To adults who are proficient readers and writers, it seems quite obvious that spoken words are composed of strings of discrete phonological segments. However, for children the insight that words are composed of discrete segments comes gradually, often as a result of literacy instruction. To many young children, and even to some adolescents and adults, a word is a word, distinguishable from other words holistically in terms of pitch contours, rhythm, stress, increased duration, and strong–weak template; however, the segmental details of the word are obscured by the overall phonological shape of the word. Knowledge of the discrete sound segments of language, that is, phonological awareness, has been found to relate to poor performance on a number of language processing tasks, such as naming objects and pictures (Catts, 1989; Manis, Seidenberg, & Doi, 1999; Snowling, van Wagtendonk, & Stafford, 1988; Swan & Goswami, 1997; Wolf & Bowers, 1999), holding verbal information in working memory (Brady, 1991), perceiving fine speech distinctions (Adlard & Hazan, 1998; Brady, Poggie, & Rapala, 1989; de Weirdt, 1988; Manis et al., 1997; Reed, 1989), repeating nonsense words (Catts, 1986; Metsala, 1999), reading (Goswami, 2002; Stanovich & Siegel, 1994), and learning new words (de Jong, Seveke, & van Veen, 2000; Hu, 2003).

A distinction between phonological awareness and phonemic awareness is often made in the developmental literature, particularly to quantify the depth with which a child is aware of the sound structure of language (Adams, 1990; Goswami, 2000; Goswami & Bryant, 1990; Stahl & Murray, 1994). The participants in the current study were native speakers of Chinese, and this distinction between phonological and phonemic awareness may not be particularly relevant with these children. The dominant syllabic structure in Mandarin Chinese is consonant–vowel (CV). In most cases, the C and the V correspond to the onset and the rime of the word. (In English, with a CVC word, for example, the onset would be C and the rime VC.) The syllable structure of Chinese makes it difficult to determine whether a child's understanding of C and V is at the onset–rime level or at the phonemic level. In this article, discussion of children's phonological awareness is used to refer to the child's insight into the subsyllabic units of the spoken word, be it at the phonemic level or at the onset–rime level.

Because poor phonological awareness has been shown to relate to poorer performance on a variety of language tasks, the current study considered the influence of native phonological awareness on another domain of language learning, the learning of nonnative words, in other words, learning of foreign language words. Native phonological awareness (phonological awareness in first language [L1]) was examined in this study for several reasons. First, phonological awareness is basically transferable from one language to another language (e.g., Cisero & Royer, 1995; Comeau, Cormier, Grandmaison, & Lacroix, 1999; Durgunoglu, Nagy, & Hancin–Bhatt, 1993). If a child realizes that words in his or her native language lexicon can be divided into component sounds, the child can apply that awareness to foreign language. Because for most children phonological awareness develops initially in the L1, the initial learning of foreign language would be influenced by phonological awareness in the L1 and not the foreign language. If a child has little or no knowledge of foreign language, then there would not be phonological awareness specific to foreign language that would assist in learning. This situation may differ for children whose literacy instruction is provided in their second language (L2) or for children who are given sufficient exposure to the second language (Guron & Lundberg, 2003).

Second, although phonological awareness is essentially transferable from one language to another language, there are studies indicating that experience with a particular language affects how phonological representations are implicitly and explicitly structured (Caravolas & Bruck, 1993; Cheung, Chen, Lai, Wong, & Hills, 2001). Given that the purpose of this study is to evaluate the role of phonological awareness in the very early learning of novel nonnative words and given that the child participants differed considerably in their experiences with the nonnative language, a finding of the relationship between nonnative phonological awareness and nonnative vocabulary learning would be less interesting. The relation may be nothing more than the spurious effect of the common variance in experience with the nonnative language.

Third, we sought to delineate the relationship between native and nonnative language processing abilities. It is not surprising that researchers have reported that the overt or subtle native language learning differences of high school and college students with language learning disabilities affect learning of a foreign language (Chen, 1999; Ganschow, Sparks, & Javorsky, 1998; Ganschow, Sparks, Javorsky, Pohlman, & Bisho–Marbury, 1991; Simon, 2000; Sparks & Ganschow, 1993, 1995). Specifically, foreign language learners who have poor phonological skills are likely to experience immediate and significant difficulty when they try to learn a foreign language (Ganschow et al., 1991; Sparks, Ganschow, & Pohlman, 1989; see also Meschyan & Hernandez, 2002). The concomitant existence of native and foreign language difficulties among high school and college students, however, does not say much about how native language abilities affect foreign language learning in the first place. To be able to model the interplay between native and foreign language acquisition comprehensively, it is important to know which aspect(s) of native language processing influences which aspect(s) of foreign language learning. This relation must be evaluated in the absence of obvious language impairment in the L1. Given that phonological awareness has been identified as a locus of many language problems, it is logical to explore foreign language learning differences with respect to native phonological awareness abilities in populations with typical language learning abilities.

Theoretically, the ability to represent a spoken word segmentally should facilitate word learning in a nonnative language. Hearing and repeating unfamiliar sequences of nonnative sounding syllables and segments is a familiar task to those who try to learn the lexicon of a nonnative language. However, although it is not so difficult for learners to process the overall prosodic features of the new word, the subsyllabic segments are often elusive. To be able to access individual segments in a novel sequence of sounds (i.e., a new word) thus seems the first, and to many learners, the most frustrating step in acquiring nonnative vocabulary. Children who have more developed phonological awareness in their native language can be expected to have greater access to the segments of the nonnative language than children who have poorer phonological awareness.

Recent work by Hu (1999, 2003) provides some direct support for the importance of native phonological awareness in nonnative word learning. Phonological memory was found to be concurrently and predictively related to the word recall and pronunciation learning of a nonnative language initially tested at age 5, whereas the measures of phonological awareness were not. However, the phonological awareness measures emerged as significant predictors when the words were taught again 6 months later, after the variance due to the native reading ability, phonological memory, and initially tested nonnative word learning ability was controlled. These results provide preliminary evidence that phonological awareness may play a specific role in nonnative vocabulary acquisition.

There were three limitations in Hu's study (2003), however. First, the participants were young; most were prereaders when they were tested. These children demonstrated very limited phonological awareness at the subsyllabic level. The relationship between phonological awareness and nonnative word learning may not be replicated in older children with more mature and stable phonological awareness. Perhaps the acquisition of nonnative vocabulary depends on a certain minimum of insight into the internal components of a spoken word, beyond which individual differences in phonological awareness are less crucial. Second, the children's existing vocabulary knowledge was not measured. Even though test words in Hu (2003) were chosen from rather obscure nonnative words that children were unlikely to have encountered, it is plausible that the relationship between native phonological awareness and nonnative word learning was mediated more broadly by children's nonnative vocabulary knowledge. Under the pressure to discriminate between two phonological systems and to sort out the phoneme constancy of two languages, children who knew more nonnative words might have developed a deeper insight into the phonological structure of language, be it specific to the nonnative language or to language in general. This awareness, arising from experience with nonnative vocabulary, in turn, might have simultaneously facilitated the development of native phonological awareness and the nonnative word learning ability. Third, to succeed in the associative learning task of nonnative vocabulary employed in Hu (2003), a child had to be able to establish a stable phonological representation for the new word for subsequent production and had to be able to map the phonological representation onto the correct referent. Given that Hu (2003) did not differentiate these two processes, it remained unclear how poor phonological awareness can affect nonnative vocabulary learning.

The current study aimed to provide a more direct investigation of the relationship between native phonological awareness and nonnative word learning. The three questions that were left unanswered in Hu (2003) were addressed by examining the following three hypotheses. First, phonological awareness also plays a role in nonnative word learning among children whose phonological awareness has reached a more stable, mature stage. Second, the relation between native phonological awareness and nonnative vocabulary learning is unique and independent of the child's extant vocabulary knowledge of the nonnative language. Third, poor native phonological awareness affects the acquisition of new nonnative words via deficiency in the construction of new phonological forms rather than difficulty in mapping a phonological form onto a referent.

In this study, participants were third grade Chinese-speaking children, older than the 5-year-old preschoolers recruited in Hu (2003). Many studies have shown that the experience with reading an alphabetic orthography is crucial to the development and stabilization of phonological awareness (Foy & Mann, 2003; Goswami & Bryant, 1990; Morais, Cary, Alegria, & Bertelson, 1979; Morais, Castro, Scliar–Cabral, Kolinsky, & Content, 1987; Prakash, Rekha, Nigam, & Karanth, 1993). Experience with the reading of a logographic orthography has also been shown to be related to the development of phonological awareness (Cheung & Chen, 2004; Ho & Bryant, 1997; Ho, Law, & Ng, 2000; Hu, 2004; Hu & Catts, 1998; Siok & Fletcher, 2001). Reading instruction in Taiwan starts with an introduction of zhuyin fuhao, a set of phonetic symbols, which are printed alongside Chinese characters as an aid for pronouncing characters. With at least 3 years of experience in reading zhuyin fuhao and Chinese characters, children's phonological awareness was expected to have reached a fairly stable stage when they were tested. Moreover, by third grade it is likely that the phonological awareness tasks reliably tapped into children's varying skill levels and that floor effects would not be a confounding issue.

Factors that might complicate the interpretation of the relationship such as phonological memory and nonnative vocabulary knowledge were directly assessed and used as covariates in statistical analyses. Phonological memory was included because it was found to support vocabulary acquisition in L1 as well as in foreign languages (Avons, Wragg, Cupples, & Lovegrove, 1998; Cheung, 1996; Gathercole & Baddeley, 1990; Gathercole, Service, Hitch, Adams, & Martin, 1999; Service & Kohonen, 1995). However, studies also showed that such skills such as phonemic segmentation are crucial if new phonological forms are to be accurately encoded in phonological memory (Metsala, 1999; Snowling, Chiat, & Hulme, 1991).

Finally, in addition to the associative learning task of nonnative vocabulary, three additional associative learning tasks that allowed for an examination of the locus of the effect of phonological awareness on nonnative vocabulary learning were created. If phonological awareness influences nonnative vocabulary learning through its effect on the construction of new phonological forms, then children with poor phonological awareness should have difficulty learning to associate novel native names and novel nonnative names with novel referents, both of which involved the construction of new phonological forms. However, they should not have difficulty learning to associate familiar names or novel visual shapes with novel referents because these two tasks do not require the construction of new phonological forms. If, on the other hand, phonological awareness influences nonnative vocabulary learning through its effect on semantic mapping rather than through its effect on the construction of new phonological forms, then children with poor phonological awareness should have difficulty with all the four associative learning tasks.

METHODS

Participants

A group of 148 Chinese-speaking third graders from eight classes participated in the study over the course of third and fourth grade. Their mean age was 8 years, 10 months (8;10, range=8;2–9;3) when they were first tested. The fall testing ran from early November to late December and the spring testing ran from mid-April to early June. The participant selection tests were given in the fall of third grade and the control tasks in the spring of third grade. In the fall of fourth grade the nonnative name learning and native name learning tests were given and in the spring the familiar name learning and visual name learning tests were given. Permission for participation was obtained from the school and the parents. The children were administered a set of participant selection tests in groups of 15–20. The selection tests were a native sound oddity task, a native sound deletion task, and a zhuyin fuhao spelling task. The children were selected and categorized according to their performances on these three native phonological awareness tasks. The score on each task was converted to a z score, and then each child's scores on the three tasks were averaged. Children with z scores below the 25th percentile were classified as having low phonological awareness (Low PA), and those with z scores above the 75th percentile were classified as having high phonological awareness (High PA). Forty-one children were categorized as Low PA and 38 were categorized as High PA. Three children in the Low-PA group and 1 child in the High-PA group left the school during the study, leaving 38 in the Low-PA group and 37 in the High-PA group completing all the tasks. One child in the Low-PA group was excluded from the analyses because of his particularly low verbal IQ as measured by the Chinese version of the Peabody Picture Vocabulary Test—Revised (PPVT-R; Lu & Liu, 1998). Thus, data are reported only for the remaining 74 children (37 in the Low-PA group and 37 in the High-PA group), who completed all the tasks. The Low-PA group did not represent a group of children who were slow in general verbal processing. In this study, the two groups of children did not differ on the PPVT-R, a test of verbal IQ, t (72)=1.9, p > .05, or digit span, t (72)=1.6, p > .05 (Table 1). Tests in this study were given by two graduate students who were blind to the purposes of the present study and the grouping status of the children.

Note that in this study the participation selection tasks (i.e., phonological awareness) were administered a year before the associative learning task was given. This arrangement was made for two reasons. The first was purely pragmatic. Time constraints did not allow the administration of the various measures within one semester. In the second, it was assumed that the children would not change phonological awareness group during the intervening year because studies have revealed that deficits in phonological awareness persist from early childhood to adulthood (Bruck, 1992; Pennington, van Orden, Smith, Green, & Haith, 1990) and phonological awareness abilities are quite stable across the early elementary years (Torgesen, Wagner, & Rashotte, 1994; Wagner, Torgesen, Laughton, Simmons, & Rashotte, 1993; Wagner, Torgesen, & Rashotte, 1994). An extended period between the phonological awareness tasks and the outcome measures also allowed us to investigate the long-term effect of poor phonological awareness on the associative learning of foreign words. Indeed, the two groups of children identified in the fall of the third grade were given two measures of L1 phonological awareness individually in the spring of the third grade to make sure that the initial grouping procedure was valid. Group performances differed significantly in these two individually administered measures of phonological awareness (initial consonant deletion and initial consonant substitution), indicating the initial grouping status was stable over time.

In Taiwan, English education officially begins at different ages in different counties. In Taipei where the children in the study resided, English education began in the third grade. (Currently, as a result of policy change, English education starts in the first grade in Taipei.) The children in this study participated in a 40-min English class in school each week over the course of the 2 years of this study; in addition, many children went to private after-school English programs. Thus, children widely differed in terms of their experience with the English language.

Participant selection tasks

Native sound oddity

This test was similar to the oddity task adopted in Hu and Catts (1998) and Siok and Fletcher (2001), but it was adapted as a group test with a forced choice procedure with a chance level of 33%. In this task, the children heard 16 sets of three one-syllable words (e.g., bi [pen], ban [plank], gou [dog]). In half of the sets, one word in each set of three words did not start with the same consonant as the other two (in this case gou). In the other half, the odd word did not rhyme with the others. The experimenter orally presented the three test words in each set three times, each time pointing to the corresponding numbers 1, 2, and 3 on an enlarged version of the answer sheet. The child's task was to circle the number on an answer sheet that represented the odd word. The position of the odd word was varied randomly in positions 1–3 across trials (max=16).

Native sound deletion

The experimenter read a disyllabic word twice (e.g., shou-pa [handkerchief]) and asked the children what was left if the initial consonant of the word was deleted. The initial consonant was read as its sound value rather than its corresponding zhuyin fuhao name. The experimenter read the two choices, pointing to the corresponding numbers 1 and 2 on an enlarged version of the answer sheet. The child was asked to circle on an answer sheet the number that represented the correct choice. There were 10 trials (max=10).

Zhuyin fuhao spelling

In this task, the child was required to spell in zhuyin fuhao a wordlike stimulus read by the experimenter. This task was included because of its close association with phonological awareness among Chinese speakers (Hu & Catts, 1998; Read, Zhang, Nie, & Ding, 1986; Siok & Fletcher, 2001). All the wordlike stimuli were monosyllabic: 10 were high-frequency words, 13 low-frequency words, and 10 pseudowords (max=33).

Control tasks

Phonological memory

In this task, children repeated bisyllabic Chinese-sounding pseudowords. Thirty-six syllables were constructed by combining one of six consonants b, d, k, g, zh, and sh with one of six rimes u, a, ai, au, an, and ang. The consonants and vowels could be freely combined with each other without violating the phonological rules of Mandarin. Two syllables were combined into a bisyllabic pseudoword, and each pseudoword was assigned a tonal structure. Thus, a total of 36 syllables were combined to form 18 words. None of the successive syllables in a pseudoword were assigned Tone 3. In Mandarin, when two syllables with Tone 3 occur in succession within a word, the tone of the first syllable changes to Tone 2. Because the stimuli used in this task were pseudowords, the change of tone would make it difficult to document the child's response. There was a total of six trials with three pseudowords in each trial (18 different words). In each trial, the child listened to the bisyllabic pseudowords spoken by the experimenter. The child was instructed to repeat the pseudowords immediately in the order that they had been presented. The child's responses were audiorecorded and later were transcribed by two college graduates. The child's performance was scored in terms of the number of syllables recalled correctly in the correct position (max=6 trials × 6 syllables per trial=36). The test–retest reliability of the task obtained for first graders in an earlier study (Hu & Catts, 1998) was .85.

Digit span

The experimenter read a sequence of digits (from 4 to 12 digits) to the child at the rate of 1 digit per second, and the child was asked to repeat the sequence of digits in the same order. There were two consecutive trials at each sequence length. The task was discontinued when the child failed to repeat two trials of the same length consecutively. The scoring was based on the maximum number of digits that the child could recall.

Native vocabulary

The Chinese version of the PPVT-R (Lu & Liu, 1998) was used to measure the child's receptive vocabulary of the native language. The Chinese PPVT-R consists of a series of 175 plates, each containing four line drawings of objects or actions. For each plate, the experimenter provided a stimulus word orally. The child was asked to respond by pointing to the line drawing on the plate that best illustrated the meaning of the stimulus word. Test administration proceeded until the test error criterion was reached.

Nonnative vocabulary

The nonnative vocabulary task was developed by Lee (2000). The task consisted of a series of 40 plates, each containing four line drawings of objects or actions. For each plate, the experimenter provided an English word orally. The child was required to respond by circling the line drawing on the plate that best illustrated the meaning of the stimulus word (max=40). The reliability of the task (Spearman–Brown) is .90.

Associative learning tasks

There were four multitrial associative learning tasks: (a) a nonnative name learning task, (b) a native name learning task, (c) a familiar name learning task, and (d) a visual name learning task. The first two tasks required the child to verbally produce three pseudowords as names for three cartoon figures, thus involving the construction and production of new phonological forms. The third task also involved reproduction of the given names, but the names were already familiar to the child. The fourth task required the child to learn to associate three unfamiliar visual shapes as names for three cartoon figures. Given that the purpose of the study was to determine whether children with poorer phonological awareness had difficulty in learning nonnative names due to their particular difficulty in constructing new phonological forms rather than in mapping a set of forms onto a set of referents, we made the four tasks comparable in terms of the demands in mapping (i.e., mapping three forms onto three cartoon figures) but different in the forms required to be learned.

A pool of 12 cartoon figures was created and judged by an independent group of third graders as novel figures with which no known names were associated. The 12 figures were then randomly divided into four sets of three figures so that a unique set of figures could be used in each learning task. In each task, there were three names to be learned; the three names were paired with three cartoon figures. The pairing of the novel figures and names for each task was systematically varied so that the names to be learned in each task had an equal chance of being paired with the four sets of cartoon figures. By varying the pairing in this way, there was a control for any inherent factors (e.g., the distinctiveness of a particular cartoon, the salience of a particular pairing) that might affect children's associative learning. Within each task, the order of the three names to be learned varied from trial to trial in the same fixed way for all children.

Nonnative name learning

Each child was individually taught three new nonnative names with the accompanying pictures of the referents. The nonnative names were derived from the three English names Patsy, Roger, and Ellis by interchanging the syllables across the names. Two sets of three novel English names were created. One included Rotsy, Elger, and Palis; the other included Rolis, Pager, and Eltsy. The phonological structures of these names are different from Chinese and thus, they would not be treated as unfamiliar Chinese new names. To ensure that Chinese-speaking children did not have difficulties in articulating these names, the names were first tested on a different group of Chinese third-graders (N=10) by eliciting immediate repetition of the names. These children did not demonstrate obvious difficulties in articulating these names. Half of the children learned one set of the English names and the other half learned the other set. For each of the three names, the examiner showed the child a picture of the referent and labeled the picture with its name. The child was asked to repeat the name. No feedback was given as to the correctness of the child's pronunciation, so that every child would receive an equivalent amount of instruction (i.e., input). Nonetheless, the accuracy in immediate repetition was recorded. When the child had been given the opportunity to repeat the names of all three pictures, the three pictures were shown again in a different order and the child was asked again to name the pictures without an immediately preceding model. The accuracy of the child's production attempts to name the pictures was scored and audiorecorded for later transcription by a college student who was blind to the purposes of the study. The interscorer reliability was .96 for 20% of the sample. Thus, a single trial consisted of the adult saying each of the three names, the child attempting to repeat each name, and the child being shown the pictures a second time and asked to produce the name without an immediately preceding model. A maximum of 10 trials was administered; the trials were discontinued if the child named all three pictures correctly in 2 consecutive trials. A correct name was defined as a production with no obvious phoneme replacements, omissions, or additions (max=30). The children did not show difficulty in articulating these foreign names as evidenced by their 98% accuracy rate in the immediate repetition of the test stimuli. Group differences in immediate repetition were not significant, t (72)=1.5, p > .05. The reliability of the task (Spearman–Brown) is .93.

Native name learning

The task was similar to the nonnative name learning task, except that the targets were the three Chinese names Pi-su, Zhai-le, and Li-cao, which were created according to the phonotactic and prosodic constraints of the Chinese language. Like the nonnative name learning task, a maximum of 10 trials was administered; the trials were discontinued if the child named all three pictures correctly in 2 consecutive trials. A correct name was defined as a production with no obvious phoneme replacements, omissions, or additions (max=30). The reliability of the task (Spearman–Brown) is .90.

Familiar name learning

In this task, the child learned to associate familiar native names with unfamiliar cartoon figures. The task was included to test whether the two groups of children differed in their ability to attach a familiar linguistic label to an unfamiliar referent. The words to be learned were three familiar fruit names: ping-guo (apple), li-zi (pear), and liu-ding (orange). The task was similar to the nonnative name learning task. A maximum of 10 trials was administered. The task was discontinued after the 10th trial or if the child named the three cartoon figures correctly in 2 consecutive trials (max=30). The reliability of the task (Spearman–Brown) is .86.

Visual name learning

The visual name learning task was similar to the aforementioned associative learning tasks, but this task involved no verbal learning. The task evaluated whether the two groups of children differed in associative learning abilities when the learning involved no verbal materials. The child was presented with three unfamiliar cartoon figures, each accompanied by a doodle drawing. Neither the cartoon figures nor the doodle drawings were readily nameable in the child's native language. The child was told that the doodle drawings were the graphs of a strange language, which represented the names of the cartoon figures. The child had to learn to associate the cartoon figure with its corresponding graphic name. After the initial presentation of the pairing of the cartoon figures and the doodle drawings, the child was presented again with the same set of cartoon figures though in a different order and with the accompanying doodle drawings. The child was then given three cards, each with one of the doodle drawings on it. The child had to match each of the doodle drawings with its corresponding cartoon figure. The number of correct matches was scored. Thus, a single trial involved the presentation of the pairing of the cartoon figures and the doodle drawings (the presentation phase), the presentation of the cartoon figures again and the child's being asked to match the cartoon figures and the doodle drawings (the test phase). The task stopped after the 10th trial, or if the child correctly paired drawings to figures in 2 consecutive trials (max=30). The reliability of the task (Spearman–Brown) is .88.

RESULTS

Table 1 presents statistics about participant characteristics. The two groups of children differed on the three phonological awareness measures, t (72)=9.7, p<.001, for native sound oddity; t (72)=5.9, p<.001, for native sound deletion; t (72)=8.9, p<.001, for zhuyin fuhao spelling, validating the grouping procedure of the present study. As expected, the two groups of children differed in phonological memory, t (72)=2.5, p<.05, and nonnative vocabulary, t (72)=4.5, p<.001. Individual differences in these two abilities were controlled statistically in the following analyses. In spite of differences in phonological awareness, phonological memory, and nonnative vocabulary knowledge, children in the Low-PA group and in the High-PA group did not differ in age, native receptive vocabulary, or digit span.

In order to guard against Type I errors due to multiple univariate testing, a repeated-measures multiple analysis of covariance (MANCOVA) with group status as a fixed factor, associative learning task as a repeating factor, and memory as a covariate was conducted to evaluate whether children with High PA consistently performed better than children with Low PA across the four associative learning tasks. Memory was included in the model because the four tasks all involved memory learning. Vocabulary knowledge was not included as a covariate in the overall MANCOVA because not all of the four tasks involved the knowledge of the native or nonnative vocabulary. Follow-up analyses were conducted using separate univariate analyses of variance (ANOVAs) and ANCOVAs for the four associative learning tasks to evaluate further differences between groups within each associative learning. Vocabulary knowledge was included as a covariate in the relevant individual ANCOVA model.

Given that the groups were extreme (below the 25th and above the 75th percentiles), it is important to check that the covariates had the same relation to the dependent variables. An informal examination of the scatterplots, separately for the four associative learning tasks against each of the covariates (i.e., memory or vocabulary knowledge), indicated parallel linear bands across the graphs with no obvious departures from linearity. We also tested whether there were interaction effects between group and the covariates for each of the associative learning tasks. An interaction effect would indicate that the group regression lines had different slopes. However, no significant interaction effects were identified (all ps > .05).

The results of the MANCOVA on the two groups (High PA vs. Low PA) and the four task variables (nonnative name, native name, familiar name, and visual name) yielded a significant main effect of group status, F (1, 71)=14.7, p<.001, and a main effect of task, Wilks' lambda, F (3, 69)=16.12, p<.001. There was also a significant interaction effect between group status and task, Wilks' lambda, F (3, 69)=3.0, p<.05. As displayed in Table 2, individual ANOVAs revealed that children with High PA performed significantly better than children with Low PA in nonnative name learning, F (1, 72)=17.5, p<.001, and native name learning, F (1, 72)=8.6, p<.01. However, there was no evidence that children with High PA performed better than children with Low PA in familiar name learning, F (1, 72)=3.1, p > .05, and visual name learning, F (1, 72)=3.5, p > .05. In addition, group differences in nonnative learning remained significant after controlling for the variance due to phonological memory, F (1, 71)=12.0, p<.001, for the variance due to nonnative vocabulary, F (1, 71)=8.7, p<.01, or for the variance due to the combined effect of phonological memory and nonnative vocabulary, F (1, 70)=5.4, p<.05. Group differences in novel native name learning remained significant after controlling for the variance due to phonological memory, F (1, 71)=4.6, p<.05, or that due to native vocabulary, F (1, 71)=5.4, p<.05, although not after controlling for the variance due to the combined effect of phonological memory and native vocabulary, F (1, 70)=3.0, p > .05. These results indicated that children with poorer phonological awareness had difficulty in associative learning only when the materials to be learned involved novel, verbal information.

Figures 1–4 present the cumulative percentage of children who reached the criterion (successful learning in two consecutive trials) from Trial 1 to Trial 10 in the four associative learning tasks. Figure 1 presents the performance of children with High PA and those with Low PA in the nonnative name learning task. Children with High PA demonstrated a quicker and more accurate acquisition of the three nonnative names within the testing trials of the study. There was no evidence that group differences diminished with the increased learning practice of additional trials. Figure 2 presents the performance of the two groups of children in the native name learning task. As in the nonnative learning task, children with Low PA demonstrated slower mastery and more error-prone performance in the native name learning task. Again, there was no evidence that group differences diminished with the increased learning practice of additional trials. However, compared with Figure 1, it is clear that both groups of children learned the native names more quickly than the nonnative names.

The cumulative percentages of children reaching the criterion in nonnative name learning

The cumulative percentages of children reaching the criterion in native name learning

The cumulative percentages of children reaching the criterion in familiar name learning

The cumulative percentages of children reaching the criterion in visual name learning

Figure 3 presents children's performance in the learning of familiar names. In contrast to the two novel name learning tasks, children with Low PA were comparable to children with High PA in learning to associate familiar phonological forms with unfamiliar referents. Figure 4 presents children's performance in the visual name learning task. There was no obvious evidence that children with Low PA were slower and less accurate than children with High PA in learning to associate unfamiliar visual forms with unfamiliar referents.