The Relationship Between Cognate Awareness and English Comprehension Among Spanish—English Bilingual Fourth Grade Students

The Relationship Between Cognate Awareness and English Comprehension Among Spanish–English Bilingual Fourth Grade Students C.. Do Spanish-speaking bilinguals perform as well or better

The Relationship Between Cognate Awareness and

English Comprehension Among Spanish–English

Bilingual Fourth Grade Students

C PATRICK PROCTOR

Boston College

Chestnut Hill, Massachusetts, United States

ELAINE MO

Center for Applied Special Technology

Wakefi eld, Massachusetts, United States

Second language (L2) learners are, to varying degrees, represented in

enrollment percentages in school districts around the world Immigration

trends in the United States, Canada, France, Spain, the Netherlands, and

other countries have been on the rise for decades, and educators the

world over are pressed to incorporate the linguistic needs of L2 learners

with the more general curricular requirements that align with

expecta-tions of literacy achievement in their respective countries In recent years,

efforts have been afoot to understand processes of literacy development

among bilingual learners, particularly in the arena of cross-linguistic

transfer, so that instruction might be better tailored to meet the distinct

linguistic needs of bilingual children

Although it appears clear that, for alphabetic languages, skills such as

phonemic awareness, alphabetic knowledge, and word reading are quite

robust in transferring between languages that share a common alphabet

(e.g., Spanish and English, Spanish and French, English and Dutch), the

extent of transferability for skills such as vocabulary knowledge is far less

clear (Proctor, August, Carlo, & Snow, 2006) It is widely accepted that

vocabulary knowledge plays an important role in fi rst language (L1) and

L2 reading achievement (Anderson & Freebody, 1981; Proctor, Carlo,

August, & Snow, 2005), yet research is only just beginning to show that

depth of vocabulary knowledge (e.g., morphological awareness) can

leverage L1 literacy skills in the service of L2 reading comprehension

One component of lexical depth is cognate recognition A cognate is a

word that shares similar orthographic and semantic characteristics in two

languages, such that the spelling and meaning of a word and its cognate

are highly similar For example, rapid and rápido are Spanish–English

nates, and escuela , école , and escola are Spanish–French–Portuguese

cog-nates Cognate recognition may be especially useful for L1-literate students

who are reading academic L2 texts, because high frequency

conversa-tional words in Spanish, such as rápido , are often low frequency academic

terms in English (e.g., rapid ) This phenomenon is consistent for Spanish–

English bilinguals in English-dominant countries worldwide

The shared Latin base between high frequency Spanish words, like

rápido , and low frequency English words, like rapid , theoretically may

allow bilingual students to draw on their L1 knowledge in making sense

of more challenging English vocabulary without any explicit vocabulary teaching (though explicit teaching does appear to help matters, as Nagy, García, Durguno˘glu, & Hancin-Bhatt, 1993, have shown) Indeed, teach-ers and researchteach-ers alike have reported on L1-literate students who appear to benefi t from cognate recognition during L2 reading (Hancin-Bhatt & Nagy, 1994; Jiménez, García, & Pearson, 1996) At present, how-ever, relatively little is known about how cognate knowledge interacts with the English reading performance of Spanish–English bilinguals Hancin-Bhatt and Nagy (1994) have suggested that students’ cognate awareness is developmentally constrained The researchers investigated the cognate and noncognate translation abilities of Grade 4, 6, and

8 Latino bilingual students and found a signifi cant interaction be-tween grade and cognate status They concluded that an age-based devel-opmental trend in recognizing Spanish cognates accounted for a signifi cant proportion of translation ability It is quite possible, however, that the discriminating effect of age may have been confounded with English language profi ciency, opening a different, though clearly related, window on the developmental parameters of Spanish–English cognate recognition

RESEARCH QUESTIONS

Using English reading comprehension as the developmental bench-mark within a group of monolingual and bilingual Grade 4 students, this study sought to answer two research questions:

1 Do Spanish-speaking bilinguals perform as well or better than English monolingual students on a reading vocabulary test that has a high percentage of Spanish–English cognates?

2 What is the relationship between Spanish-speaking bilinguals’ knowledge of cognate reading vocabulary and English reading comprehension?

METHOD

Participants

The participants were 30 Grade 4 students in two classrooms in a medium-sized school district in Southern California The students were

recruited from two connected classrooms to participate in a 4-week

inter-vention designed to promote vocabulary knowledge and reading

com-prehension (for a report on the larger study, see Proctor, Dalton, &

Grisham, 2007) The two teachers worked as a team with both of the

classes, and they selected all of their Spanish–English bilinguals as well as

monolingual English students whom they considered to be struggling

with reading Thus, at the beginning of the intervention, the

participat-ing students were performparticipat-ing on average at the 31st percentile in

read-ing comprehension on the Gates–MacGinitie Readread-ing Comprehension

subtest (MacGinitie, MacGinitie, Maria, & Dreyer, 2000) Performance at

this level means 69% of the norming sample population outscored the

students participating in the study The students were comprised of

16 Spanish–English bilinguals and 14 English monolinguals There were

11 boys and 19 girls The district in which the school was located was

rela-tively affl uent; however, this particular school was the sole recipient of

Title I funds in the district, which were based on the percentage of

stu-dents receiving free or reduced-price lunches

Measures

Reading Comprehension

The Gates-MacGinitie Reading Comprehension subtest, a whole

group, timed, multiple-choice assessment, was administered as a test

of the students’ reading comprehension The assessment consisted of

14 passages followed by a series of questions, totaling 48, designed to

cap-ture main ideas and details described in a particular passage Students

had 35 minutes to complete the assessment All scores are reported as

percentiles

Spanish–English Cognate Awareness

The Cognate Awareness Test (Malabonga, Kenyon, August, Louguit, &

Carlo, 2004) is a reading vocabulary assessment that consists of 48 items

of both cognate and noncognate English words matched for frequency in

English All words were of low frequency in English and 50% of them had

Spanish cognates that are considered high frequency in Spanish (e.g.,

English tranquil and Spanish tranquilo ); the other half had no Spanish

cognate (e.g., frenzied ) Students were presented with a target word and

had to choose a correct synonym from a list of four possible responses

This test was scored in four different ways Monolingual and bilingual

stu-dents were compared by (a) their percentage correct on the test (variable

RAW), (b) their percentage correct on noncognate (variable NONCOG) items on the assessment, (c) their percentage correct on the cognate items only (variable COG), and (d) the COG/RAW ratio (variable RATIO), which represents the percentage of total correct items on the assessment that were cognates

Procedures

The Cognate Awareness Test Malabonga and the Gates-MacGinitie Reading Comprehension subtest were administered in whole group set-tings to all 30 students following the intervention During this interven-tion, students read 8 short Internet-based texts, 4 narrative and 4 informational Interactive supports that targeted depth and breadth of vocabulary development (Lively, August, Carlo, & Snow, 2003; Nagy, 1988) were embedded into the digital text (Proctor et al., 2007)

In the digital environment, for each text, prereading vocabulary

activi-ties targeted fi ve power words, all of which were Spanish–English cognates

Students were exposed to the Spanish–English cognate relationship between each word by Al, a bilingual literacy coach embedded in the

environment For example, Al’s script for the power word decorate was, The Spanish word for decorate is decorar Do you notice that these words look similar to each other? That means they’re cognates: words that look and mean the same in Spanish and English! Students were exposed to between 20 and 40

of these cognate alerts in total Both bilingual and monolingual students

viewed the cognate alerts The fact that the monolinguals viewed Spanish translations and received cognate awareness instruction was considered benefi cial by the students, their teachers, and their parents because the district considered foreign language instruction a priority

A set of analyses was conducted to investigate performance differences and associations between bilingual and monolingual students on the Gates-MacGinitie and the four variables of interest: RAW, NONCOG, COG, and RATIO

RESULTS

Table 1 displays the means, standard deviations, ranges, t -statistics, and

signifi cance levels for reading comprehension and the cognate aware-ness overall test scores (RAW) by bilingual (BIL) status Individual t

tests for equality of means indicated that bilinguals and monolinguals

dif-fered signifi cantly on reading comprehension ( t = −2.61, p < 0.05) in

favor of the monolingual students However, the total score for the Cognate Awareness Test showed no signifi cant performance differences

( t < 1.96)

The bivariate scatterplot in Figure 1 displays results depicting the

rela-tionship between percentage correct on the entire test and the

percent-age correct on cognates only, disaggregated by bilingual status This

fi gure graphically displays how bilinguals (indicated by stars) who

per-formed similarly to monolinguals on total percentage correct posted

gen-erally higher proportions of correct cognates, suggesting that cognate

TABLE 1 Comparing Bilinguals and Monolinguals on Comprehension Scores and the Cognate

Awareness Test ( N = 30)

Outcome Language status Mean (SD) Min-Max t

Gates MacGinitie

Comprehension percentile ranking

BIL 21.53 (16.80) 6–61 −2.61*

MONO 41.47 (24.38) 10–95 Cognate Awareness

Test Total percentage correct (RAW)

BIL 35.28 (0.12) 19–54 −0.61 MONO 37.78 (0.11) 17–58

* p < 0.05; BIL = bilingual; MONO = monolingual; SD = standard deviation.

FIGURE 1 Bivariate Scatterplot of Percentage of Correct Cognate Items by Total Percentage of

All Items on the Cognate Awareness Test

presence positively infl uenced the overall performance of the bilingual students

Table 2 displays additional analyses using the ratio of correct cognates

to total percentage correct (RATIO), and percentage of correct cognates overall (COG) In the fi rst case, there is a signifi cant difference for ratio

of cognates correct to total score, in favor of the bilingual students

( t = 2.77, p < 0.05) That is, when assessing the correct items from the

Cognate Awareness Test, a signifi cantly larger percentage of the bilingual

students’ correct responses were on the cognate items A t test indicated

no signifi cant difference between the groups for the percentage of cor-rect cognates, further suggesting that cognate presence may be a factor

in narrowing the vocabulary performance difference between these groups

Table 3 displays the correlations between the different variables There was a signifi cant, strong, negative correlation between bilingual status and reading comprehension score on the Gates–MacGinitie, further indi-cating that monolingual students outperformed their bilingual counter-parts Similarly, bilingual status and noncognate scores showed a signifi cant, strong, negative correlation such that monolinguals were out-performing their bilingual counterparts on noncognate items What was

TABLE 2 Comparing Bilinguals and Monolinguals on the Cognate Awareness Test Using Two Outcome

Measures

Cognate Awareness Test Language status Mean (SD) Min-Max t

Percentage of correct cognates to BIL 49.95 (0.01) 31–64 2.77* total correct (RATIO)

MONO 40.49 (0.01) 27–63 Percentage of correct cognates (COG) BIL 36.11 (0.15) 17–58 1.52

MONO 29.44 (0.01) 17–46

* p < 0.05; BIL = bilingual; MONO = monolingual; SD = standard deviation

TABLE 3 Correlations Between Predictor (Bilingual), Control (Reading Comprehension), And Possible

Outcome Variables

1 Bilingual (0 = no, 1 = yes) 2 Reading comprehension 3 NONCOG 4 COG 5 RATIO

* p < 0.05 ∼ p < 0.10; ** p < 0.01; *** p < 0.001

surprising was the absence of a signifi cant correlation between bilingual

status and percentage of correct cognates That is, monolinguals did not

outperform their bilingual peers, suggesting an additive effect of word

type (cognate vs noncognate) on the vocabulary performance of the

bilinguals in the sample In addition, the strong, positive correlation

between bilingual status and the ratio of correct cognates to total score

also suggests that the bilingual students’ vocabulary performance may

have been affected by the presence of cognate items

Given that these results suggested that cognate presence may be infl

u-encing test performance, two regression models were tested to

investi-gate the relationship between cognate awareness and English reading

comprehension We chose to run regression analyses despite our small

sample size because the primary focus of the study was to test the

devel-opmental relationship between English reading comprehension and

cog-nate recognition, which is diffi cult to adequately assess through bivariate

correlations alone

Each regression model controlled for English reading comprehension

and assessed the effect of bilingual status on the cognate recognition

vari-ables: COG and RATIO Both models in Table 4 showed interesting

results for the interactive relationship between language status (bilingual

or monolingual) and English reading comprehension In each case, it

appeared that increased reading comprehension was associated with

higher cognate recognition for bilinguals but not for monolinguals

A model predicting total score (RAW) on the Cognate Awareness Test

resulted in no signifi cant main effects for English reading

comprehen-sion or bilingual status nor of any signifi cant interaction between them

Thus, overall on the test, monolinguals and bilinguals were performing

comparably

TABLE 4 Final Regression Models Explaining Variation in Scores for Two Outcome Measures

on the Cognate Awareness Test

Predicting RATIO Predicting COG

* p < 0.05 ∼ p < 0.10; ** p < 0.01; *** p < 0.001

Note Small regression coeffi cients are due to the difference in scale between the outcome

mea-sures (RATIO and COG), inputted as decimal representations of percentages, and the reading

comprehension measure, inputted as whole numbers

Figure 2 graphs the interaction between comprehension score and language status in predicting RATIO For lower levels of English compre-hension scores, bilinguals and monolinguals resembled each other with respect to RATIO, but positive changes in English comprehension were associated with increasing differences in group RATIOs For example, a bilingual and monolingual with English comprehension scores at the 10th percentile would have similar predicted RATIO scores of 47% and 49% However, a bilingual and monolingual student with English reading comprehension at the 60th percentile would have predicted RATIOs of 53% and 36%, respectively, an appreciable group difference

A similar phenomenon was noted in predicting the percentage correct for cognate items only (COG) Gates–MacGinitie comprehension score

and language status interacted signifi cantly ( t = 2.130, p < 0.05) such that

bilingual and monolingual students with comprehension scores at the 5th percentile would have similar predicted cognate scores of 31% and 33%, respectively, but dissimilar predicted cognate scores of 62% and 28% when comprehension is at the 60th percentile, or just above average performance in reading (see Figure 3 )

DISCUSSION

This study was designed to investigate the effect of English reading comprehension on cognate knowledge among a sample of Spanish-speaking bilinguals alongside their monolingual English-Spanish-speaking coun-terparts In order to understand this developmental relationship, the effects of English reading comprehension were controlled in regression analyses, and the impact of language status (i.e., bilingual or monolin-gual) was assessed

FIGURE 2 Predicted Values of RATIO as a Function of Comprehension Scores by Bilingual (ELL)

or Monolingual (non-ELL) Status

We found a signifi cant effect of bilingual status predicting cognate

awareness Bilingual students signifi cantly outperformed their

monolin-gual counterparts on the ratio of correct items that were represented by

cognates Current trends in research on bilingual populations, dating to

Cummins’s (1979, 1984) theories of a common underlying profi ciency

and L1 thresholds for adequate L2 acquisition argue strenuously that L1

literacy skills are important, if not crucial, in understanding variation in

L2 comprehension outcomes

Almost paradoxically, however, research among adolescent and adult

L2 learners, who are more likely to have well-developed L1 literacy

skills than their younger counterparts, has focused on the role of improved

L2 reading achievement as the key to fostering the cross-linguistic

trans-fer for improved L2 literacy outcomes Specifi cally, as L2 lexical and

grammatical knowledge improve, adult L2 reading increasingly

resem-bles the L1 reading process, and the reader is thus able to use

compre-hension strategies developed in the L1 to L2 reading (Laufer, 1997;

Nation, 2001) The results reported here are in line with this

phenomenon

We have related instructional recommendations for teachers of English

to speakers of other languages Instruction that promotes depth and

breadth of vocabulary knowledge, including cognate awareness for

bilin-guals who speak typologically similar languages, deserves further

consider-ation and implementconsider-ation An example of such an approach to improve

young readers’ comprehension is the recent work by Carlo et al (2004)

The authors worked in several urban classrooms with high numbers of

immigrant bilingual and English language learners implementing a

vocab-ulary improvement program targeting depth and breadth of vocabvocab-ulary

development The program was specifi cally designed to teach students

FIGURE 3 Predicted Values of COG as a Function of Comprehension Scores

by Bilingual (ELL) or Monolingual (Non-ELL) Status

word-learning strategies that focused on fi ve components of word knowl-edge: depth of meaning, polysemy, morphology, cross-linguistic features (including cognate recognition), and spelling and punctuation By focus-ing on depth of vocabulary knowledge, fewer words were targeted in defer-ence to learning “useful words and word-learning strategies” (p 192) The program proved effective for both bilingual and monolingual students Coady (1993), writing from an English as a foreign language perspec-tive, argues that strategies such as cognate recognition may be more effec-tive after students have gained some L2 profi ciency Indeed, more genuine metalinguistic insights are likely possible when the two language and literacy systems are relatively well established so that the learner can begin to make meaningful comparisons between them Surely, then, native language literacy is of crucial importance in thinking about how

fi rst and second languages interact, and it is a limitation of this study that

we did not collect Spanish language literacy data Thus, future research

in this area would do well to collect data among larger samples of stu-dents and a more complete assessment battery, including measures of L1 literacy development Additionally, more inquiry is needed into the rela-tionships between other orthographically linked languages where cog-nate relationships may be exploited Does a similar phenomenon exist among French–English bilinguals, for example? Answering these ques-tions would shed light on whether there exists a degree of universality relative to cognate recognition and improve thinking on how to best exploit the use of cognates for instructional and academic gain

ACKNOWLEDGMENTS

We thank the students and teachers for their participation and dedication to this proj-ect In addition, we thank Bridget Dalton for her support as well as her keen eye on previous versions of this manuscript

THE AUTHORS

C Patrick Proctor is an assistant professor of literacy and language arts at the Lynch School of Education at Boston College, Chestnut Hill, Massachusetts, United States His primary focus is on the literacy development of bilingual students at risk for lit-eracy diffi culties His current research focuses on vocabulary instruction for students from various linguistic backgrounds to improve reading comprehension

Elaine Mo is a doctoral student at the Harvard Graduate School of Education, Wakefi eld, Massachusetts, United States Her work centers on language and literacy development of English language learners, and her interests include investigating how culture infl uences learning and teaching Her current research focuses on the vocabulary development of both English monolingual and bilingual students