing of derivational words. Two experiments were conducted. In Experiment 1, we used a masked priming paradigm, and participants completed a lexical decision task. If morphological knowledge plays a key role in morphological processing, then the group with higher morphological knowledge would tend to apply rule-governed decomposition and the group with lower morphological knowledge would tend to directly access whole forms. We predicted that the morphological priming effect in the high morphological knowledge group would be greater than that in the low morphological knowledge group. Also, a morphological priming effect would be present in the high morphological knowledge group but absent in the low morphological knowledge group.

Paradis (2004) suggested that single word studies are not the most suitable to understand actual language processing, because words do not usually appear in isolation in authentic language contexts. Therefore, it is more ecologically valid to investigate morphological processing at the sentence level, in order to reveal the nature of derivational processing by L2 learners. Thus, in Experiment 2, the target derivational words were embedded in plausible simple structure sentences, and the reading times on the key segments were analyzed. Participants with high morphological knowledge were hypothesized to automatically apply rule-based decomposition mechanisms online, having strengthened their understanding of the word formation rules in exposure to the complex words. Therefore, a decomposition processing cost, as described by Pliatsikas and Marinis (2013), could be present in this group. Compared to processing monomorphemic words, the processing cost for derivational words can be explained by the activation of the formation rule, which automatically leads to the decomposition of the derivational words. Thus, if morphological knowledge really affects L2 morphological processing, we predict that a greater processing cost would be evident in the group with high morphological knowledge.

Experiment 1

Experiment 1 aimed to investigate the effect of morphological knowledge on the processing of suffix-derivations at the single-word level.

Method

Participants

Participants were 40 Chinese (L1)–English (L2) college students enrolled at Beijing Normal University. They either passed the College English Test 4 (CET 4) with scores higher than 550 or passed the CET 6. The CET is administered by the Ministry of Education of China for non-English major college students, with 710 as the highest score and 425 as the lowest. It includes two levels of tests; CET 6 is the higher level one. All participants were right-handed, healthy, native speakers of Chinese with normal or corrected-to-normal vision.

The participants were divided into two groups according to their median score on a morphological knowledge test. The group with high morphological knowledge (HG) consisted of 20 participants (11 female), and the group with low morphological knowledge (LG) consisted of 20 participants (13 female).

The morphological knowledge test was administered to test the derivational knowledge of the morphological structure of derived words in L2 English. It was designed to assess the L2 learners’ awareness of the relations of bases and their derived forms. The items in this morphological knowledge test were adapted from previous studies or reconstructed according to the rules adopted by these studies (Carlisle, 2000; Wang, Cheng, & Chen, 2006). In the studies by Carlisle (2000) and Wang et al. (2006), the derivational morphology task was used to assess students’ awareness of the derivational structure of English words and ability to define morphologically complex words. This task requires the knowledge about the grammatical roles and meanings of suffixes, not just relational knowledge about the suffix and stems. Carlisle’s (2000) participants were asked to complete a sentence based on the clue words. For example, for the clue word “farm,” the sentence to be completed was, “My uncle is a __” (correct answer: “farmer”). As these items were designed to measure the morphological awareness of monolingual English-speaking children, they may not be completely appropriate for adult L2 learners in the present study. Therefore, we revised and reconstructed some items. The whole morphological knowledge test for the present study included two parts. The first part required the production of a derived word in order to finish a sentence (hence called derivation). For example, for the clue word “bold,” the sentence to be completed was, “Linda’s __always gets her into trouble” (correct answer: “boldness”). This task was used to assess the L2 learners’ awareness of the morphological structure of words and placed emphasis on their ability to produce the complex forms given the base word and a sentence context. The second part of the morphological knowledge test was a true/false judgment task requiring participants to judge whether the given word was the correct derivational form of the base word, for example “deaf”-“deafment” (incorrect) or “deaf”-“deafness” (correct). These items assessed the correct understanding of the word relations between the base word and its derived word. After a relatively large-scale testing with a separate group of 137 students who had a similar background to the participants, items with a passing rate higher than 0.8 or lower than 0.2 were removed, resulting in 100 items in total. Each part of the test for morphological knowledge included 50 items, and each correct answer was awarded 1 point, so that scores on the complete test could range from 0 to 100. The Cronbach’s alpha coefficient of the test was 0.92. For the grouping criterion, we adopted the median split method to classify the participants into two groups. First, the median of all the participants was established. Then, those whose scores were below the median were classified as the group with low morphological knowledge; those whose scores were equal or above the median were classified as the group with high morphological knowledge.

English proficiency scores were obtained through the Oxford Placement Test (OPT) as well as a six-point scale self-rating questionnaire. The OPT includes 25 multiple-choice questions and a cloze test, where the highest possible score is 50. The English listening, speaking, reading, and writing skills of the participants were assessed by their self-ratings (1 for “quite poor”, 6 for “highly proficient”). The participants’ mean age, years of learning English, scores on the morphological knowledge test, self-assessment scores, and OPT scores are presented in Table 1.

Table 1.

Age and second language history and proficiency rating and tests (standard deviations) for both groups.

Group	LG	HG
Age	22.70 (1.75)	23.10 (1.43)
Years of learning English	10.30 (1.01)	10.80 (0.82)
Morphological knowledge	58.28 (4.25)	76.70 (2.86)
Self-rating: listening	4.15 (0.51)	4.41 (0.29)
Self-rating: speaking	4.33 (0.33)	4.38 (0.34)
Self-rating: reading	5.74 (0.45)	5.56 (0.22)
Self-rating: writing	4.88 (0.23)	5.01 (0.32)
Oxford Placement Test	40.90 (4.08)	40.30 (5.24)

Note: LG = low morphological knowledge group; HG =high morphological knowledge group.

A paired-sample t-test verified that there were statistically significant differences in the scores of morphological knowledge between the two groups, t(19) = 7.64, p < 0.01, d = 0.83. There were no differences in age, t(19) = 0.91, p > 0.01, and in years of English learning, t(19) = 1.64, p > 0.01, which indicated that both groups were matched on age and years of English learning. The two groups were not significantly different in their OPT scores, self-rated L2 speaking, reading, and writing (all ps > 0.05), but did differ in listening self-ratings, t(19) = 2.46, p < 0.05, d = 0.17. So, for the most part, the two groups were matched on English proficiency.

Materials and design

Thirty-six derivational words with the suffixes -ness, -ity, or -ment, were chosen as the prime words because of their high-frequency and rich morphological productivity (Kuo & Anderson, 2006). The corresponding stems were taken as the target words (e.g., darkness-dark). Three priming conditions were designed for the derived suffix condition: derived prime (darkness-dark), repetition prime (dark-dark), and unrelated prime (object-dark). Another 36 pairs of nouns similar in forms (e.g., scandal-scan, the initial letters repetition between the prime and the target was 51.2% on average) were selected as the form control condition to exclude the possible effect of word-form priming. Correspondingly, three types of primes were created for the form control condition: form prime (scandal-scan); repetition prime (scan-scan); and unrelated prime (nature-scan). Three well-matched wordlists were constructed to guarantee that participants did not see the same target twice.

The frequencies of the priming stimuli were calculated based on the frequency dictionary of Brysbaert and New (2009). A separate group of 35 Chinese–English bilinguals, who did not participate in the lexical decision tasks, rated the familiarity of these English primes on a 5-point scale, with 1 being not familiar and 5 being very familiar. The matching results of priming words in different conditions are reported in Table 2.

Table 2.

Means on the word length, frequency and familiarity of the priming stimuli (standard deviations).

Prime type	Derived	Unrelated	Form	Unrelated
Length	8.31 (0.95)	6.67 (3.38)	6.53 (0.81)	7.36 (3.18)
Frequency	10 (12.27)	28 (53.91)	42 (58.10)	35 (53.78)
Familiarity	4.22 (0.30)	4.23 (0.35)	4.50 (0.25)	4.81 (0.25)

Paired sample t-tests showed that no significant differences were found between derived primes and unrelated primes in length, frequency, and familiarity (all ps > 0.05). Also, no significant differences were found between form primes and unrelated primes in length, frequency, and familiarity (all ps > 0.05). By changing 2 to 3 letters of the target words, 72 nonwords were created as fillers. Thus, each wordlist included 144 trials with 72 real words and 72 nonwords.

Procedure

Each participant was tested individually and completed the priming experiment first followed by the self-rating questionnaires, the morphological test, and the OPT.

Stimuli were presented using E-Prime software version 1.1. A trial began with the central presentation of a forward mask (a row of #####). After 500 ms, the mask disappeared and the prime was presented for 50 ms; the prime was immediately followed by the target, which stayed on the screen until the participants made a judgment. All the trials were presented randomly, with 1000 ms as the interval between trials. Participants were asked to decide as quickly and accurately as possible whether the target was a real word in English or a nonword and to make their response with a corresponding button press on the keyboard. Participants completed 10 practice trials prior to the start of the formal experiment.

Results and discussion

Analyses were performed on the mean judgment accuracy and on mean response times (RTs) of correct trials. RTs beyond three standard deviations (SDs) of the grand mean were excluded (5.29%).

Derived suffixed primes

Separate two factor analyses of variance (ANOVAs) were performed both by participants (F₁) and by items (F₂) on the mean RTs and accuracy. Group (high morphological knowledge, low morphological knowledge) was a between subject variable, prime type (repetition, derived, unrelated) was a within subject variable. The mean RTs and accuracy of the derived suffixed primes are listed in Table 3.

Table 3.

Mean response times (RTs) (ms) and accuracy (%) on the three different primes (standard deviations).

	LG			HG
	Repetition	Derived	Unrelated	Repetition	Derived	Unrelated
RT	655 (123)	719 (149)	736 (121)	664 (114)	685 (95)	812 (110)
Accuracy	98 (4.80)	98 (4.31)	95 (7.82)	95 (6.28)	98 (4.67)	94 (5.50)

Note: LG = low morphological knowledge group; HG = high morphological knowledge group.

An ANOVA on the mean RTs showed that the main effect of prime types was significant, F₁(2,76) = 53.31, p < 0.01, η² = 0.63; F₂(2,70) = 71.09, p < 0.01, η² = 0.51, the main effect of group was not significant, F₁(1,38) = 0.23, p = 0.64, F₂(1,35) = 0.04, p = 0.90, and the interaction between prime types and group was significant F₁(2,76) = 33.86, p < 0.01, η² = 0.29; F₂(2,70) = 7.44, p < 0.01, η² = 0.65.

To clarify the nature of the interaction, a paired-sample t-test was separately conducted for each group (t₁ and t₂ signify the participant analysis and item analysis). The differences in RTs of the three different prime types are given in Table 4. For the low morphological knowledge group, significant differences were found between derived prime and repetition prime, t₁(19) = 3.51, p < 0.01, d₁=0.83; t₂(35) = 5.29, p < 0.01, d₂=0.63, with the RTs in the repetition prime condition being significantly shorter than those of the derived prime condition. The difference between unrelated primes and repetition primes was significant, t₁(19) = 6.08, p < 0.01, d₁ = 0.43; t₂(35) = 6.98, p < 0.01, d₂=0.66, with the RTs in the repetition prime condition being significantly shorter than those of the unrelated primes. However, there were no differences between the unrelated prime condition and the derived prime condition, t₁(19) = 1.29, p = 0.21; t₂(35) = 2.15, p = 0.06. For the high morphological knowledge group, the priming effect between the unrelated prime condition and the repetition prime condition was significant, t₁(19) = 12.10, p < 0.01, d₁ = 0.77; t₂(35) = 7.44, p < 0.01, d₂ = 0.51, with the RTs in the repetition prime condition being significantly shorter than those of the unrelated primes. The priming effect between unrelated primes and derived primes was also significant, t₁(19) = 8.60, p < 0.01, d₁ = 0.29; t₂(35) = 7.67, p < 0.01, d₂ = 0.80, with RTs in the derived prime condition being significantly shorter than those of the unrelated prime condition. More importantly, the difference between derived and repetition primes was not significant in the participant analysis, t₁(19) = 1.37, p = 0.19; t₂(35) = 2.17, p = 0.04, d₂ = 0.36.

Table 4.

The differences in response times of three different prime types (standard deviations).

	LG	HG
Derived – repetition	64^** (81)	21 (68)
Unrelated – derived	16 (54)	127^** (66)
Unrelated – repetition	81^** (59)	148^** (55)

Note: ^*p<0.05;^**p<0.01; LG =group with low morphological knowledge; HG = group with high morphological knowledge.

An ANOVA on the mean accuracy scores showed that the main effect of prime types was significant, F₁(2,76) = 10.66, p < 0.01, η² = 0.22; F₂(2,70) = 6.59, p < 0.05, η² = 0.33, the main effect of group was not significant, F₁(1,38) = 1.11, p = 0.29, F₂(1,35) = 0.48, p = 0.49, and the interaction between prime types and group was not significant, F₁(2,76) = 0.04, p = 0.85, F₂(2,70) = 0.47, p = 0.50.

In sum, for the low morphological knowledge group, no response time difference was found between the derived and unrelated prime conditions; this means that there was no priming effect. For the high morphological knowledge group, a full priming effect was observed, since there was no difference in response times between the derived prime and repetition prime conditions, and the RTs in both conditions were significantly shorter than in the unrelated prime condition.

Form control primes

The mean RTs and accuracy of the form control primes are listed in Table 5.

Table 5.

Response times (RTs) (ms) and accuracy (%) on the form control primes (standard deviations).

	LG			HG
	Repetition	Form	Unrelated	Repetition	Form	Unrelated
RT	634 (175)	722 (139)	718 (102)	629 (98)	734 (109)	745 (130)
Accuracy	88 (8.12)	91 (10.91)	92 (9.70)	90 (7.68)	90 (8.15)	91 (13.92)

Note: LG =group with low morphological knowledge; HG = group with high morphological knowledge.

An ANOVA on the mean RTs showed that the main effect of prime type was significant, F₁(2,76) = 25.97, p < 0.01, η² = 0.41, F₂(2,70) = 3.24, p = 0.08, η² = 0.21. The main effect of group was not significant, F₁(1,38) = 0.11, p = 0.75, F₂(1,35) = 0.16, p = 0.68, and the interaction between prime type and group was also not significant, F₁(2,76) = 0.35, p = 0.56, F₂(2,70) = 0.42, p = 0.52. The significant main effect of prime type was driven by the fact that the RTs in the repetition prime condition were significantly shorter than those in the form prime condition and the unrelated prime condition, while there was no significant difference between the form prime condition and unrelated prime condition. This pattern was found in both groups.

The ANOVA on accuracy showed that there were no significant effects, including the main effect of prime type, F₁(2,76) = 0.26, p = 0.61, F₂(2,70) = 0.34, p = 0.56, the main effect of group, F₁(1,38) = 0.02, p = 0.90, F₂(1,35) = 0.45, p = 0.50, and also the interaction between prime type and group, F₁(2,76) = 0.10, p = 0.76, F₂(2,70) = 0.04, p = 0.84.

In sum, no significant difference was found between the form condition and the unrelated prime condition, and the RTs of both conditions were significantly longer than those of the repetition prime condition. These results indicate that priming effects in the derived prime condition could not be reduced to just the formal overlap between primes and targets.

In Experiment 1, a morphologically related priming effect was only observed in the high morphological knowledge group. As pointed out by Pinker (1999), in morphological processing, the morphologically complex words were decomposed into subcomponents as morphemes, which can facilitate the recognition of the target that completely overlaps with the morphemes. If the size of the derived priming effect, caused by complete decomposition of the derived words, is equivalent to the repetition priming effect, a full priming effect has been obtained (Pinker, 1999; Silva & Clahsen, 2008). The full priming effect in the high morphological knowledge group indicated that they do apply rule-governed decompositions in L2 morphological processing.

Experiment 2

In Experiment 2, we use a self-paced sentence reading paradigm to investigate the morphological knowledge on processing of L2 derivational words.

Method

Participants

Forty participants who completed Experiment 1 also completed Experiment 2. The time interval of the two experiments was at least four months in order to reduce the possible impact of Experiment 1 on Experiment 2.

Materials and design

Thirty derivational words selected from Experiment 1 and 30 monomorphemic nouns that matched with these derivational words in word length, frequency, and familiarity were used in the self-paced reading task. Each word was embedded in a plausible and syntactically simple sentence. Therefore, 60 experimental sentences, along with 60 filler sentences, and 5 practice sentences, make the number of sentences 125 in total. Each sentence contained 8 words. Most of derivational words and corresponding monomorphemic words were located in the fifth position in each sentence. The examples are as follows:

The/president/gave/ a / statement / to /the/public. (Sentence with a derivational word in the fifth position.)

This/method/has/ the / advantage / of /saving/energy. (Sentence with a monomorphemic word in the fifth position.)

Jane/loves/the/small/restaurant/in/the/hotel. (Filler sentence.)

The bold and italicized words in the above sentences were marked as segments 1, 2, and 3. Segment 2 was the derivational word or a monomorphemic word and was analyzed as the key segment of interest.

The familiarity of the derivational word or monomorphemic words was rated on a 5-point scale (1 = very unfamiliar, 5 = quite familiar) by a separate group of 44 college students from the same background as the participants. The frequencies were calculated based on the frequency dictionary of Brysbaert and New (2009). There were no significant differences in word length (8.67 vs. 8.61), familiarity rating (4.67 vs. 4.59), and word frequency (10.37 vs. 9.87) between these two types of words (all ps > 0.05).

To match the sentence context constraints and semantic plausibility between these two types of experimental sentences, the same 44 participants were asked to rate the aspects of all 60 sentences. The proportion of the raters who filled in the blank of the sentences with the target words was taken as an indicator of sentence context constraints. The semantic plausibility of the sentences was rated on a 5-point scale (1 = very implausible, 5 = plausible). No significant differences were found in the sentence constraints (0.25 vs. 0.21) or semantic plausibility (4.79 vs. 4.80) between the two types of sentences (ps > 0.05).

Procedure

Sentences were presented on a computer using E-prime software version 1.1. A self-paced sentence reading task was employed. Participants read a sentence by pressing the spacebar to advance word by word. As each new word appeared, the preceding word disappeared. After the final word of each trial, a true/false comprehension question appeared. Participants answered by pressing “J” to respond “true” or “F” to respond “false.” Half of the answers were true, and the other half were false. After answering the question, a star appeared again in the center of the screen indicating the beginning of the next trial.

The amount of time the participants spent on each word and their accuracy on the comprehension questions were recorded. To be specific, reading time was measured from the onset to the disappearance of each word. Prior to the experiment, five practice trials were provided for each participant.

Results and discussion

Separate ANOVAs were performed by participants (F₁) and by items (F₂) on the mean RTs in the regions of interest and on the mean comprehension accuracy. Analyses of mean RTs included only those trials in which the comprehension questions were answered correctly. RTs beyond three SDs of the grand mean were excluded (8.73%). Mean RTs in each region of interest are reported in Table 6.

Table 6.

Mean response times (ms) in three regions between high and low knowledge groups (standard deviations).

Word type	LG			HG
Word type	Segment 1	Segment 2	Segment 3	Segment 1	Segment 2	Segment 3
Derivational	601 (99)	997 (375)	517 (63)	567 (109)	1099 (413)	498 (98)
Monomorphemic	582 (102)	1029 (384)	518 (85)	596 (122)	1005 (454)	484 (107)
Difference	19	32	−1	−29	94*	14

Note: ^*p<0.05;^**p<0.01; LG =group with low morphological knowledge; HG = group with high morphological knowledge.

For each region of interest (segment 1, segment 2, segment 3), we performed separate two-way ANOVAs including group (high morphological knowledge, low morphological knowledge) and word type (derivational, monomorphemic). These results are reported in Table 7.

Table 7.

Results of analyses of variance at three critical regions.

Region	Source of variance	By participant		By item
Region	Source of variance	F₁(1,38)	p	F₂(1,58)	p
S1	Word type	2.62	0.11	2.01	0.13
	Group	0.78	0.34	0.56	0.46
	Group×word type	0.33	0.55	0.10	0.78
S2	Word type	1.23	0.27	3.55	0.06
	Group	0.10	0.76	1.06	0.31
	Group×word type	4.94*	<0.05	5.82**	<0.01
S3	Word type	0.23	0.64	0.89	0.35
	Group	1.13	0.30	3.11	0.06
	Group×word type	2.96	0.08	7.08**	<0.01

Note: ^*p < 0.05; ^**p < 0.01; S = segment.

The only significant interaction both by subjects and by items between word type and group was found in segment 2 (F₁ (1,38) = 4.94, p < 0.05, η² = 0.18; F₂ (1,58) = 5.82, p < 0.01, η² = 0.22). To clarify the nature of the interaction, paired-sample t-tests were conducted separately for the high morphological knowledge group and for the low morphological knowledge group. The results showed that the mean RTs in the derivational word condition were significantly longer than those in the monomorphemic word condition for the high morphological knowledge group (t₁ (19) = 2.17, p < 0.05, d₁ = 0.64; t₂ (29) = 2.38, p < 0.05, d₂ = 0.78), but no such differences were found for the low morphological knowledge group (t₁ (19) = -1.07, p = 0.30; t₂ (29) = -0.17, p = 0.86).

All the participants achieved at least 90% accuracy on the comprehension questions, indicating that they carefully read the sentences for comprehension. The average comprehension accuracy of the high morphological knowledge group was 96% (SD = 0.02) in the derivational words condition and 94% (SD = 0.03) in the monomorphemic words conditions. In the low morphological knowledge group, accuracy was 94% (SD = 0.02) in the derivational words condition and 94% (SD = 0.04) in the monomorphemic words conditions. An ANOVA on the accuracy data showed no significant differences by group or condition (ps > 0.1), indicating that both groups comprehended all the sentences equally well.

The results of Experiment 2 showed that for the high morphological knowledge group, the time they spent on reading the derivational words in the sentence was significantly longer than that of the monomorphemic words. This processing cost suggested that the high morphological knowledge group applied the rule-based decompositions to process derivational words. There were no differences in RTs of reading segment 1 between these two groups, indicating that the observed differences in segment 2 were not due to any wrap-up effect of segment 1.

General Discussion

The present study investigated the relationship between the morphological knowledge and L2 online processing of derivational words. The findings from these two experiments consistently showed that participants with high morphological knowledge are sensitive to morphological structure. These results suggest that L2 learners can and do decompose morphologically complex derivational words during both single-word and sentence-level reading, as long as the L2 learners have reached a higher level of morphological knowledge. The study provided direct evidence that morphological knowledge plays an important role during automatic L2 derivational decomposition processing and further deepens our understanding of the relationship between morphological knowledge and L2 online processing of derivational words.

The performance of the group with high morphological knowledge suggested that for the L2 learners, it is possible to apply automatic rule-based decomposition. The findings were in accordance with the results of Diependaele et al. (2011). In their study, the full priming effect was observed in the L2 learners, indicating that L2 learners can engage in rule-governed decompositions. Therefore, with the accumulation of their morphological knowledge, L2 learners can gradually realize automatic decompositions. However, when L2 learners have a low level of morphological knowledge, they may have to rely on the whole-word processing. So, morphological knowledge is a very important variable in helping L2 learners acquire the automatic morphological decomposition process. The present finding was in accordance with the result of our study that used event-related potentials to explore the effect of morphological knowledge on L2 derivational processing (Deng, Shi, Dunlap, Bi, & Chen, 2016).

Why does morphological knowledge influence the way L2 learners’ process derivational words? In English, derivational processing seems to be associated with the linguistic properties of the derivational words. Derivations create new words and are context-independent (Bickel & Nichols, 2007; Bozic et al., 2013). The derivational suffixes can change the meaning of the stems they attach to, which can help to access and decompose the derived words. As L2 learners accumulate morphological knowledge through the cumulative exposure and usage, they may strengthen their understanding of the word formation rules. As we have stated in the introduction, L2 learners may tend to decompose the morphological words into components when they encounter them. Then, the more words they read, the higher their ability to decompose words becomes. For the group with high morphological knowledge, as they accumulate more morphological knowledge, they may greatly improve their ability to decompose these words. Therefore, the group with high morphological knowledge may be able to decompose the morphologically complex words online.

Certainly, not all words in English follow the pattern of stem plus suffix. Derivation can change the meaning of the words with the affixes as the semantic cues, and may make the online derivational processing relatively easy. But this may not be the same case for inflected words. Inflections are determined by the grammatical context where they are expressed and do not change the semantic meaning of the stems. Inflected words with the affixes are taken to be a surface variant of the underlying stem but with different grammatical information (Bickel & Nichols, 2007; Bozic et al., 2013). Therefore, correctly applying the rules of inflections requires the acquisition of grammatical knowledge such as tense and subject–verb agreement, which may lead to difficult decomposition by L2 learners. Therefore, even if participants have high morphological knowledge, it may not be sufficient for them to apply the combinatorial operations of the inflected words. This may partly give an explanation for the inconsistency with the findings of Silva and Clahsen (2008) that L2 learners rely more heavily on whole-word processing for inflected words and decompose derivational words in a less automatic way. As the present study focused on the processing of derivational words, we concede that we have not directly compared the processing mechanism between derivational morphology and inflectional morphology in L2. Further studies are required to test these issues.

Clahsen and colleagues (Neubauer & Clahsen, 2009; Silva & Clahsen, 2008) have argued that it is difficult for adult L2 learners to establish native-like rule-based decomposition. They have to rely much more on whole-form storage. With an increase in proficiency, they can gradually acquire rule-governed decompositions but still fail to attain automatic or native-like proficiency. The main difference between our results and theirs may lie in the variable investigated, as morphological knowledge was not considered in their studies. According to our results, L2 learners with high morphological knowledge can employ rule-based decomposing mechanisms for processing derivational words. Heyer and Clahsen (2015) found that priming effects for derived words in non-native speakers are not morphological in nature but are heavily influenced by surface form. In the present study, we had a form control condition and found no significant form priming effects. This result indicates that priming effects for derived words could not be reduced to just the form overlap between primes and targets. Some previous studies have also indicated priming effects for morphologically related primes, which could not be attributed to the form or semantic overlap between primes and targets (Boudelaa & Marslen-Wilson, 2005; Frost, Kugler, Deutsch, & Forster, 2005). Therefore, we do not deny the possibility that the priming effects for L2 learners may be influenced by surface form, but the derived priming effects in the present study cannot be attributed solely to the form overlap between primes and targets.

The declarative/procedural model proposed by Ullman (Ullman, 2001, 2004, 2005) highlighted that, in L1, processing aspects of morphologically complex words depends largely upon procedural memory, and native speakers can easily decompose morphologically complex words into the stems and suffixes. However, in a later-acquired L2, it is argued that the procedural memory system may be impaired in learning or computing aspects of morphologically complex words. With insufficient input, late L2 learners may have to depend largely on declarative memory in processing L2. According to Ullman ( 2004, 2005), the shift of dependence from declarative to procedural memory is expected to be a function of language proficiency or exposure; then the combinatorial operations similar to native speakers can be gradually realized. The results from our study suggest that, in addition to proficiency, the morphological knowledge L2 learners have acquired is also an important variable in combinatorial processing.

As the present study did not measure or manipulate vocabulary knowledge as a variable, we conceded this would be a limitation. Our data cannot address any hypothesized relationship between vocabulary knowledge and morphological knowledge. Future studies could include related vocabulary tests to further explore the relationship between vocabulary, morphological knowledge, and online morphological processing.

In summary, the present study provides behavioral evidence that L2 morphological knowledge plays an important role on L2 derivational processing. Learners with high morphological knowledge were able to use a decomposing mechanism, whereas learners with low morphological knowledge depended much more on a whole-word processing mechanism of L2 derivational processing. These results suggest that improvements in morphological knowledge can facilitate the rule-based processing of L2 derivational words.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by the Open Fund of State Key Lab of Cognitive Neuroscience and Learning (CNLYB1309) to Baoguo Chen.

Notes

Author biographies

Taiping Deng is a PhD student from School of Psychology at Beijing Normal University and a university lecturer at the Zhejiang University of Finance and Economics. Her research focuses on morphological processing and syntactic acquisition of L2 learners.

Jiawei Shi is a master degree’s student from School of Psychology at Beijing Normal University. Her research focuses on morphological processing of L2 learners.

Hongyan Bi is a professor from Institute of Psychology, Chinese Academy of Sciences. Her research focuses on orthographic and phonological processing of developmental dyslexia.

Susan Dunlap is a postdoctoral research fellow at the University of Texas Health Science Center at Houston. Her research focuses on the kindergarten teacher classroom behavior, phonological awareness; second language acquisition.

Baoguo Chen is a professor from School of Psychology at Beijing Normal University. His research focuses on lexical, syntactic and semantic processing of L2 learners, language switching.

References

Bickel

Nichols

(2007). Inflectional morphology. In Shopen

(Ed.), Language typology and syntactic description, Volume 3: grammatical categories and the lexicon (pp. 169–240). Cambridge, UK: Cambridge University Press.

Boudelaa

Marslen-Wilson

W. D.

(2005). Discontinuous morphology in time: incremental masked priming in Arabic. Language and Cognitive Processes, 20(1/2), 207–260.

Bowers

P. N.

Kirby

J. R.

Deacon

S. H.

(2010). The effects of morphological instruction on literacy skills: a systematic review of the literature. Review of Educational Research, 80(2), 144–179.

Bozic

Tyler

L. K.

Wingfield

Marslen-Wilson

W. D.

(2013). Neurobiological systems for lexical representation and analysis in English. Journal of Cognitive Neuroscience, 25(10), 1678–1691.

Brysbaert

New

(2009). Moving beyond Kucera and Francis: a critical evaluation of current word frequency norms and the introduction of a new and improved word frequency measure for American English. Behavior Research Methods, 41(4), 977–990.

Carlisle

J. F.

(2000). Awareness of the structure and meaning of morphologically complex words: impact on reading. Reading and Writing, 12(3), 169–190.

Carlisle

J. F.

Nomanbhoy

D. M.

(1993). Phonological and morphological awareness in first graders. Applied Psycholinguistics, 14(02), 177–195.

Clahsen

(1999). Lexical entries and rules of language: a multidisciplinary study of German inflection. Behavioral and Brain Sciences, 22(6), 991–1013.

Clahsen

Felser

Neubauer

Sato

Silva

(2010). Morphological structure in native and non-native language processing. Language Learning, 60(1), 21–43.

10.

Clahsen

Sonnenstuhl

Blevins

J.P.

(2003). Derivational morphology in the German mental lexicon: a dual mechanism account. In Baayen

Schreuder

(Eds.), Morphological structure in language processing (pp. 125–155). Berlin, Germany: Mouton de Gruyter.

11.

Clark

M. D.

Gilbert

Anderson

M. L.

(2011). Morphological knowledge and decoding skills of deaf readers. Psychology, 2(2), 109–116.

12.

Deacon

S. H.

Benere

Pasquarella

(2013). Reciprocal relationship: children’s morphological awareness and their reading accuracy across grades 2 to 3. Developmental Psychology, 49(6), 1113–1126.

13.

Deng

T. P.

Shi

J. W.

Dunlap

H. Y.

Chen

B. G.

(2016). Morphological knowledge affects processing of L2 derivational morphology: An event-related potential study. Journal of Neurolinguistics, 37, 47–57.

14.

Diependaele

Duñabeitia

J. A.

Morris

Keuleers

(2011) Fast morphological effects in first and second language word recognition. Journal of Memory and Language, 64(4), 344–358.

15.

Feldman

L. B.

Prostko

(2002). Graded aspects of morphological processing: task and processing time. Brain and Language, 81(1), 12–27.

16.

Frost

Kugler

Deutsch

Forster

K. I.

(2005). Orthographic structure versus morphological structure: principles of lexical organization in a given language. Journal of Experimental Psychology: Learning, Memory and Cognition, 31(6), 1293–1326.

17.

Goodwin

Huggins

Carlo

August

Calderon

(2013). Minding morphology: how morphological awareness relates to reading for English language learners. Reading and Writing, 26(9), 1387–1415.

18.

Heyer

Clahsen

(2015). Late bilinguals see a scan in scanner AND in scandal: dissecting formal overlap from morphological priming in the processing of derived words. Bilingualism: Language and Cognition, 18(03), 543–550.

19.

Hoover

W. A.

Gough

P. B.

(1990). The simple view of reading. Reading and Writing, 2(2), 127–160.

20.

Kieffer

Lesaux

(2012). Direct and indirect roles of morphological awareness in the English reading comprehension of native English, Spanish, Filipino, and Vietnamese speakers. Language Learning, 62(4), 1170–1204.

21.

Kirkici

(2005). Words and rules in L2 processing: an analysis of the dual-mechanism model. Unpublished manuscript. Middle East Technical University, Ankara. Retrieved from https://etd.lib.metu.edu.tr/upload/12605980/index.pdf

22.

Kraut

(2015). The relationship between morphological awareness and morphological decomposition among English language learners. Reading and Writing, 28(6), 873–890.

23.

Kuo

Anderson

R. C.

(2006). Morphological awareness and learning to read: a cross-language perspective. Educational Psychologist, 41(3), 161–180.

24.

Lehtonen

Laine

(2003). How word frequency affects morphological processing in monolinguals and bilinguals. Bilingualism: Language and Cognition, 6(3), 213–225.

25.

Liang

L. J.

Chen

B. G.

(2014). Processing morphologically complex words in second-language learners: the effect of proficiency. Acta Psychologica, 150, 69–79.

26.

Longworth

C. E.

Marslen-Wilson

W. D.

Randall

Tyler

L. K.

(2005). Getting to the meaning of the regular past tense: evidence from neuropsychology. Journal of Cognitive Neuroscience, 17(7), 1087–1097.

27.

Mahony

Singson

Mann

(2000). Reading ability and sensitivity to morphological relations. Reading and Writing, 12(3), 191–218.

28.

Marslen-Wilson

Tyler

L. K.

Waksler

Older

(1994). Morphology and meaning in the English mental lexicon. Psychological Review, 101(1), 3–33.

29.

Miozzo

(2003). On the processing of regular and irregular forms of verbs and nouns: evidence from neuropsychology. Cognition, 87(2), 101–127.

30.

Murrell

Morton

(1974). Word recognition and morphemic structures. Journal of Experimental Psychology, 102(6), 963–968.

31.

Nagy

W. E.

Berninger

V. W.

Abbott

R. D.

(2006). Contributions of morphology beyond phonology to literacy outcomes of upper elementary and middle-school students. Journal of Educational Psychology, 98(1), 134–147.

32.

Nagy

W. E.

Carlisle

J. F.

Goodwin

A. P.

(2014). Morphological knowledge and literacy acquisition. Journal of Learning Disabilities, 47(1), 3–12.

33.

Neubauer

Clahsen

(2009). Decomposition of inflected words in a second language. Studies in Second Language Acquisition, 31(03), 403–435.

34.

Ouellette

G. P.

(2006). What’s meaning got to do with it: the role of vocabulary in word reading and reading comprehension. Journal of Educational Psychology, 98(3), 554–566.

35.

Padeliadu

Antoniou

(2014). The relationship between reading comprehension, decoding, and ﬂuency in Greek: a cross-sectional study. Reading & Writing Quarterly: Overcoming Learning Difficulties, 30(1), 1–31.

36.

Paradis

(2004). A neurolinguistic theory of bilingualism. Amsterdam, The Netherlands: John Benjamins.

37.

Perfetti

C. A.

Hogaboam

(1975). Relationship between single word decoding and reading comprehension skill. Journal of Educational Psychology, 67(4), 461–469.

38.

Pinker

(1999). Words and rules: the ingredients of grammar. New York, NY: Basic Books.

39.

Pinker

Ullman

M. T.

(2002). The past and future of the past tense. Trends in Cognitive Sciences, 6(11), 456–463.

40.

Pliatsikas

Marinis

(2013). Processing of regular and irregular past tense morphology in highly proficient second language learners of English: a self-paced reading study. Applied Psycholinguistics, 34(05), 943–970.

41.

Pliatsikas

Johnstone

Marinis

(2014). fMRI evidence for the involvement of the procedural memory system in morphological processing of a second language. PloS ONE, 9(5), e97298. DOI: 10.1371/journal.pone.0097298.

42.

RAND Reading Study Group. (2002). Reading for understanding: toward an R&D program in reading comprehension. Santa Monica, CA: RAND Corporation.

43.

Silva

Clahsen

(2008). Morphologically complex words in L1 and L2 processing: evidence from masked priming experiments in English. Bilingualism: Language and Cognition, 11(2), 245–260.

44.

Singson

Mahony

Mann

(2000). The relation between reading ability and morphological skills: evidence from derivational suffixes. Reading and Writing: An Interdisciplinary Journal, 12(3), 219–252.

45.

Taft

(1979). Recognition of afﬁxed words and the word frequency effect. Memory & Cognition, 7(4), 263–272.

46.

Taft

Forster

(1975). Lexical storage and retrieval of preﬁxed words. Journal of Verbal Learning and Verbal Behavior, 14, 638–647.

47.

Tong

Deacon

S. H.

Kirby

J. R.

Cain

Parrila

(2011). Morphological awareness: a key to understanding poor reading comprehension in English. Journal of Educational Psychology, 103(3), 523–534.

48.

Ullman

M. T.

(2001). A neurocognitive perspective on language: the declarative/procedural model. Nature Reviews Neuroscience, 2(10), 717–726.

49.

Ullman

M. T.

(2004). Contributions of memory circuits to language: the declarative/ procedural model. Cognition, 92(1/2), 231–270.

50.

Ullman

M. T.

(2005). A cognitive neuroscience perspective on second language acquisition: The declarative/procedural model. In Sanz

(Ed.), Processing approaches to adult SLA: theory and practice (pp. 141–178). Washington, DC: Georgetown University Press.

51.

Ullman

M. T.

(2012). The declarative/procedural model. In Robinson

(Ed.), Routledge encyclopedia of second language acquisition. New York & London: Routledge.

52.

Ullman

M. T.

Pancheva

Love

Yee

Swinney

Hickok

(2005). Neural correlates of lexicon and grammar: evidence from the production, reading, and judgment of inflection in aphasia. Brain and Language, 93(2), 185–238.

53.

Wang

Cheng

Chen

(2006). Contribution of morphological awareness to Chinese–English biliteracy acquisition. Journal of Educational Psychology, 98(3), 542–553.