LSHSS
Research Article
Longitudinal Analysis of Receptive Vocabulary Growth in Young Spanish English–Speaking Children From Migrant Families Carla Wood Jackson,a Christopher Schatschneider,a and Lindsey Leacoxa
Purpose: The authors of this study described developmental trajectories and predicted kindergarten performance of Spanish and English receptive vocabulary acquisition of young Latino/a English language learners (ELLs) from socioeconomically disadvantaged migrant families. In addition, the authors examined the extent to which gender and individual initial performance in Spanish predict receptive vocabulary performance and growth rate. Method: The authors used hierarchical linear modeling of 64 children’s receptive vocabulary performance to generate growth trajectories, predict performance at school entry, and examine potential predictors of rate of growth. The timing of testing varied across children. The ELLs (prekindergarten to 2nd grade) participated in 2–5 testing sessions, each 6–12 months apart. Results: The ELLs’ average predicted standard score on an English receptive vocabulary at kindergarten was nearly 2 SDs below the mean for monolingual peers. Significant growth in the ELLs’ receptive vocabulary was observed between preschool and 2nd grade, indicating that the ELLs were slowly closing the receptive vocabulary gap, although their average score remained below the standard score mean
S
panish-speaking families are among the fastest growing linguistic group of English language learners (ELLs), representing more than 65% of non-Englishspeaking immigrants in the United States (National Center for Education Statistics [NCES], 2004). Approximately 56% of public school teachers in the United States have at least one child who is characterized as an ELL in their classrooms (Waxman, Tellez, & Walberg, 2004). Spanish-speaking ELLs
for age-matched monolingual peers. The ELLs demonstrated a significant decrease in Spanish receptive vocabulary standard scores over time. Initial Spanish receptive vocabulary was a significant predictor of growth in English receptive vocabulary. High initial Spanish receptive vocabulary was associated with greater growth in English receptive vocabulary and decelerated growth in Spanish receptive vocabulary. Gender was not a significant predictor of growth in either English or Spanish receptive vocabulary. Conclusion: ELLs from low socioeconomic backgrounds may be expected to perform lower in English compared with their monolingual English peers in kindergarten. Performance in Spanish at school entry may be useful in identifying children who require more intensive instructional support for English vocabulary growth. Findings substantiate the need for progress monitoring across the early school years. Key Words: English language learner, bilinguals, dual-language learners, Spanish–English speakers, vocabulary comprehension, Spanish vocabulary, English acquisition, language loss
from families of low socioeconomic status (SES) are at high risk for underachievement in language and literacy (MancillaMartinez & Lesaux, 2010). The disproportionate prevalence of childhood poverty among Latino/a ELLs (U.S. Census Bureau, 2008) complicates language and literacy achievement for these children (Suarez-Orozco & Suarez-Orozco, 2001). A disproportionate number of Latino/a children demonstrate low reading achievement, with 50% of Latino/a fourth graders scoring at or below the basic level in reading achievement (Lee, Grigg, & Donahue, 2007).
a
Florida State University, Tallahassee
Correspondence to Carla Wood Jackson: [email protected] Editor: C. Melanie Schuele Associate Editor: Megan Dunn Davison Received December 3, 2012 Revision received May 31, 2013 Accepted December 17, 2013 DOI: 10.1044/2013_LSHSS-12-0104
40
Influence of SES ELLs come from a variety of socioeconomic backgrounds; however, research findings suggest that young Disclosure: The authors have declared that no competing interests existed at the time of publication.
Language, Speech, and Hearing Services in Schools • Vol. 45 • 40–51 • January 2014 • A American Speech-Language-Hearing Association
ELLs who are from low-SES families and who begin school with low levels of English proficiency are more likely to perform below average in reading achievement than are children who are from families with more resources (Kieffer, 2008). ELLs from families with low resources may be among the most vulnerable for school failure (Kieffer, 2008; Snow, Burns, & Griffin, 1998; Wright, Diener, & Kay, 2000). Young ELLs whose parents report low levels of educational attainment for themselves are less likely to have frequent print exposure—with fewer books in the home and less frequent shared reading opportunities—than are children of parents reporting higher educational levels (Koskinen et al., 2000; Moss & Puma, 1995; Suarez-Orozco & Suarez-Orozco, 2001). Taken together, the existing literature suggests that greater shortfalls in language and literacy achievements may be expected for ELLs from low-SES families.
Spanish-Speaking Children of Migrant Family Backgrounds ELLs from migrant families represent a subgroup of ELLs from low-SES backgrounds. The term migrant describes a person who is employed by the agricultural industry and who relocates within a yearly cycle, usually to follow crop seasonal employment. Compared with children of nonmigrant families, the children of migrant families are more likely to attend low-SES schools and to have a high dropout rate (Stanton-Salazar, 2001; Suarez-Orozco & Suarez-Orozco, 2001). Children of Spanish-speaking migrant families often present with multiple risk factors for poor academic outcomes. Mexican immigrant mothers face exceptional risk and significant disadvantages (Padilla, Dalton Radey, Hummer, & Kim, 2006). The Fragile Families and Child Wellbeing Study (Reichman, Teitler, Garfinkel, & McLanahan, 2001), which involved interviews of 4,898 mothers in 16 randomly selected cities, reported that 75% of Mexican immigrant mothers had less than a high school education, 79% worked in labor or service jobs, and 70% were below the poverty line. Financial hardship has been identified as a powerful influencing factor on child health and development as it can affect the adequacy of nutrition, shelter, child care, and medical care (Gershoff, Aber, & Raver, 2003; Mayer, 1997). For some families, economic disadvantages may be attenuated by the support of extended family members and/or access to public assistance or other social supports (Guendelman, 2000). Although low maternal education and low-paying occupations affect individual families differently, multiple risk factors such as harsh living conditions, inadequate access to health services, and unstable neighborhoods can have a combined negative effect on child and family wellbeing (Gershoff et al., 2003; Shonkoff & Phillips, 2000). With respect to neighborhood instability, in the Fragile Families and Child Wellbeing Study, Reichman and colleagues (2001) reported that a significantly higher proportion of immigrant families of Mexican origin had lived in their neighborhoods for less than a year when compared with families of White and African American descent. Frequency
of moving, as an influencing factor, may not by itself be concerning, but the impact on a child’s development may be intensified when combined with cultural differences, language barriers, and lack of access to public assistance (Padilla et al., 2006). The adjustments required in relocating to a new community and changing schools are associated with increased stress for families in general; for Spanish-speaking migrant families, the language barriers and limited resources associated with poverty might be expected to intensify the risk of academic failure. Potential trends in high-risk conditions of Spanishspeaking migrant families are further substantiated by a recent report (Crosnoe, 2007) that aggregated data from the Early Childhood Longitudinal Study—Kindergarten Cohort (ECLS–K; NCES, 2002), a nationally representative sample of 22,782 American kindergartners. The ECLS–K included immigrant children born in Mexico and children born in the United States to Mexican-born parents. Based on ECLS–K data, the children of Mexican immigrant families had the lowest level of SES and highest level of poverty when compared with children of White and African American descent. Additionally, the children of Mexican immigrant families scored lower than their peers on all factors related to educational enrichment in the home, including having the least number of books in the home and being read to less frequently than children from other backgrounds.
Need for Progress Monitoring Due to ELLs’ increased risk for school failure, longitudinal studies of language and literacy achievement are needed to provide information about typical growth trajectories for progress monitoring. Kieffer (2008) conducted a longitudinal study examining the reading growth trajectories of a large sample of students from a variety of socioeconomic, ethnic, and linguistic backgrounds from both high- and low-SES schools from kindergarten through fifth grade using a national data set of the ECLS–K (Tourangeau, Lê, & Nord, 2005). The English reading measure included items assessing basic reading skills (e.g., print awareness, letter recognition, phoneme awareness, and decoding), vocabulary, and reading comprehension. Although the children’s native languages differed and their skills in their first language (L1) were not assessed, English proficiency at kindergarten entry was predictive of the rate of reading growth. Children who entered kindergarten with limited English proficiency (LEP), as measured by the Pre-Language Assessment Scale (Duncan & De Avila, 1998), diverged from the national average and fell farther behind over time on the reading measure. In contrast, dual-language learners who entered kindergarten proficient in English converged with monolingual peers from similar demographic backgrounds.
Importance of Vocabulary Among influencing factors on language and literacy achievement, vocabulary knowledge serves an important role in reading comprehension and academic achievement
Jackson et al.: Longitudinal Vocabulary Development
41
(August, Carlo, Dressler, & Snow, 2005; August & Shanahan 2006; Davison, Hammer, & Lawrence, 2011; Droop & Verhoeven, 2003; Gersten et al., 2007; Hammer, Lawrence, & Miccio, 2007; Hammill, 2004; Scarborough, 2002). Vocabulary is fundamental to reading comprehension in both monolingual and ELL children, serving as a link between oral and written language (Proctor, Carlo, August, & Snow, 2005; Scarborough, 2002). Language comprehension in either the L1 or second language (L2) during the preschool years has been shown to be a good predictor of later literacy performance in English (Davison et al., 2011; Hammer et al., 2007). Furthermore, English vocabulary knowledge has been shown to be a strong proximal and distal predictor of reading comprehension, specifically for Spanishspeaking ELLs (Proctor et al., 2005). It has been suggested that vocabulary knowledge may account for even greater variability in listening comprehension performance for ELLs than for monolingual peers (Droop & Verhoeven, 2003). Building English vocabulary knowledge has been highlighted specifically as a potential mediating factor in improving ELLs’ comprehension and metalinguistic skills (August & Shanahan, 2006). English vocabulary occurring in children’s literature and reading texts has been historically problematic for ELLs (Fitzgerald, 1995; Garcia, 1991). Children learning English as an L2 commonly show restricted depth of lexical knowledge in their L2 and have particular difficulty acquiring labels for words when the phonological representations differ from the sound patterns of their native language (Feldman & Healy, 1998; Ordóñez, Carlo, Snow, & McLaughlin, 2002). Deducing the meaning of novel words based on incidental exposures found in typical reading exposures may be especially problematic for ELLs, who lack the grammatical knowledge needed to determine meaning from context alone (Carlo et al., 2004). There is increasing recognition of the need for more longitudinal studies that (a) examine vocabulary growth in young ELLs (Davison et al., 2011; Hammer et al., 2007), (b) consider factors that influence vocabulary growth, and (c) consider how vocabulary growth influences literacy achievement. One such study by Uccelli and Páez (2007) evaluated the vocabulary skills of 24 Spanish English– speaking children in kindergarten and first grade using the Picture Vocabulary subtest in English and Spanish from the Woodcock Language Proficiency Battery—Revised (Woodcock & Muñoz-Sandoval, 1995). Although significant gains were made in English expressive vocabulary, 20 of the 24 participants had below average standard scores at both time points; however, performance varied widely. Hammer et al. (2007) investigated the influence of English exposure before school entry on children’s receptive vocabulary growth. They tested 88 bilingual children in a Head Start program twice a year across 2 years. Some participants had English communication in the home before school entry (n = 53); others were without substantial English exposure before school entry (n = 35). No between-group differences were found in the rate of change on a measure of English vocabulary (i.e., Peabody Picture Vocabulary
Test—III [PPVT–III]; Dunn & Dunn, 1997) or a measure of Spanish vocabulary (i.e., Test de Vocabulario en Imagenes Peabody [TVIP]; Dunn, Lugo, Padilla, & Dunn, 1986). Thus, early exposure to English was not found to be a factor in preschool vocabulary growth. Bilingual students’ emergent literacy skills were predicted by the growth in English receptive vocabulary skills as opposed to English skill at one static time point. English vocabulary growth across the 2 years predicted children’s performance on the Test of Early Reading Ability—2 (Reid, Hresko, & Hammill, 1991) in the spring of the kindergarten school year. It has also been suggested that dual-language growth may be predicted by early language skills. Davison et al. (2011) conducted a study to describe the language performance of young ELLs and examine the predictive nature of early language performance on later literacy skills, including vocabulary comprehension. The authors tested 81 bilingual children of Puerto Rican descent in Head Start programs in Pennsylvania. The children showed significant growth in their English receptive vocabulary and Spanish receptive vocabulary as measured by raw scores on the PPVT–III and TVIP across multiple time points. Further, the children demonstrated an increasing rate of growth during their 2 years of attendance in the preschool Head Start programs. Children who experienced some exposure to English at home showed parallel receptive vocabulary growth rates to children who were exposed to only Spanish at home, indicating that both groups were developing at the same rate. Thus, the need for additional longitudinal studies is well recognized, given the high risk for English vocabulary word learning delays and the recognized importance of vocabulary knowledge to support reading and academic achievement for ELLs broadly, and particularly for ELLs from low-SES backgrounds, such as migrant families. More studies are needed to examine the patterns of vocabulary growth and to identify influences on the rates of growth. Therefore, we aimed to examine the growth trajectory of ELLs’ vocabulary performance in English and Spanish across the early schoolage years, with a focus on children from low-SES migrant families. We asked three research questions: •
What are the typical developmental trajectories of Spanish and English receptive vocabulary acquisition of ELLs from low-SES migrant families?
•
What is the predicted receptive vocabulary performance of ELLs from low-SES migrant families at the age of kindergarten entry (60–61 months old)?
•
Does early Spanish or English receptive vocabulary performance on standardized vocabulary measures in kindergarten predict later vocabulary performance in English?
Method The study protocol was approved by the Institutional Review Board at Florida State University, and parental consent was obtained for the children to participate in the study.
42 Language, Speech, and Hearing Services in Schools • Vol. 45 • 40–51 • January 2014
Participants Young children in a seasonal education program (Panhandle Area Education Consortium [PAEC] for Migrant Education) were invited to participate in a longitudinal study of vocabulary knowledge. The PAEC program primarily serves Mexican and Mexican American children living in a rural area of northern Florida. All of the children in the program meet the criteria for the federal free-lunch program, indicating low SES. Program participants receive hearing screenings, dental evaluations, and well-child checkups. The children’s parents have migrant status (i.e., are registered as migrant farm workers). The consented sample included 111 children, but high attrition was observed; 47 of the consented children were not included in the study because they were tested only once. Thus, the study sample included 64 children who had at least two data points. Eligible participants had no identified disabilities (i.e., no remarkable history of sensory disorders or other physiological or neurological disorders). An equal number of female and male participants was desired due to mixed research findings on the influence of gender on the growth rates of ELLs (Tong, Irby, Lara-Alecio, Yoon, & Mathes, 2010). The study participants included 38 girls (59% of the sample) and 26 boys (41%). Children entered the study in preschool or kindergarten and were 37 to 82 months of age (M = 61 months; SD = 12.2) at the outset of the study. Nearly half (48%) of the participants were younger than 5 years of age at study outset; the other half were older than five but were attending kindergarten due to repeating kindergarten or entering school at an older age. Children were tested at 6- to 12-month time intervals from study entry until second grade. The timing of testing was allowed to vary across participants so as to minimize attrition. Family relocations resulted in variable timing when children were available for assessment. Thus, not all children were represented at each time point. All 64 participants (100%) were tested on at least two testing time points or waves; 31% had at least three waves, 13% had at least four waves, and 6% had five waves. In other words, 44 participants contributed a maximum of two data points, 12 discontinued after three data points, four discontinued after four data points, and four contributed a total of five data points.
Parental History Families were contacted by phone to gather background information; extended family members were informants when multiple attempts to contact the parent were unsuccessful or the children were residing temporarily with extended family members who served as guardians. Maternal education level was reported as 12th grade or below, with the exception of one mother who reported attending a course at a community college. With one exception (i.e., high school diploma), no fathers reported having completed 12th grade. Seven fathers and four mothers reported only having completed elementary school (third grade in most cases). All of the children were born in the United States (Florida,
Texas, Georgia, and North Carolina). Grandparents were reported to be born in Mexico (75%), the United States (18%), Guatemala (4%), or Honduras (4%). Parents were born in Mexico (79%), the United States (14%), Guatemala (4%), or Honduras (4%); most parents reported immigrating to the United States from Mexico.
Linguistic Environment at Home and at School Families reported Spanish to be the predominant language spoken at home; children likely had exposure to English before study enrollment in formal educational settings, in speaking with older siblings, and through incidental exposure in the community. While testing the children in kindergarten, the investigators asked about the language that family members spoke at home. When asked about their mother’s primary language, 75.0% of the children reported Spanish, 10.0% reported both Spanish and English, and the remaining 15.0% reported English. When asked about their father’s primary language, 63.0% of the children reported Spanish, 9.0% reported both Spanish and English, 18.0% reported English, and 10.0% reported that they did not know. When asked about siblings’ primary language when talking to the child, 46.0% reported Spanish, 21.5% reported Spanish and English, and 32.5% reported English. The investigators also asked the children additional questions regarding language use and environment. When asked to identify which language they knew better, 41% of the children identified Spanish, 20% identified Spanish and English as equivalent, 38% identified English, and 1% indicated that they did not know which language they knew best. Finally, when asked about television, 24% reported that they watch television primarily in Spanish, 38% said in both English and Spanish, and 38% said primarily in English. We included the children’s self-reported language because it may be indicative of which language a child identifies with more readily; however, the accuracy of the children’s perceptions was not validated, for example, by reports from parents. To provide further descriptive information about the language environment of the children’s preschool classrooms, the investigators recorded audio samples using Language Environmental Analysis (LENA) digital language processors with nine preschoolers on 5 separate days to produce descriptive snapshots of the language environment. Each child wore the processor throughout the school day. Review of the daylong samples was conducted by sampling the first minute of each 5-min consecutive audio segment that was categorized as Spanish-only, English-only, or mixed Spanish and English productions. Spanish was spoken in the preschool approximately 44% of the time (95 out of 215 audio units in 5-min segments), English was spoken approximately 42% of the time (90 out of 215 audio units in 5-min segments), and productions that included a mix of Spanish and English were used 14% of the time (30 out of 215 audio units in 5-min segments). Spanish was spoken primarily within children’s interactions with teaching assistants and Spanish-speaking peers.
Jackson et al.: Longitudinal Vocabulary Development
43
Measures We administered the Peabody Picture Vocabulary Test, Fourth Edition (PPVT–4; Dunn & Dunn, 2007), a norm-referenced measure of receptive vocabulary in English (normed for ages 2 to 90 years), to monitor the children’s receptive vocabulary in English. The assessment takes 10–15 min to administer; the child is asked to point to a picture that matches the examiner’s spoken word given a choice of four pictures. The measure was normed on 3,540 individuals in the United States, reflecting the U.S. population distribution with regard to sex, race/ethnicity, geographic region, SES, and clinical diagnosis. Split-half reliability by age for Form A and Form B was M = 0.94 (SD = 3.6). We administered the TVIP, a norm-referenced measure of receptive vocabulary in Spanish (normed for ages 2;6 [years;month] to 17;11), to monitor the children’s receptive vocabulary in Spanish. The TVIP format parallels that of the PPVT–4. The TVIP was normed on 2,707 monolingual Spanish-speaking children from Mexico and Puerto Rico. Because the majority of participants in the current study were of Mexican descent, we used the normative tables for the children from Mexico to derive standard scores.
Procedure Research assistants accompanied the children to the testing classroom, where multiple testing stations were set up with the PPVT–4, TVIP, or Expressive One
Research Article
Longitudinal Analysis of Receptive Vocabulary Growth in Young Spanish English–Speaking Children From Migrant Families Carla Wood Jackson,a Christopher Schatschneider,a and Lindsey Leacoxa
Purpose: The authors of this study described developmental trajectories and predicted kindergarten performance of Spanish and English receptive vocabulary acquisition of young Latino/a English language learners (ELLs) from socioeconomically disadvantaged migrant families. In addition, the authors examined the extent to which gender and individual initial performance in Spanish predict receptive vocabulary performance and growth rate. Method: The authors used hierarchical linear modeling of 64 children’s receptive vocabulary performance to generate growth trajectories, predict performance at school entry, and examine potential predictors of rate of growth. The timing of testing varied across children. The ELLs (prekindergarten to 2nd grade) participated in 2–5 testing sessions, each 6–12 months apart. Results: The ELLs’ average predicted standard score on an English receptive vocabulary at kindergarten was nearly 2 SDs below the mean for monolingual peers. Significant growth in the ELLs’ receptive vocabulary was observed between preschool and 2nd grade, indicating that the ELLs were slowly closing the receptive vocabulary gap, although their average score remained below the standard score mean
S
panish-speaking families are among the fastest growing linguistic group of English language learners (ELLs), representing more than 65% of non-Englishspeaking immigrants in the United States (National Center for Education Statistics [NCES], 2004). Approximately 56% of public school teachers in the United States have at least one child who is characterized as an ELL in their classrooms (Waxman, Tellez, & Walberg, 2004). Spanish-speaking ELLs
for age-matched monolingual peers. The ELLs demonstrated a significant decrease in Spanish receptive vocabulary standard scores over time. Initial Spanish receptive vocabulary was a significant predictor of growth in English receptive vocabulary. High initial Spanish receptive vocabulary was associated with greater growth in English receptive vocabulary and decelerated growth in Spanish receptive vocabulary. Gender was not a significant predictor of growth in either English or Spanish receptive vocabulary. Conclusion: ELLs from low socioeconomic backgrounds may be expected to perform lower in English compared with their monolingual English peers in kindergarten. Performance in Spanish at school entry may be useful in identifying children who require more intensive instructional support for English vocabulary growth. Findings substantiate the need for progress monitoring across the early school years. Key Words: English language learner, bilinguals, dual-language learners, Spanish–English speakers, vocabulary comprehension, Spanish vocabulary, English acquisition, language loss
from families of low socioeconomic status (SES) are at high risk for underachievement in language and literacy (MancillaMartinez & Lesaux, 2010). The disproportionate prevalence of childhood poverty among Latino/a ELLs (U.S. Census Bureau, 2008) complicates language and literacy achievement for these children (Suarez-Orozco & Suarez-Orozco, 2001). A disproportionate number of Latino/a children demonstrate low reading achievement, with 50% of Latino/a fourth graders scoring at or below the basic level in reading achievement (Lee, Grigg, & Donahue, 2007).
a
Florida State University, Tallahassee
Correspondence to Carla Wood Jackson: [email protected] Editor: C. Melanie Schuele Associate Editor: Megan Dunn Davison Received December 3, 2012 Revision received May 31, 2013 Accepted December 17, 2013 DOI: 10.1044/2013_LSHSS-12-0104
40
Influence of SES ELLs come from a variety of socioeconomic backgrounds; however, research findings suggest that young Disclosure: The authors have declared that no competing interests existed at the time of publication.
Language, Speech, and Hearing Services in Schools • Vol. 45 • 40–51 • January 2014 • A American Speech-Language-Hearing Association
ELLs who are from low-SES families and who begin school with low levels of English proficiency are more likely to perform below average in reading achievement than are children who are from families with more resources (Kieffer, 2008). ELLs from families with low resources may be among the most vulnerable for school failure (Kieffer, 2008; Snow, Burns, & Griffin, 1998; Wright, Diener, & Kay, 2000). Young ELLs whose parents report low levels of educational attainment for themselves are less likely to have frequent print exposure—with fewer books in the home and less frequent shared reading opportunities—than are children of parents reporting higher educational levels (Koskinen et al., 2000; Moss & Puma, 1995; Suarez-Orozco & Suarez-Orozco, 2001). Taken together, the existing literature suggests that greater shortfalls in language and literacy achievements may be expected for ELLs from low-SES families.
Spanish-Speaking Children of Migrant Family Backgrounds ELLs from migrant families represent a subgroup of ELLs from low-SES backgrounds. The term migrant describes a person who is employed by the agricultural industry and who relocates within a yearly cycle, usually to follow crop seasonal employment. Compared with children of nonmigrant families, the children of migrant families are more likely to attend low-SES schools and to have a high dropout rate (Stanton-Salazar, 2001; Suarez-Orozco & Suarez-Orozco, 2001). Children of Spanish-speaking migrant families often present with multiple risk factors for poor academic outcomes. Mexican immigrant mothers face exceptional risk and significant disadvantages (Padilla, Dalton Radey, Hummer, & Kim, 2006). The Fragile Families and Child Wellbeing Study (Reichman, Teitler, Garfinkel, & McLanahan, 2001), which involved interviews of 4,898 mothers in 16 randomly selected cities, reported that 75% of Mexican immigrant mothers had less than a high school education, 79% worked in labor or service jobs, and 70% were below the poverty line. Financial hardship has been identified as a powerful influencing factor on child health and development as it can affect the adequacy of nutrition, shelter, child care, and medical care (Gershoff, Aber, & Raver, 2003; Mayer, 1997). For some families, economic disadvantages may be attenuated by the support of extended family members and/or access to public assistance or other social supports (Guendelman, 2000). Although low maternal education and low-paying occupations affect individual families differently, multiple risk factors such as harsh living conditions, inadequate access to health services, and unstable neighborhoods can have a combined negative effect on child and family wellbeing (Gershoff et al., 2003; Shonkoff & Phillips, 2000). With respect to neighborhood instability, in the Fragile Families and Child Wellbeing Study, Reichman and colleagues (2001) reported that a significantly higher proportion of immigrant families of Mexican origin had lived in their neighborhoods for less than a year when compared with families of White and African American descent. Frequency
of moving, as an influencing factor, may not by itself be concerning, but the impact on a child’s development may be intensified when combined with cultural differences, language barriers, and lack of access to public assistance (Padilla et al., 2006). The adjustments required in relocating to a new community and changing schools are associated with increased stress for families in general; for Spanish-speaking migrant families, the language barriers and limited resources associated with poverty might be expected to intensify the risk of academic failure. Potential trends in high-risk conditions of Spanishspeaking migrant families are further substantiated by a recent report (Crosnoe, 2007) that aggregated data from the Early Childhood Longitudinal Study—Kindergarten Cohort (ECLS–K; NCES, 2002), a nationally representative sample of 22,782 American kindergartners. The ECLS–K included immigrant children born in Mexico and children born in the United States to Mexican-born parents. Based on ECLS–K data, the children of Mexican immigrant families had the lowest level of SES and highest level of poverty when compared with children of White and African American descent. Additionally, the children of Mexican immigrant families scored lower than their peers on all factors related to educational enrichment in the home, including having the least number of books in the home and being read to less frequently than children from other backgrounds.
Need for Progress Monitoring Due to ELLs’ increased risk for school failure, longitudinal studies of language and literacy achievement are needed to provide information about typical growth trajectories for progress monitoring. Kieffer (2008) conducted a longitudinal study examining the reading growth trajectories of a large sample of students from a variety of socioeconomic, ethnic, and linguistic backgrounds from both high- and low-SES schools from kindergarten through fifth grade using a national data set of the ECLS–K (Tourangeau, Lê, & Nord, 2005). The English reading measure included items assessing basic reading skills (e.g., print awareness, letter recognition, phoneme awareness, and decoding), vocabulary, and reading comprehension. Although the children’s native languages differed and their skills in their first language (L1) were not assessed, English proficiency at kindergarten entry was predictive of the rate of reading growth. Children who entered kindergarten with limited English proficiency (LEP), as measured by the Pre-Language Assessment Scale (Duncan & De Avila, 1998), diverged from the national average and fell farther behind over time on the reading measure. In contrast, dual-language learners who entered kindergarten proficient in English converged with monolingual peers from similar demographic backgrounds.
Importance of Vocabulary Among influencing factors on language and literacy achievement, vocabulary knowledge serves an important role in reading comprehension and academic achievement
Jackson et al.: Longitudinal Vocabulary Development
41
(August, Carlo, Dressler, & Snow, 2005; August & Shanahan 2006; Davison, Hammer, & Lawrence, 2011; Droop & Verhoeven, 2003; Gersten et al., 2007; Hammer, Lawrence, & Miccio, 2007; Hammill, 2004; Scarborough, 2002). Vocabulary is fundamental to reading comprehension in both monolingual and ELL children, serving as a link between oral and written language (Proctor, Carlo, August, & Snow, 2005; Scarborough, 2002). Language comprehension in either the L1 or second language (L2) during the preschool years has been shown to be a good predictor of later literacy performance in English (Davison et al., 2011; Hammer et al., 2007). Furthermore, English vocabulary knowledge has been shown to be a strong proximal and distal predictor of reading comprehension, specifically for Spanishspeaking ELLs (Proctor et al., 2005). It has been suggested that vocabulary knowledge may account for even greater variability in listening comprehension performance for ELLs than for monolingual peers (Droop & Verhoeven, 2003). Building English vocabulary knowledge has been highlighted specifically as a potential mediating factor in improving ELLs’ comprehension and metalinguistic skills (August & Shanahan, 2006). English vocabulary occurring in children’s literature and reading texts has been historically problematic for ELLs (Fitzgerald, 1995; Garcia, 1991). Children learning English as an L2 commonly show restricted depth of lexical knowledge in their L2 and have particular difficulty acquiring labels for words when the phonological representations differ from the sound patterns of their native language (Feldman & Healy, 1998; Ordóñez, Carlo, Snow, & McLaughlin, 2002). Deducing the meaning of novel words based on incidental exposures found in typical reading exposures may be especially problematic for ELLs, who lack the grammatical knowledge needed to determine meaning from context alone (Carlo et al., 2004). There is increasing recognition of the need for more longitudinal studies that (a) examine vocabulary growth in young ELLs (Davison et al., 2011; Hammer et al., 2007), (b) consider factors that influence vocabulary growth, and (c) consider how vocabulary growth influences literacy achievement. One such study by Uccelli and Páez (2007) evaluated the vocabulary skills of 24 Spanish English– speaking children in kindergarten and first grade using the Picture Vocabulary subtest in English and Spanish from the Woodcock Language Proficiency Battery—Revised (Woodcock & Muñoz-Sandoval, 1995). Although significant gains were made in English expressive vocabulary, 20 of the 24 participants had below average standard scores at both time points; however, performance varied widely. Hammer et al. (2007) investigated the influence of English exposure before school entry on children’s receptive vocabulary growth. They tested 88 bilingual children in a Head Start program twice a year across 2 years. Some participants had English communication in the home before school entry (n = 53); others were without substantial English exposure before school entry (n = 35). No between-group differences were found in the rate of change on a measure of English vocabulary (i.e., Peabody Picture Vocabulary
Test—III [PPVT–III]; Dunn & Dunn, 1997) or a measure of Spanish vocabulary (i.e., Test de Vocabulario en Imagenes Peabody [TVIP]; Dunn, Lugo, Padilla, & Dunn, 1986). Thus, early exposure to English was not found to be a factor in preschool vocabulary growth. Bilingual students’ emergent literacy skills were predicted by the growth in English receptive vocabulary skills as opposed to English skill at one static time point. English vocabulary growth across the 2 years predicted children’s performance on the Test of Early Reading Ability—2 (Reid, Hresko, & Hammill, 1991) in the spring of the kindergarten school year. It has also been suggested that dual-language growth may be predicted by early language skills. Davison et al. (2011) conducted a study to describe the language performance of young ELLs and examine the predictive nature of early language performance on later literacy skills, including vocabulary comprehension. The authors tested 81 bilingual children of Puerto Rican descent in Head Start programs in Pennsylvania. The children showed significant growth in their English receptive vocabulary and Spanish receptive vocabulary as measured by raw scores on the PPVT–III and TVIP across multiple time points. Further, the children demonstrated an increasing rate of growth during their 2 years of attendance in the preschool Head Start programs. Children who experienced some exposure to English at home showed parallel receptive vocabulary growth rates to children who were exposed to only Spanish at home, indicating that both groups were developing at the same rate. Thus, the need for additional longitudinal studies is well recognized, given the high risk for English vocabulary word learning delays and the recognized importance of vocabulary knowledge to support reading and academic achievement for ELLs broadly, and particularly for ELLs from low-SES backgrounds, such as migrant families. More studies are needed to examine the patterns of vocabulary growth and to identify influences on the rates of growth. Therefore, we aimed to examine the growth trajectory of ELLs’ vocabulary performance in English and Spanish across the early schoolage years, with a focus on children from low-SES migrant families. We asked three research questions: •
What are the typical developmental trajectories of Spanish and English receptive vocabulary acquisition of ELLs from low-SES migrant families?
•
What is the predicted receptive vocabulary performance of ELLs from low-SES migrant families at the age of kindergarten entry (60–61 months old)?
•
Does early Spanish or English receptive vocabulary performance on standardized vocabulary measures in kindergarten predict later vocabulary performance in English?
Method The study protocol was approved by the Institutional Review Board at Florida State University, and parental consent was obtained for the children to participate in the study.
42 Language, Speech, and Hearing Services in Schools • Vol. 45 • 40–51 • January 2014
Participants Young children in a seasonal education program (Panhandle Area Education Consortium [PAEC] for Migrant Education) were invited to participate in a longitudinal study of vocabulary knowledge. The PAEC program primarily serves Mexican and Mexican American children living in a rural area of northern Florida. All of the children in the program meet the criteria for the federal free-lunch program, indicating low SES. Program participants receive hearing screenings, dental evaluations, and well-child checkups. The children’s parents have migrant status (i.e., are registered as migrant farm workers). The consented sample included 111 children, but high attrition was observed; 47 of the consented children were not included in the study because they were tested only once. Thus, the study sample included 64 children who had at least two data points. Eligible participants had no identified disabilities (i.e., no remarkable history of sensory disorders or other physiological or neurological disorders). An equal number of female and male participants was desired due to mixed research findings on the influence of gender on the growth rates of ELLs (Tong, Irby, Lara-Alecio, Yoon, & Mathes, 2010). The study participants included 38 girls (59% of the sample) and 26 boys (41%). Children entered the study in preschool or kindergarten and were 37 to 82 months of age (M = 61 months; SD = 12.2) at the outset of the study. Nearly half (48%) of the participants were younger than 5 years of age at study outset; the other half were older than five but were attending kindergarten due to repeating kindergarten or entering school at an older age. Children were tested at 6- to 12-month time intervals from study entry until second grade. The timing of testing was allowed to vary across participants so as to minimize attrition. Family relocations resulted in variable timing when children were available for assessment. Thus, not all children were represented at each time point. All 64 participants (100%) were tested on at least two testing time points or waves; 31% had at least three waves, 13% had at least four waves, and 6% had five waves. In other words, 44 participants contributed a maximum of two data points, 12 discontinued after three data points, four discontinued after four data points, and four contributed a total of five data points.
Parental History Families were contacted by phone to gather background information; extended family members were informants when multiple attempts to contact the parent were unsuccessful or the children were residing temporarily with extended family members who served as guardians. Maternal education level was reported as 12th grade or below, with the exception of one mother who reported attending a course at a community college. With one exception (i.e., high school diploma), no fathers reported having completed 12th grade. Seven fathers and four mothers reported only having completed elementary school (third grade in most cases). All of the children were born in the United States (Florida,
Texas, Georgia, and North Carolina). Grandparents were reported to be born in Mexico (75%), the United States (18%), Guatemala (4%), or Honduras (4%). Parents were born in Mexico (79%), the United States (14%), Guatemala (4%), or Honduras (4%); most parents reported immigrating to the United States from Mexico.
Linguistic Environment at Home and at School Families reported Spanish to be the predominant language spoken at home; children likely had exposure to English before study enrollment in formal educational settings, in speaking with older siblings, and through incidental exposure in the community. While testing the children in kindergarten, the investigators asked about the language that family members spoke at home. When asked about their mother’s primary language, 75.0% of the children reported Spanish, 10.0% reported both Spanish and English, and the remaining 15.0% reported English. When asked about their father’s primary language, 63.0% of the children reported Spanish, 9.0% reported both Spanish and English, 18.0% reported English, and 10.0% reported that they did not know. When asked about siblings’ primary language when talking to the child, 46.0% reported Spanish, 21.5% reported Spanish and English, and 32.5% reported English. The investigators also asked the children additional questions regarding language use and environment. When asked to identify which language they knew better, 41% of the children identified Spanish, 20% identified Spanish and English as equivalent, 38% identified English, and 1% indicated that they did not know which language they knew best. Finally, when asked about television, 24% reported that they watch television primarily in Spanish, 38% said in both English and Spanish, and 38% said primarily in English. We included the children’s self-reported language because it may be indicative of which language a child identifies with more readily; however, the accuracy of the children’s perceptions was not validated, for example, by reports from parents. To provide further descriptive information about the language environment of the children’s preschool classrooms, the investigators recorded audio samples using Language Environmental Analysis (LENA) digital language processors with nine preschoolers on 5 separate days to produce descriptive snapshots of the language environment. Each child wore the processor throughout the school day. Review of the daylong samples was conducted by sampling the first minute of each 5-min consecutive audio segment that was categorized as Spanish-only, English-only, or mixed Spanish and English productions. Spanish was spoken in the preschool approximately 44% of the time (95 out of 215 audio units in 5-min segments), English was spoken approximately 42% of the time (90 out of 215 audio units in 5-min segments), and productions that included a mix of Spanish and English were used 14% of the time (30 out of 215 audio units in 5-min segments). Spanish was spoken primarily within children’s interactions with teaching assistants and Spanish-speaking peers.
Jackson et al.: Longitudinal Vocabulary Development
43
Measures We administered the Peabody Picture Vocabulary Test, Fourth Edition (PPVT–4; Dunn & Dunn, 2007), a norm-referenced measure of receptive vocabulary in English (normed for ages 2 to 90 years), to monitor the children’s receptive vocabulary in English. The assessment takes 10–15 min to administer; the child is asked to point to a picture that matches the examiner’s spoken word given a choice of four pictures. The measure was normed on 3,540 individuals in the United States, reflecting the U.S. population distribution with regard to sex, race/ethnicity, geographic region, SES, and clinical diagnosis. Split-half reliability by age for Form A and Form B was M = 0.94 (SD = 3.6). We administered the TVIP, a norm-referenced measure of receptive vocabulary in Spanish (normed for ages 2;6 [years;month] to 17;11), to monitor the children’s receptive vocabulary in Spanish. The TVIP format parallels that of the PPVT–4. The TVIP was normed on 2,707 monolingual Spanish-speaking children from Mexico and Puerto Rico. Because the majority of participants in the current study were of Mexican descent, we used the normative tables for the children from Mexico to derive standard scores.
Procedure Research assistants accompanied the children to the testing classroom, where multiple testing stations were set up with the PPVT–4, TVIP, or Expressive One
