Rehearsal and aptitude in foreign vocabulary learning.

Rehearsal and Aptitude in ForeignVocabulary LearningKRISTINA DAHLENSharon Public SchoolsForeign Language Department181 Pond StreetSharon, MA 02067Email: [email protected]

CATHERINE CALDWELL–HARRISBoston UniversityDepartment of Psychology64 Cummington StreetBoston, MA 02215Email: [email protected]

Foreign language learners rehearse the language they are learning, both vocally and subvocally, as part ofinner speech development (Guerrero, 2005). To add to the research on which method of rehearsal ismore effective for initial foreign vocabulary learning, this study examined two hypotheses: first, that themanner of rehearsal of foreign vocabulary words influences subsequent recall and recognition of thosewords and second, that foreign language aptitude, as measured by the Modern Language Aptitude Test,can predict the level of success in initial learning of foreign vocabulary. English speakers (n! 88) heard aTurkish noun said three times while looking at a picture of that noun. They then rehearsed the 20Turkish nouns in 1 of 4 conditions: vocal rehearsal with their own auditory feedback, vocal rehearsal withtheir auditory feedback suppressed by white noise, subvocal rehearsal as part of inner speech, and norehearsal. Participants were then asked to recall the Turkish word for each picture and recognizethe words when said in a sentence. We found that participants with high foreign language (FL) aptituderecalled and recognized more target words than participants with low FL aptitude. Also, the ability torehearse undisturbed, either vocally or subvocally, yielded the best results in recall and recognition ofthe foreign vocabulary across all levels of foreign language aptitude. The results suggest that studentswith L1 cognitive factors correlated with low FL aptitude should receive early training in phonologicalsensitivity and learning strategies to help them be successful language learners (de Jong, Seveke, & vanVeen, 2000).

Keywords: rehearsal; aptitude; inner speech; phonological awareness; working memory

FOREIGN LANGUAGE (FL) LEARNERS CANrehearse a word aloud either to themselves or aspart of a class drill, but they can also rehearseusing their inner speech, an approach that isoutwardly silent and is considered to be primarilyphonological as opposed to articulatory (seeMacKay, 1992). During the initial stages of FLlearning, learners may use inner speech to playwith funny sounding words or to rehearse wordsand utterances to be learned, perhaps evenhearing the teacher’s voice or an idealized nativeaccent in their heads (Guerrero, 2005); however,controversy exists concerning the role of rehears-

al for FL vocabulary learning and which type ofrehearsal is most effective (Beaton et al., 2005;Ellis & Beaton, 1993; Lawson & Hogben, 1996).Some FL educators and researchers recommendspeaking and practicing the target languagealoud, as soon as possible, for optimum recalland pronunciation practice (Pimsleur, 2007).However, others propose an extended listeningperiod, with comprehension preceding produc-tion, to create a more authentic inner voice andultimately a more native-like accent (Asher, 1977;Postovsky, 1974, 1981).

Although rehearsal seems to be an importantstrategy for most FL learners, whether vocalizedrepetition is superior to silent practice is notpresently known. The current study examinedthis question, focusing on the effects of differenttypes of rehearsal on foreign vocabulary learningin relation to learners’ FL aptitude, as measured

The Modern Language Journal, 97, 4, (2013)DOI: 10.1111/j.1540-4781.2013.12045.x0026-7902/13/902–916 $1.50/0© 2013 The Modern Language Journal

by the Modern Language Aptitude Test (MLAT;Carroll & Sapon, 1959). In the next section, wereview evidence regarding the role of phonologi-cal awareness and rehearsal in working memory,relate these features to the construct of FLaptitude as measured by the MLAT, and exploreresearch findings that explicate their link toaspects of FL vocabulary learning.

WORKING MEMORY AND FOREIGNLANGUAGE APTITUDE

Phonological Awareness and Rehearsal as Part ofWorking Memory

It is commonly assumed that all typicallydeveloping individuals can learn a foreign lan-guage.However, the amount of time required andthe best teaching method or learning environ-ment may differ from person to person. Severalresearchers (e.g., Miyake & Friedman, 1998; Wen& Skehan, 2011) believe that such individualdifferences in FL learning ability, particularlywhen they are related to the construct of FLaptitude, may be accounted for by differences inworking memory, a temporary storage system thatrelies on rehearsal in the phonological loop tomove information into short-term memory (Bad-deley, 2003).

To explain how this process might workBaddeley and others (e.g., Baddeley, 1990,1996, 2003; Baddeley, Gathercole, & Papagno,1998; Baddeley & Logie, 1992) posit a phonologi-cal loop, which is comprised of a phonologicalstore that temporarily holds phonologically en-coded information and a rehearsal process torefresh and maintain this information in thephonological store. Specifically, a word is held inthe phonological store and repeated via rehearsal(either vocal or subvocal) long enough for thenew word to be phonologically encoded in short-termmemory. While auditory input automaticallygains access to the phonological loop, non-speechinputs such as printed words, pictures, andthoughts are entered into the phonologicalloop by the individual saying the word aloud orby thinking the word silently as part of innerspeech.1 Once entered into the phonologicalloop, the phonological representation of theinput is believed to decay after about two secondsif it is not rehearsed (Gathercole & Thorn, 1998).Therefore, preventing word learners from re-hearsing inhibits word learning (Ellis & Sinclair,1996; Papagno, Valentine, & Baddeley, 1991).This inhibiting effect has been shown in studieswhere participants tried to learn a new word while

performing a competing task such as repeating‘bla’ aloud to a metronome, counting aloud from1 to 5, or tapping their fingers. The morecognitively taxing the task, the more disruptiveit is to rehearsal and thus to the future recall of thewords.

There are also item-specific factors that caninfluence word learning such as word length,phonemic complexity, foreignness, or distinc-tiveness of the item, such as time or mode ofpresentation, unusual sound, rich semanticknowledge, or word class as compared to thewords around it (Hulme et al., 2004; Hulme et al.,2006). A well-documented item-specific factor isword length. Word length effects mean thatlonger words, being harder to apprehend andtaking longer to rehearse in the phonologicalloop, are more difficult to recall than shorterwords. Baddeley, Thomson, and Buchanan(1975) showed that recall for a string of fivewords went from 90% to 50% when the list ofwords was changed from 5 one-syllable to 5 five-syllable words. The decrease in recall could bebecause it is easier to form a phonologicalrepresentation of shorter words and they can berehearsed more times in the phonological loopbefore they decay or due to forgetting the longerwords during recall. Baddeley (1996) states thatthe processes of rehearsal and response produc-tion are assumed to happen in real time.Therefore, longer words take longer to articulate(both subvocally as part of rehearsal and vocallyduring recall).

A reverse word length effect also exists: Tehanand Tolan (2007) replicated Hendry and Tehan’s(2005, as cited inTehan&Tolan, 2007)finding onreverse word length effects, indicating that longwords are recognized better than short words onitem recognition tasks. Service (1998) suggestedthat word length effects result from phonologicalcomplexity and not from syllable number orarticulatory duration, that is, the extra cognitiveattention required to encode the number andquality of phonemes places an added burden onthe phonological loop mechanism. She demon-strated that this burden can be lessened when thepseudoword to be recalled is rated high in wordlikeness or when additional semantic cues areprovided by the word’s similarity to a known word.She explained that longer words have fewerphonological neighbors, making them easier torecognize, even when the representation of theword has decayed in the phonological store,because by virtue of having more parts and beingmore distinct they can bemore accurately guessedeven when parts are missing from memory.

Kristina Dahlen and Catherine Caldwell–Harris 903

Especially important for the present research ishow the ability to form an accurate phonologicalrepresentation of a novel word influences effec-tive rehearsal of the word and eventual recall. Thisability is a function of phonological awareness, anindividual’s metalinguistic understanding of thesound structure of language, for example, under-standing that a sentence can be divided intowords, a word can be divided into syllables, andsyllables can be divided into onsets and rimes (thenucleus and coda; Rubba, 2003). Phonologicalawareness is initially acquired through exposureto language via listening and does not depend onthe kind of sound–letter correspondence (pho-nemic awareness) that children develop as part ofreading. Even so, poor phonological awareness iscorrelated with native language (L1) readingfailure (Rubba, 2003). Several researchers (e.g.,de Jong, Seveke, & van Veen, 2000; Hu & Schuele,2005) have found that, although learners withpoor L1 phonological awareness may also havepoor short-term verbal memory, when short-termverbal memory is statistically controlled for, it ispoor L1 phonological awareness that adverselyaffects the acquisition of phonologically unfamil-iar words. It is thus generally accepted that poorphonological awareness skills in the L1 are carriedover to the task of learning the L2, resulting in FLlearning difficulties.

Foreign Language Aptitude

In linking the construct of working memory toFL aptitude, FL aptitude must be distinguishedfrom general intelligence; two individuals withthe same IQmay differ significantly regarding theease with which they learn a FL (Stansfield, 1989).In a theory called the Linguistic Coding DeficitHypothesis (LCDH), Sparks and Ganschow(1993) claim that L1 phonological deficits are anoteworthy cause of FL learning difficulties. Theyposit that students with average or superior IQwho struggle to learn a FL have a specificphonological coding deficit that is responsiblefor difficulties in both L1 reading and FLlearning. High school and university studentsnot labeled as learning disabled, but who werefailing or having difficulty with their FL course,were found to have weak L1 phonological skillsand poor FL aptitude as measured by the MLAT(Sparks & Ganschow, 1993).

The MLAT is a standard instrument formeasuring cognitive abilities that are consideredpivotal to learning a FL in a classroom setting. Thetest assesses phonetic coding ability, grammatical

sensitivity, and rote memory and has correlationcoefficients between .40 and .65 with L2 achieve-ment (Carroll, 1962, 1963, 1966, as cited inWinke, 2013). While the MLAT does not measureintelligence, use of learning strategies, or affectivevariables such as motivation or anxiety, it is highlypredictive of FL learning difficulties, as well as ofpresumed learner success. For example, a score ator below the 20th percentile on the MLAT hasbeen used by schools to identify learners with “ahistory of dyslexia or auditory processing prob-lems” (Carroll & Sapon, 2002, p. 22). MLATscores have also been used to predict anindividual’s likely success in a FL class (Grigor-enko, Sternberg, & Ehrman, 2000). A notableexample for such use is that the U.S. andCanadian governments and some missionaryprograms use the MLAT to assess individualsand assign them to learn those languages that aredeemed to be appropriate for their aptitude level,where the languages themselves are categorizedaccording to four difficulty levels (Stansfield,1989). The MLAT was also used by Grigorenkoet al. (2000) to validate their theory and testinginstrument called the “Cognitive Ability forNovelty in Acquisition of Language as Appliedto Foreign Language Test (CANAL–FT).” Inother words, the MLAT has long been viewed asthe benchmark of FL aptitude tests (Grigorenkoet al., 2000).

Initial FL learning involves being sensitive tounfamiliar sounds, occurring in unfamiliar pat-terns and with different stresses and syllableconfigurations (Hu, 2003). French and O’Brien(2008) found that phonological working memorycapacity does not change over time, regardlessof increasing L2 proficiency or learning contexts.Phonological awareness, however, which facili-tates effective rehearsal necessary for wordlearning, especially at the beginning level, canbe improved with training (e.g., French &O’Brien, 2008; Sparks & Ganschow, 1993; Stano-vich, 1986; Troia, 1999).

A method based on the Orton–Gillinghamapproach (Bradley, Danielson, &Hallahan, 2002)and used for teaching FLs is called multisensorystructured learning (MSL). It combines differentsensory learning approaches (e.g., auditory,visual, tactile), provides opportunities for fre-quent practice and review, makes connectionsbetween old and new information, provides directinstruction on grapheme–phoneme correspond-ences of the language, and shows students how tobreak words down and put them back together(Ganschow & Schneider, 2002). Sparks et al.(1998) compared the effects of multisensory

904 The Modern Language Journal 97 (2013)

structured learning (MSL) to traditional text-book-based instruction in a high school setting.L1 speakers of English received Spanish instruc-tion for two years and were compared on ultimateSpanish proficiency, improvement in nativelanguage literacy skills, and FL aptitude (pre-versus posttest MLAT). It was found that themultisensory structured learning group achievedhigher Spanish proficiency than did comparisongroups. The training in the grapheme–phonemecorrespondence of Spanish improved learners’reading, writing, speaking, and listening skills. Allof the students made gains in L1 literacymeasures.

Although FL aptitude at the beginning level iswell measured by the MLAT, recent work on theconcept of FL aptitude has expanded on thepurely cognitive traits that have been used topredict the ease and speed with which someonecan learn a second language. Winke (2013)explored the plausibility of a structural equationmodel of FL aptitude to include cognitive factorsof rote memory, phonemic coding ability, gram-matical sensitivity (assessed by the MLAT), andphonological working memory, using a verbalworking memory span test. Along with this shemeasured language learning strategy use andaffective factors such as motivation. She conclud-ed that, although the MLAT predicts languagelearning success at lower levels, the role ofaptitude and rote memory in particular lessensas the learner becomes more proficient. Withgrowing L2 proficiency, the learner relies less onthe basic cognitive skills associated with FLaptitude for continued progress and interactsmore with the social environment of the L2,drawing on a different set of incidentally learnedskills and abilities to move to the advanced level(Robinson, 2013).

SITUATING AND MOTIVATING THE STUDY

Research interest regarding word learning hasidentified as a particularly important distinctionthe nature of learners’ rehearsal of vocabularyitems. Thus, Ellis and Beaton (1993) compareddifferent instructional strategies for FL vocabu-lary learning on the part of L1 English universitystudents with no knowledge of German. At theoutset, students viewed a written English wordnext to its German translation while hearing theGerman word pronounced. Students then re-hearsed the word pairs according to theirrandomly assigned learning strategy group: (a)vocal rehearsal, in which participants repeated theEnglish–German word pairs aloud until the next

pair of words was presented; (b) keyword technique,in which participants were instructed to imagine aparticular scene that visually linked the sound ofthe English word to the sound of the Germanword; or (c) learners’ own strategy, in whichparticipants were told to do their best to learnthe word pairs, but were not told how. Thestudents were then tested for recall in two ways:first, by typing the English translation for thewritten German word while also hearing it spoken(receptive) and, second, by typing the Germanword when they saw the written English word(productive). A month after completing theinitial learning task, participants were again testedon their recall from both English to German andGerman to English and were asked to indicatehow they had managed to remember the words(chosen from a list of prompts).

The study found that participants’MLAT scorewas predictive of their ability to learn foreignwords; productive German recall was moredifficult than receptive English recall; and nounswere rememberedmore often than verbs. The twobest learning strategies were vocal rehearsal (71%correct), and ‘own strategy’ (62% correct).Participants in the vocal rehearsal conditionoutscored the other two groups in the productiveGerman condition, whereas the keyword/visualimagery technique worked best when they had totype the English translation while reading andhearing the German word. In their discussion, theresearchers wrote, “it appears that the repetitiongroup benefited from hearing themselves repeatthe words to be learned and, possibly, from thearticulatory representations established throughrepetition” (Ellis & Beaton, 1993, p. 552). It is notsurprising that those in the vocal learningcondition were able to recall the most Germanvocabulary words, since they could experiencefour learning modalities: rehearsal, vocalization,auditory feedback from their own verbal output,and visualization (if they also used a visualstrategy). However, in order to determine theprecise function of vocal rehearsal, these contrib-uting aspects need to be disaggregated.

In a follow-up study, Ellis and Sinclair (1996)began to tackle this task by explicitly investigating,among other issues, whether beginning FLlearners should be encouraged to vocally repeatand rehearse the target language or whether theyshould observe a silent listening period in thebeginning. This time L1 English university stu-dents rehearsed Welsh words as FL vocabularyitems. Participants were randomly assigned tothree conditions: (a) silent, (b) vocal rehearsal of theWelsh utterances whenever they heard the words


played on the computer loudspeaker, or (c)articulatory suppression, a condition under whichparticipants whispered the numbers one to fivecontinuously as they listened to the Welsh utter-ances. After the rehearsal condition, the partic-ipants completed a well-formedness test, a ruletest, and a speech production test. Those who hadvocally rehearsed the words and phrases scoredbetter in all areas, except receptive grammaticalityjudgments. Members of the silent group scoredbetter than those using articulatory suppression.They concluded that vocal rehearsal promotesretention and that in hearing and repeating thewords and phrases, learners inferred and acquiredknowledge about the language above and beyondthe particular utterances they were repeating.

A drawback of Ellis and Sinclair’s (1996) study isthat the participants in the silent group wereneither instructed to rehearse subvocally norprevented from rehearsing subvocally using artic-ulatory suppression. As a consequence, theirinternal rehearsal state is unknown. A second issueis that learners in the vocally rehearsing groupcould have benefited from the output of their ownverbal articulation, or from the input of hearinghow well their own pronunciationmatched that ofthe Welsh speaker on the computer. In otherwords, while research seems to indicate that vocalrehearsal is more effective than not rehearsing atall (articulatory suppression), the precise cause ofthat effectiveness remains unclear (Ellis & Beaton,1993; Ellis & Sinclair, 1996).

The current study addressed these issues in twoways. First, we split Ellis and Beaton’s (1993)vocal group into two separate groups, one inwhich participants were able to hear their ownauditory feedback and the other with auditoryfeedback masked by white noise. This modifica-tion separated the effect of vocal rehearsal fromthe effect of hearing oneself speak. Furthermore,the silent group was given specific instructions torehearse subvocally. In modifying the previouslydescribed modes of rehearsal, our goal was todetermine which rehearsal modalities contributemost to the recall and recognition of foreignvocabulary words. A second research question waswhether FL aptitude, as measured by the MLAT,would have an effect on the recall and recognitionof newly learned words.

THE STUDY

Overview of the Study

Each research participant was tested separatelyin a laboratory room, taking about two hours for

the whole experiment (one hour for the MLATand one hour for the Turkish vocabulary learningportion). Participants completed a brief question-naire on language knowledge andmotivation (seeAppendix A) before beginning the study, whichproceeded in four stages:

1. Each participant took the timed MLAT test.2. Participants were led through the rehearsal

of the Turkish vocabulary according to theirrandomly assigned rehearsal condition. Par-ticipants viewed a picture of a concrete nounwhile hearing the Turkish word for thatnoun repeated 3 times through their head-phones. They then repeated the Turkishword or performed a control task for 20seconds.

3. Participants took a recall test where theywere tested for verbal recall via a productivepicture naming task.

4. In a final recognition task, participantsheard each Turkish word embedded in anovel aurally presented Turkish sentenceand indicated on a picture grid which wordthey thought they heard.

Participants

The 88 participants (60% female, 40% male;mean age 19) were undergraduates in anintroductory psychology class at BostonUniversitywho received course credit for taking part in theexperiment. None of the students accepted forthe study had previous experience with learningTurkish. The majority were native English speak-ers (N! 56) who reported never having mastereda second language; a sizable number werebilingual from birth (N! 25), and 7 reportedhaving learned a second language to the pointwhere it could be used in daily interactions,including 2 international students (nonnativespeakers of English).

Questionnaire

Participants completed a language knowledgequestionnaire containing items concerning theirage, sex, past experience with FLs, self-assessed FLability, anxiety regarding language learning, andmotivation to learn a FL (see Appendix).

MLAT

Participants spent an hour taking the MLATusing the CD, a test booklet, and an answer sheet.In Part I: Number Learning (aural) the individual


learns Kurdish numbers and then listens to andtranscribes Kurdish numbers of up to three digits.This tests auditory alertness and phonologicalmemory span. In Part II: Phonetic Script (aural)the individual hears a word and must circle thephonetic transcription that represents the word asa measure of the person’s sound–symbol correla-tion or phonemic awareness. In Part III: SpellingClues, the individual first reads and decodes anEnglish word that is spelled as it sounds and thenmust select a synonym for that word. This testssound–symbol correlation in written form. In PartIV: Words in Sentences, the individual reads anunderlined word or phrase in a sentence andmust find the word or phrase in another sentencethat serves the same grammatical function. Thismeasures grammatical sensitivity without relyingon grammatical terminology. In the last part, PartV: Paired Associates, the individual has a fewminutes to study a list of Kurdishwords and is thengiven a multiple-choice test on the meaning ofthose words. This is a test of rote memorization.Four of the five MLAT sections, excluding thesection on grammatical sensitivity, measure anindividual’s phonological awareness, phonologi-cal memory span, and phonemic awareness.These four sections are thus the most plausiblyrelated to the learning task used in the currentstudy to examine whether vocalized repetition issuperior to silent inner speech in learning foreignvocabulary words. However, we used MLAT meanpercentile scores based on all 5 sections as ourmeasure of FL learning aptitude, since theseholistic scores were used to predict vocabularylearning in previous similar studies.

Stimuli

The Turkish stimuli consisted of 20 concretenouns that varied in length from 1 to 4 syllables(see Table 1). All nouns were comprised ofphonemes that exist in the English language. The

spoken Turkish words were paired with picturesfrom Rossion and Pourtois’ (2004) coloredline drawings, which added color and texture tothe pictures used by Snodgrass and Vanderwart(1980) in order to improve object recognition.All Turkish stimuli were spoken by a male nativeTurkish speaker and were recorded using a Maccomputer on Audacity 1.2.5 software. Theaudio and visual stimuli were presented usingPsyScope experimental software (Cohen et al.,1993). The stimuli were clear and understandablebased on the judgment of another native Turkishspeaker.

Vocabulary Rehearsal Procedure

Participants were randomly assigned to one ofthe four vocabulary rehearsal conditions. Vocab-ulary presentation began with the touch of acomputer key, whereupon participants saw thepicture of the target word on the computer screenwhile hearing the Turkish word spoken 3 times.This was followed by 20 seconds of silence, duringwhich the participants rehearsed according totheir randomly assigned rehearsal condition. All20 Turkish words were presented in randomorder twice, with each presentation followed by a20-second silence for the rehearsal protocol.

Vocal Plus Auditory Feedback. Participants spokethe words aloud during the 20-second rehearsalwindow and were able to hear their own voicesrepeating the Turkish word.

Vocal With White Noise (meaning no or minimalauditory feedback). Participants heard white noisethrough the earphones during the 20 secondsthey were repeating the word aloud and could nothear how accurate or inaccurate their pronuncia-tion was. This method allowed for assessment ofthe effects of vocal rehearsal on FL learningseparate from the effects of auditory feedback.

TABLE 1Twenty Turkish Stimuli and Five Foil Words With English Gloss

One Syllable Two Syllables Three Syllables Four Syllables

at ‘horse’ catal ‘fork’ anahtar ‘key’ ayakkabı ‘shoe’kol ‘arm’ kopek ‘dog’ araba ‘car’ buzdolabı ‘refrigerator’kus ‘bird’ sapka ‘hat’ eldiven ‘glove’ kursun kalem ‘pencil’muz ‘banana’ ucak ‘airplane’ portakal ‘orange’ yerfıstı!g ı ‘peanut’top ‘ball’ yuzuk ‘ring’ sandalye ‘chair’ "aycice!g i ‘sunflower’"fil ‘elephant’ tencere ‘saucepan’ "bilgisayar ‘computer’"mum ‘candle’ "dondurma ‘ice cream’

Note. "indicates foil.


Subvocal Rehearsal. Participants were in-structed to repeat the word in their heads,subvocally, for the full 20 seconds. They obviouslydid not receive auditory feedback, but presumablyheard the rehearsal words in their heads, as partof inner speech.

Articulatory Suppression. Participants readaloud from a few short, stimulating magazinearticles during the 20-second rehearsal windowbetween the presentation of Turkish words. Thisprotocol was developed following a pilot study forarticulatory suppression (N! 13) in which par-ticipants, just as in the Ellis & Sinclair (1996)study, counted aloud from 1 to 5 during the 20-second silence to prevent rehearsal. The pilotresults showed that the ability to recall andrecognize the Turkish words was not impairedby this procedure. When debriefed, our pilotparticipants said that they had counted andrehearsed at the same time or that they hadcounted to 5, repeated the target word in theirheads, and then counted to 5 aloud again. Thus,reading aloud was selected to create an activitywith higher cognitive load to prevent vocal orsubvocal rehearsal.

The 4 rehearsal conditions are summarized inTable 2 along with the modalities that theparticipants in each group experienced: overtarticulation (with and without auditory feed-back), subvocal rehearsal, and articulatory sup-pression. Due to equipment failure, the auditorydata from 11 of the Subvocal participants could

not be coded. Their data appears in the Recogni-tion mean, but not in the Recall mean.

Recall Procedure and Scoring

Following the vocabulary rehearsal task, partic-ipants immediately completed a recall test thatassessed their ability to recall and accuratelypronounce the 20 Turkish words that had beenpresented during the rehearsal phase. Each of the20 pictures appeared on a computer screen one ata time and in random order. The participantswere instructed to say the Turkish word for thatpicture. They were told that they could take asmuch time as they needed and that they couldgive partial answers or say whatever they remem-bered about the word, e.g., if it was long, short,started with a /k/, etc. The recall procedurecontinued in this fashion for all 20 pictures.Responses were recorded using a digital voicerecorder and a high quality stereo headset with abuilt-in microphone.

The first author and a native Turkish speaker(not the same person who had recorded thestimuli) coded the participants’ Turkish recallresponses for correctness according to Service’s(1998) method. Each syllable could receive 3points, 1 point each for the onset, nucleus, andcoda (Ladefoged, 1993). A one-syllable wordsuch as muz “banana” received a point each for/m/, /u/, /z/, but only two points for /buz/ or/miz/ where one phoneme was incorrect. A one-

TABLE 2Vocabulary Learning Conditions and Modalities Experienced

Manner of Rehearsal Modalities ExperiencedNumber of

Modalities Experienced

Vocal: N! 22Vocal rehearsal plus auditory feedback

# articulation 3# rehearsal# auditory# feedback

White Noise: N! 23Vocal rehearsal minus auditory feedback

# articulation 2# rehearsal$ auditory$ feedback

Subvocal Rehearsal: N! 23Silent rehearsal using inner speech

$ articulation 1# rehearsal$ auditory$ feedback

Articulatory Suppression: N! 22Reading aloud during the 20- secondsilence between presentation of stimuli

$ articulation 0$ rehearsal$ auditory$ feedback


syllable word such as at “horse” received a pointeach for the null onset, the nucleus, and the coda(2 points if a phoneme was added, deleted, orreplaced). We modified Service’s coding systemby counting null phonemes as one point,subtracting a point if a phoneme was insertedinto the null spot, but not distinguishing betweenshort vowels and diphthongs. Following thisprocedure, two-syllable words were worth 6points, three-syllable words were worth 9 points,and four-syllable words were worth 12 points.

The interrater reliability between the firstauthor and the native Turkish speaker was99.45%; such a high level of agreement isprobably attributable to a very clearly definedprotocol for rating each word. We totaled thepoints earned across the 20 words and divided bythe total points possible to yield the percentage ofsyllables correctly recalled. Then we averaged thescore given by the two raters to determine eachsubject’s total recall score. Across our sample of 88participants, total scores ranged from 1% ofsyllables correctly recalled to 85%, indicatingsubstantial variation in word learning. Since recallscores were a continuous variable, we had theoption of analyzing them with parametric statis-tics. However, a histogram of recall scores showedconsiderable skew with a long rightward tail. Thesample had a modal score of 10% of syllablescorrect, but 4 of the 88 participants recalled 70%or more of syllables. We thus analyzed the datawith nonparametric statistics. We used theKruskal–Wallace H when testing for a differencebetween more than two conditions, the Mann–Whitney U for comparing two groups, andSpearman’s rho for identifying a significantcorrelation between two continuous variables.

Recognition Procedure and Scoring

A recognition task was employed to measurewhether participants could recognize the Turkishwords that they had just practiced when the wordswere embedded in a Turkish sentence andwhether they could retrieve the meaning of theword from memory. This task corresponds to thecommonplace report of FL learners that theforeign vocabulary words they know seem to “popout” of sentences they hear, even if the sentencesare otherwise unintelligible to them. Neverthe-less, detecting a recently learned word in an entiresentence may be a more difficult task for someindividuals than identifying the target Turkishword in isolation. Individuals with poor phono-logical awareness organize their phonologicalrepresentations of words in chunks larger than

the phoneme level. This inefficient segmentingbecomes especially problematic when individualsare presented with foreign words with unfamiliarphonological structures (Hu, 2003). One mighttherefore expect participants with low MLATscores and poor phonological awareness to havemore difficulty recognizing the Turkish targetwords in a sentence than participants with highMLAT scores.

The novel sentences were written and recordedby the same native Turkish speaker who recordedthe test words (see Sentence Example 1 and 2).Twenty sentences contained a noun that had justbeen rehearsed and tested via recall. However,five sentences contained a noun that had notbeen taught during the rehearsal task; thesenouns are referred to as foils. All 25 sentenceswere of similar syllabic length (average 8.5syllables), regardless of whether the target nounwas short (one syllable) or long (four syllables).The target nouns appeared at the beginning,middle, or end of the Turkish sentence with equalfrequency.

SENTENCE EXAMPLE 1

ucak ‘airplane’

Ucak az sonra kalkacak.

The airplane will take off soon. (Target word at thebeginning)

SENTENCE EXAMPLE 2

kopek ‘dog’

Bu cok sakin bir kopek.

This dog is very calm. (Target word at the end)

Participants scanned a grid containing thepictures of the 20 nouns on the computer screenin front of them. Through the headphones, theyheard a Turkish sentence. Their task was to clickon the picture of the word they thought theyheard in the sentence, or to type the zero (“0”) keyon the keyboard if none of the target words waspresent. They were told that some sentences didnot contain a target vocabulary word but not howmany. Participants had as much time as theywanted to make a decision, allowing them time topotentially replay the sentence in their innerspeech. As soon as a participant clicked on apicture, or hit “0,” the next Turkish sentenceplayed. This continued for all 25 trials (20 targetwords and 5 foil words). Scores were the


percentage of target words the participantscorrectly matched with the picture of the noun.The foil items were correctly rejected on 85% oftrials, an accuracy rate that did not differ byrehearsal condition or MLAT score. For thatreason we did not further analyze the foil trials.Recognition total scores ranged from 0%of targetwords correctly recognized (2 subjects) to 70% oftarget words correctly recognized. Results werenormally distributed and thus parametric statis-tics were used in the analysis.

RESULTS

Preliminary Analyses of Language BackgroundVariables

We analyzed students’ language backgroundquestionnaire to determine if any self-reportvariables correlated with the MLAT or our twomeasures of vocabulary learning, the recall andrecognition scores.

Bilingual/Monolingual Status. Contrary to ex-pectations, MLAT scores (which ranged fromthe 15th to the 99th percentile in our sample)were higher for the 56 monolinguals (medianscore of 70) than for the 32 bilinguals (medianscore of 52), a statistically significant difference,t(86)! 2.3, p< .03; also significant at p< .03 usingthe nonparametric Kruskal–Wallis test. It isbeyond the scope of this paper to speculate onthe cause of the difference, other than to notethat the MLAT measures cognitive abilities thathelp learn a FL in a classroom setting and thusmay be most valid for monolingual Englishspeakers, the group that constituted the majority(63%) of our sample. Bilingual/monolingualstatus did not correlate with scores on the recallof recognition tasks; thus we will not consider thisbilingual status further.

Self-rated Language Learning Ability, Anxiety, andMotivation. Neither anxiety over learning a FLnor self-reported motivation to learn a languagewas correlated with MLAT scores or recall andrecognition totals. Self-reported language learn-ing ability did correlate weakly with total recallscores, rho! .25, p< .05, but not with recognitionscores, Spearman’s rho! .15, p> .05. MLATscores correlated with recognition scores, rho! .28, but not recall scores, which had rho! .19.Self-reported ability and MLAT were only weaklycorrelated, rho! .27 (Pearson’s r was also .27 forMLAT and recognition). Given that MLAT scoreshad slightly higher correlations than self-reportedability when averaged across our two measures of

vocabulary learning, and is a validated construct,we set aside the simple report of ability and usedMLAT scores in our analyses. Graphs plottingvocabulary learning as a function of MLATshowed that those participants in the top-thirdof the MLAT distribution had better vocabularylearning compared to those in the bottom third(as discussed and depicted in Figures 1 and 2).

Recall Scores by Rehearsal Condition

The participants in this study received a rawscore on the MLAT and were then given apercentile rating according to the normed MLATscales for male and female freshmen in college.The mean percentile score for all participants was64 and themedian score was 65, the rangewas fromthe 15th to the 99th percentile. Participants weredivided into three groups for descriptive purposes:Percentile scores from 1–45 were labeled Low,percentile scores from 50–80 were labeled Medi-um, and from 85–100 were labeled High.

Figure 1 depicts recall scores according torehearsal condition and MLAT group (Low,Medium, and High scores). As noted in thepreliminary analysis, participants with highMLATscores recalled more words than those with lowerMLAT scores.

An initial expectation was that the learninggroup experiencing the most modalities (thevocal group experienced 3: articulation, rehears-al, auditory feedback) would recall more Turkishwords than groups experiencing fewermodalities.The groups’ median syllable recall was 43% for

FIGURE 1Percent of Syllables Recalled by Rehearsal Conditionand MLAT


subvocal rehearsal, 40% for vocal rehearsal, 32%for those who rehearsed with white noise, and 9%for those who were required to read out loudrather than being allowed to rehearse (articulato-ry suppression group). These group mediansdiffered significantly, with the Kruskal–Wallis TestH! 33, p< .0001. Pair wise comparisons with theMann–Whitney U test were conducted to isolatestatistically significant differences between re-hearsal conditions. As is apparent from Figure1, the strongest difference among the conditionsoccurred between the very low number of syllablesrecalled in the articulatory suppression groupcompared to the other groups. The subvocalrehearsal condition differed significantly fromthe white noise condition, but not from the verbalrehearsal condition. The verbal rehearsal condi-tion did not differ significantly from the whitenoise condition. From these comparisons weconclude that the subvocal rehearsal conditionwas most conducive to high recall of syllables.

Recognition Scores by Rehearsal Conditions

Figure 2 illustrates how vocabulary learningas measured by recognition of the learnedwords in novel sentences varied by rehearsalcondition. Recognition scores were analyzed in abetween subjects ANOVA (four learning condi-tions with the MLAT score as a covariate). Themain effect of FL aptitude (actual MLAT percen-tile scores, not High/Medium/Low category) wasF (1, 72)! 10.13, p< .001, partial eta squared! .12; and the main effect of learning conditionwas F (3, 72)! 9.47, p< .001, partial eta squared

! .28; the interaction between aptitude andlearning condition was not significant, F< 1.Post hoc (Scheffe) analyses for learning conditionshowed that, although the vocal and subvocalmeans were not significantly different from eachother, they were different from articulatorysuppression (p< .001 and p< .002). The whitenoise condition was marginally different fromarticulatory suppression and the vocal conditions(p< .08), but not from the subvocal condition.

As shown in Figure 2, the vocal and subvocalmanner of rehearsal yielded the highest recogni-tion scores, followed by the white noise and thearticulatory suppression groups. The mean num-ber of words correctly recognized out of 20 was:vocal! 8.0; subvocal! 7.8; white noise! 6.0; andarticulatory suppression! 4.0. The vocal andsubvocal group means were virtually identical,but almost 20 percentage points higher than thearticulatory suppression group. This indicatesthat a disruption in rehearsal is detrimental tolearning FL vocabulary.

MLAT and Rehearsal Condition

In an exploratory analysis, we correlated MLATscores with recall and recognition totals separatelyfor each learning condition. This was done todetermine whether FL aptitude could have astronger influence on learning in some rehearsalconditions than others. In the vocal rehearsalcondition, MLAT correlated with recall, rho! 0.57, and with recognition, rho! .54. Correla-tions were weaker in the other conditions,ranging from rho! .21 to rho! .38, but thesewere not significant due to the small sample sizesand restricted range resulting from conductingcorrelations on a subgroup of participants. Anexplanation for why learning from vocal rehearsalwas more strongly influenced by FL learningaptitude (as measured by the MLAT) than theother rehearsal conditions could be sought infuture research.

Word Length Effects and Response Time

Although not the focus of this paper, data wascollected on response time and recall andrecognition by word, broken down by syllablelength (there were five one-, two-, three-, and four-syllable words). The correlation between Recallscore and Recognition score was strong, (r! .78,p< .01). The word length effect, meaning thatshorter words are more easily recalled and longerwords are more easily recognized, is a furtherindication of the role of rehearsal in word recall.

FIGURE 2Percent of Words Recognized by RehearsalCondition and MLAT


In this study, for each of the four rehearsalgroups, the most recalled class of words was theone syllable words, and the most recognizedwords were the three syllable words, as predictedby the literature. The mean recall of one-syllablewords was 59%, but only 34% for the meanrecognition. Only 24% of all participants wereable to correctly recall the three-syllable words,but 44% recognized them in the sentences. Thebest recalled words were one-syllable wordscomprised of a consonant, a vowel, and aconsonant (CVC); these were top ‘ball’ (recalledby 77.3% of participants), kus ‘bird’ (66.8%), kol‘arm’ (63.3%) and muz ‘banana’ (57.7%). Themost recognized words were sandalye ‘chair’(73.0%), tenjere ‘saucepan’ (67.0%), kus ‘bird’(61.6%), portakal ‘orange’ (55.6%), top ‘ball’(50.5%) and muz ‘banana’ (46.9%). These werea combination of one-syllable words (all CVC) andthree-syllable words,manywith interesting soundswith possibilities for making mental connectionsto English. The four-syllable words did not benefitfrom reverse word length effects, perhaps becauseparticipants were not able to form a phonologicalrepresentation of them, and thus could notrehearse them effectively, thereby compromisingtheir ability to recall and recognize them.

Although participants were not told to respondquickly, it is nevertheless interesting to analyzeresponse times for the recognition task given thatlatency to click on a response was automaticallyrecorded by the experimental control software.Deleting 14 outlying data points below 3 secondsand 1 point above 40 seconds, 1102 total datapoints were analyzed for reaction time. TheSubvocal group’s reaction time, from sentenceonset to response, was the fastest (11.6 s) com-pared to the Articulatory Suppression group(12.1 s), the White Noise group (12.6 s), and theVocal group (13.8 s). The Vocal group had thelongest reaction time. Using a one-way ANOVA,reaction time was significant, F (3, 84)! 4.02,p< .01. It is interesting to note that the Vocal andWhite Noise groups, the only two groups thatrehearsed the words aloud, had the two slowestreaction times. Pair-wise comparison between theSubvocal and Vocal groups (the twomost extrememeans) were highly significant, F (3, 84)! 10.2,p< .002.

DISCUSSION AND IMPLICATIONS

This study looked at the role of different typesof rehearsal (vocal, vocal with auditory feedbackmasked by white noise, subvocal, and no rehears-al/articulatory suppression) on the recall and

recognition of Turkish concrete nouns. A secondquestion was whether the FL aptitude of theparticipants would have predictive value on theirvocabulary learning. Phonological awareness, askill developed in the L1 and carried over to L2learning, was highlighted as a key prerequisite foreffective rehearsal in the phonological loop aspart of working memory. In the next section, wewill discuss some of the most interesting resultsand suggest directions for future research as wellas some implications for differentiated instruc-tion in the FL classroom. Because a great deal ofperipheral data was collected during the recalland recognition tests, we will also touch briefly onsignificant findings regarding response time andword length effects.

Rehearsal and Aptitude

This study addressed how short-term recall andrecognition of foreign vocabulary words would beinfluenced by type of rehearsal, and secondly, byFL learning aptitude (as assessed by the MLAT).We found that participants who were able torehearse the Turkish vocabulary had better recalland recognition than participants who wereprevented from rehearsing. The white noisecondition was employed to distinguish whetherthe facilitative effect of vocal rehearsal onvocabulary learning (Ellis & Sinclair, 1996; Service&Kohonen, 1995) was due to overt articulation orto participants’ ability to hear their vocal produc-tion based on their own auditory feedback. Eventhough the white noise group spoke the Turkishwords aloud, their performance was inferior tothat of the subvocal group (with no auditoryfeedback beyond the feedback in their innerspeech). We infer that the white noise may haveinterfered with participants’ ability to recall andrecognize words by diminishing the positive effectof the vocal rehearsal by disrupting the partic-ipants’ ability to hear themselves think in theirinner speech. The rehearsal and self-regulatingfunctions of inner speech (Guerrero, 2005) weredisrupted by the white noise in a way similar to thedifficulty one might experience while studying ata loud party or multitasking. Trying to tune outthe white noise and focus on the task may havecaused participants to divide their cognitiveresources, leaving them less able to focus onmeaningful rehearsal. The poor performance ofthe white noise group suggests the crucial role ofundisturbed rehearsal (either vocal or subvocal)in recalling and recognizing new words in a FL.

The ability of the subvocal group to recall andrecognize Turkish words better than or equal to


the vocal group was surprising given that overtvocalization offers two modalities in whichlearning can occur: rehearsing and hearingone’s own utterances. It should thus have beensuperior to subvocal rehearsal. The good vocabu-lary learning scores of those who rehearsedsubvocally suggests that vocal rehearsal was notnecessary for recall and recognition of the testvocabulary. It was not overt vocalization thatenabled participants to correctly pronounce anddetect FL vocabulary in novel sentences, but theability to rehearse without disruption of the innerspeech.

We propose the following explanation for whysubvocal rehearsal is a powerful learning mecha-nism: Silent rehearsal in inner speech is fasterthan vocal articulation, allowing the test words tobe rehearsedmore times in the phonological loopthan is possible with slower vocal rehearsal.Although not addressed in this study, we hypoth-esize that for some learners, overt vocalizationmay actually detract from learning by dividingtheir attention between the inner processingfunction of rehearsal and the performance aspectof overt pronunciation. Thus, children forced toread aloud in class may understand little of whatthey are reading because limited cognitiveresources are focused on both the decoding oftext and the output of reading aloud, a phenom-enon known as word calling (Stanovich, 1986).

The study revealed the powerful effect ofrehearsal and language learning aptitude inrecalling and recognizing foreign words. Thosewith high FL learning aptitude generally learnedmore than those with lower MLAT scores. Thehigh aptitude participants are distinguishablefrom the low aptitude participants by learningalmost twenty percent more Turkish vocabulary(see Figures 1 and 2), suggesting that FL aptitudeis predictive of FL vocabulary recall and recogni-tion. However, due to the powerful effect ofrehearsal, the low MLAT scorers in the vocal andsubvocal groups outscored the high aptitudeparticipants in the articulatory suppressiongroup, who were prevented from rehearsing.We found this by isolating what appears to be a keyfactor in vocal and subvocal rehearsal: undis-turbed rehearsal. Separating Ellis & Sinclair’s(1996) vocal rehearsal group into vocal rehearsalplus auditory feedback and vocal rehearsal minusauditory feedback distinguished the benefit ofvocal rehearsal from the additional benefit ofbeing able to hear oneself rehearse. Undisturbedvocal and subvocal rehearsal improves scores onboth recall and recognition tasks. Manner ofrehearsal showed a stronger effect than FL

aptitude, highlighting the importance of undis-turbed rehearsal, either vocal or subvocal, to FLvocabulary learning.

A possible follow-up experiment would be tocompare the results of subvocal and vocalrehearsal using a target language that containedphonemes nonexistent in the participants’ L1. Itwould be interesting to see if the current resultshold for Chinese, a language with greaterphonemic linguistic distance that relies ontones to supply meaning. Would L2 learners ofChinese need to articulate Chinese tones vocallyto pronounce them correctly, recall them, or evenrecognize them, or would undisturbed subvocalrehearsal produce similar results? A secondquestion would be whether the greater complexi-ty of the phonemes combined with the toneswould exacerbate the achievement gap betweenthose with high versus low aptitude.

Implications for Teaching

FL learners with high FL aptitudemay seem likenaturals in the FL classroom, a phenomenon thatis noticed by teachers and fellow students alike.These students seem to recall new words afterseeing or hearing themonly once, they seem to beable to recognize and pronounce new words thatthey see in written form, and they are able toextract meaning from what the teacher says in thetarget language.

This study affirms the claimed existence ofindividual differences in FL aptitude. Participantswith high FL aptitude (MLAT scores) performeddifferently from those with low aptitude on eachdependentmeasure (Recall andRecognition). Allof the participants in the current study appearedto try their best to learn the twenty Turkish wordsand to succeed at every task we presented to them.However, given their different levels of FLaptitude and the powerful effect of the rehearsalconditions they were assigned to, they were not allable to successfully recall and recognize theTurkish vocabulary words.

When debriefed after the experiment, manyparticipants were apologetic or embarrassed attheir lack of success on these tasks. Their feelingof inadequacy may have been heightened by theincreasing focus on foreign language learning inthe U.S., with many high schools now requiring atleast two years of FL study for graduation and agrowing national awareness of the need to speak asecond language. In order to meet these newexpectations, lower aptitude learners must beidentified and given the support they need inorder to learn a FL, meet graduation


requirements, and become global citizens alongwith their higher aptitude peers.

These individuals could receive specific pho-nological awareness training to help them per-form closer to the level of their higher MLATscoring, but perhaps equally intelligent, peers.Sparks and Ganschow (1993) recommend directand explicit teaching of L1 phonology in theprimary grades and before L2 instruction beginsas the skills needed for phonological awarenessare transferable from the L1 to the L2. Programssuch as multisensory structured language can beused to teach the phonology of the FL, whereasinstruction and interaction using comprehensi-ble input in the target language (Krashen, 1985)should be prioritized to give students the soundsandmeaning they need to play with in their headsas part of inner speech. Research from the studiesof phonology, L1 reading, brain research, and FLinstruction should be pooled and shared to helpdesign and implement programs for low aptitudelearners, seeing aptitude as a combination ofcognitive and affective factors. These programswould no doubt make the FL learning experienceboth more productive and more pleasant forstudents of all aptitude levels and help alleviatethe FL dread that many students come to feel.

ACKNOWLEDGMENTS

We thank Shanley Allen, Catherine O’Connor, andMarnie Reid who reviewed the doctoral dissertation,Aptitude, Rehearsal, And Skin Conductance Response InForeign Vocabulary Learning (2008), from which this datacomes. We thank Mara Anderson, Kelly Hartman, KarlLabarbara, and Megan Peppard for coordinating theresearch participants and John Lymberis and OktaySahin for assistance in coding data and statisticalanalysis. We are grateful to Ulas Kaplan who helpedchoose the Turkish vocabulary, write the sentences, andwho lent his voice to the experiment. Thanks finally tothe input of the three anonymous MLJ reviewers and toEditor-in-Chief Heidi Byrnes for their suggestions forrevisions, which made the article clearer and stronger.

NOTE

1 Although there is a wide range of possible functionsof inner speech, the working memory model proposedby Baddeley deals only with the rehearsal aspect of innerspeech. This article will treat subvocal or mentalrehearsal as a rudimentary form of inner speech usedin vocabulary learning (Guerrero, 2005), which does not

represent the full range of functions of fully developedand internalized L2 inner speech.

REFERENCES

Asher, J. J. (1977). Learning another language throughactions. Los Gatos, CA: Sky Oaks Productions.

Baddeley, A. (1990). Human memory: Theory and practice.Boston: Allyn & Bacon.

Baddeley, A. (1996). The fractionation of workingmemory. Proceedings of the National Academy ofSciences of the United States of America, 93, 13468–13472. Retrieved 27 July, 2013 from http://www.ncbi.nlm.nih.gov/pmc/articles/PMC33632/pdf/pq013468.pdf

Baddeley, A. (2003). Working memory and language:An overview. Journal of Communication Disorders, 36,189–208.

Baddeley, A., & Logie, R. (1992). Auditory imagery andworking memory. In D. Reisberg (Ed.), Auditoryimagery (pp. 179–197). Mahwah, NJ: LawrenceErlbaum.

Baddeley, A., Gathercole, S., & Papagno, C. (1998). Thephonological loop as a language learning device.Psychological Review, 105, 158–173.

Baddeley, A. D., Thomson, N., & Buchanan, M. (1975).Word length and the structure of short-termmemory. Journal of Verbal Learning and VerbalBehavior, 14, 575–589.

Beaton, A. A., Gruneberg, M. M., Hyde, C., Shufflebot-tom, A., & Sykes, R. N. (2005). Facilitation ofreceptive and productive foreign vocabulary acqui-sition using the keyword method: The role ofimage quality. Memory, 13, 458–471.

Bradley, R., Danielson, L. C., & Hallahan, D. (2002).Learning disabilities: Historical perspectives. In R.Bradley, L. Danielson, & D. P. Hallahan (Eds.),Identification of learning disabilities: Research topractice (pp. 1–65). Mahwah, NJ: LawrenceErlbaum.

Carroll, J. B., & Sapon, S. M. (1959). Modern LanguageAptitude Test. Rockville, MD: Second LanguageTesting.

Carroll, J. B., & Sapon, S. M. (2002). Modern LanguageAptitude Test Manual. N. Bethesda, MD: SecondLanguage Testing.

Cohen, J. D., MacWhinney, B., Flatt, M., & Provost, J.(1993). PsyScope: A new graphic interactiveenvironment for designing psychology experi-ments. Behavioral Research Methods, Instruments, &Computer, 25, 257–271.

de Jong, P. F., Seveke, M. J., & van Veen, M. (2000).Phonological sensitivity and the acquisition of newwords in children. Journal of Experimental ChildPsychology, 76, 275–301.

Ellis, N., & Beaton, A. (1993). Factors affecting thelearning of foreign language vocabulary: Imagerykeyword mediators and phonological short-termmemory.Quarterly Journal of Experimental Psychology,46A, 533–558.


Ellis, N., & Sinclair, S. G. (1996). Working memory inthe acquisition of vocabulary and syntax: Puttinglanguage in good order. Quarterly Journal ofExperimental Psychology, 49A, 234–250.

French, L., & O’Brien, I. (2008). Phonological memoryand children’s second language grammar learn-ing. Applied Psycholinguistics, 29, 463–487.

Ganschow, L., & Schneider, E. (2002). At-risk studentsand the study of a foreign language in school.International Dyslexia Association. Retrieved 29July, 2013 from www.ldonline.org/article/19281

Gathercole, S. E., &Thorn, A. S. C. (1998). Phonologicalshort-term memory and foreign language learn-ing. In A. F. Healy & L. E. Bourne (Eds.), Foreignlanguage and learning: Psycholinguistic studies ontraining and retention (pp. 141–158). Mahwah, NJ:Lawrence Erlbaum.

Grigorenko, E. L., Sternberg, R. J., & Ehrman, M. E.(2000). A theory-based approach to the measure-ment of foreign language learning ability: TheCANALF theory and test.Modern Language Journal,84, 390–405.

Guerrero, M. C. M. de (2005). Inner speech-L2: Thinkingwords in a second language. New York: Springer.

Hendry, L., & Tehan, G. (2005). An item/order trade-off explanation of word length and generationeffects. Memory, 13, 364–371.

Hu, C. F. (2003). Phonological memory, phonologicalawareness, and foreign language learning. Lan-guage Learning, 53, 429–462.

Hu, C. F., & Schuele, C. M. (2005). Learning nonnativenames: The effect of poor native phonologicalawareness. Applied Psycholinguistics, 26, 343–362.

Hulme, C., Neath, I., Stuart, G., Shostak, L., Surprenant,A. M., & Brown, G. D. A. (2006). The distinc-tiveness of the word-length effect. Journal ofExperimental Psychology: Learning, Memory, andCognition, 32, 586–594.

Hulme, C., Surprenant, A. M., Bireto, T. J., Stuart, G., &Neath, I. (2004). Abolishing the word-lengtheffect. Journal of Experimental Psychology: Learning,Memory, and Cognition, 30, 98–106.

Krashen, S. D. (1985). The input hypothesis: Issues andimplications. New York: Longman.

Ladefoged, P. (1993). A course in phonetics. New York:Harcourt Brace.

Lawson, M. J., & Hogben, D. (1996). The vocabulary-learning strategies of foreign-language students.Language Learning, 46, 101–135.

MacKay, D. (1992). Constraints on theories of innerspeech. In D. Reisberg (Ed.), Auditory imagery (pp.121–149). Mahwah, NJ: Lawrence Erlbaum.

Miyake, A., & Friedman, N. P. (1998). Individualdifferences in second language proficiency:Working memory as language aptitude. In A. F.Healy & L. E. Bourne (Eds.), Foreign languagelearning: Psycholinguistic studies on training andretention (pp. 339–364). Mahwah, NJ: LawrenceErlbaum.

Papagno, C., Valentine, T., & Baddeley, A. (1991).Phonological short-term memory and foreign-

language vocabulary learning. Journal of Memoryand Language, 30, 331–347.

Pimsleur, P. (2007). How the Pimsleur method works.Retrieved 26 July 2013 from http://www.pimsleur-approach.com/how-pimsleur-works/

Postovsky, V. A. (1974). Effects of delay in oral practice atthe beginning of second language learning.Modern Language Journal, 58, 229–239.

Postovsky, V. A. (1981). The priority of aural compre-hension in the language acquisition process. In H.Winitz (Ed.), The comprehension approach to foreignlanguage instruction (pp. 170–186). Rowley, MA:Newbury House.

Robinson, P. (2013). Aptitude in second languageacquisition. In C. Chapelle (Ed.), The encyclopediaof applied linguistics (pp. 198–202). Malden,MA: Wiley–Blackwell, Retrieved 27 July 2013from http://www.academia.edu/2628115/_2013_Aptitude_in_second_language_acquisition

Rossion, B., & Pourtois, G. (2004). Revisiting Snodgrassand Vanderwart’s object databank: The role ofsurface detail in basic level object recognition.Perception, 33, 217–236.

Rubba, J. (2003). Phonological awareness skills andspelling skills. Retrieved 27 August 2006 fromhttp://cla.calpoly.edu/%jrubba/phon/phona-ware.html

Service, E. (1998). The effect of word length onimmediate serial recall depends on phonologicalcomplexity, not articulatory duration. QuarterlyJournal of Experimental Psychology, 51A, 283–304.

Service, E., & Kohonen, V. (1995). Is the relationbetween phonological memory and foreign lan-guage learning accounted for by vocabularyacquisition? Applied Psycholinguistics, 16, 155–172.

Snodgrass, J. G., & Vanderwart, M. (1980). A standard-ized set of 260 pictures: Norms for nameagreement, image agreement, familiarity, andvisual complexity. Journal of Experimental Psychology:Human Learning & Memory, 6, 174–215.

Sparks, R. L., & Ganschow, L. (1993). Searching for thecognitive locus of foreign language learningdifficulties: Linking first and second languagelearning. Modern Language Journal, 77, 289–302.

Sparks, R. L., Artzer, M., Patton, J., Ganschow, L., Miller,K., Hordubay, D., & Walsh, G. (1998). Benefits ofmultisensory structured language instruction forat-risk foreign language learners: A comparison ofstudy of high school Spanish students. Annals ofDyslexia, 48, 239–270.

Stanovich, K. (1986). Matthew effects in reading: Someconsequences of individual differences in theacquisition of literacy. Reading Research Quarterly,21, 360–407.

Stansfield, C. W. (1989). Language aptitude reconsidered.Washington, DC: ERIC Clearinghouse on Lan-guages and Linguistics.

Tehan, G., & Tolan, G. A. (2007). Word length effects inlong-term memory. Journal of Memory and Lan-guage, 56, 35–48.


Troia, G. (1999). Phonological awareness interventionresearch: A critical review of the experimentalmethodology. Reading Research Quarterly, 34, 28–52.

Wen, Z., & Skehan, P. (2011). A new perspective onforeign language aptitude research: Building and

supporting a case for “workingmemory as languageaptitude.” Revista Ilha do Desterro, 60, 15–43.

Winke, P. (2013). An investigation into second languageaptitude for advanced Chinese learning. ModernLanguage Journal, 97, 109–130.

APPENDIX A

Questionnaire for Volunteers

QUESTIONNAIRE for VOLUNTEER

Age: ___________ Male/Female (circle one)Major at BU: ___________________________________________________________Native (First) Language: ___________________List any other languages that you spoke at home other than the one listed above:_______________________________________________________________________Other languages known/studied:

1. Language ____________________________________________________________Years Studied____________________________________________________________Estimated Level Spoken (circle one): Novice / Intermediate / Advanced / SuperiorEstimated Level Written (circle one): Novice / Intermediate / Advanced / Superior

2. Language _____________________________________________________________Years Studied_____________________________________________________________Estimated Level Spoken (circle one): Novice / Intermediate / Advanced / SuperiorEstimated Level Written (circle one): Novice / Intermediate / Advanced / Superior

3. Language ______________________________________________________________Years Studied______________________________________________________________Estimated Level Spoken (circle one): Novice / Intermediate / Advanced / SuperiorEstimated Level Written (circle one): Novice / Intermediate / Advanced / Superior

(Add more on back of this form if necessary.)How would you rate your ability to learn a foreign language? (check one):____ It comes quite easily to me.____ I have to work at it, but I can learn a foreign language.____ I have to really work at it and I still never really get it.____ I have never really tried to learn a foreign language.

How would you rate your anxiety over learning a foreign language? (check one):____ I have no particular anxiety over learning a foreign language.____ I have mild anxiety.____ I have quite a bit of anxiety.____ I am extremely anxious.

How would you rate your motivation to learn a foreign language? (check as many as apply):____ In general, I am very motivated to learn foreign languages.____ I am somewhat motivated to learn foreign languages.____ I am motivated to learn a foreign language that interests me or that I might need.____ I am really not motivated to learn other languages, because I can get by with English.

""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""

MLAT score: PI ________, PII___________, PIII___________ PIV__________ PV__________Total: __________! _____________ % (HIGH, MED, LOW)Rehearsal Group: _______________ Syllable Recall! _________________ RA: __________


Rehearsal and aptitude in foreign vocabulary learning.

Documents