phonological awareness and speech...

Phonological Awareness and SpeechComprehensibility: An Exploratory Study

H.S. Venkatagiri and John M. LevisIowa State University, Ames, USA

This study examined whether differences in phonological awareness were related todifferences in speech comprehensibility. Seventeen adults who learned English as aforeign language (EFL) in academic settings completed 14 tests of phonological aware-ness that measured their explicit knowledge of English phonological structures, andthree tests of phonological short term memory. The same participants also read alouda passage and narrated picture stories. These tasks were used by 12 native speakersof English to rate the EFL speakers’ comprehensibility on a 9-point scale. There wasa strong positive correlation between composite phonological awareness scores andrated comprehensibility and between composite phonological awareness scores andphonological short term memory. The correlation between rated comprehensibilityand phonological short term memory was not significant. A simple linear regressionanalysis showed that approximately 19% of the variance in rated comprehensibilityscores was accounted for by composite phonological awareness scores. The study of-fers support to the view that phonological awareness is related to differences in speechcomprehensibility and the results suggest that form-focused instruction in phonologymay contribute to the comprehensibility of EFL speakers.

doi: 10.2167/la417.0

Keywords: metaphonology, intelligibility, speech comprehensibility, focus onform, pronunciation, phonological awareness

IntroductionWhy can two learners of a foreign language have the same opportunities,

similar intelligence and similar motivation, yet differ so greatly in ultimateattainment? This differential achievement is evident in all areas of language,but perhaps nowhere is it greater than in pronunciation, in which even learnerswho are advanced in other areas of language can remain incomprehensible. Totake one example, the great English novelist Joseph Conrad, a native speakerof Polish, commanded written English to an unusual degree, but was widelyreported to have a horrible accent and be frequently incomprehensible in hisspoken English (Pousada, 1994).

This paper examines the contribution of one possible factor to speech com-prehensibility (hereafter, comprehensibility) differences, that of phonologicalawareness. By comprehensibility, we follow Munro and Derwing (1995), whodescribe comprehensibility as perceived ease of understanding. Like other meta-linguistic abilities, phonological awareness is a construct, which is measured byhow well learners can focus on the structure of the L2 system, in this case,the phonology. The development of metalinguistic ability is important in L1

0965-8416/07/04 263-15 $20.00/0 C© 2007 H.S. Venkatagiri & J.M. LevisLANGUAGE AWARENESS Vol. 16, No. 4, 2007

263

264 Language Awareness

acquisition (Tunmer et al., 1984) and L2 acquisition (White & Ranta, 2002). Phono-logical awareness has been shown to be related to the teaching of reading skillsfor L2 learners (e.g. Giambo & McKinney, 2004), and in the teaching of foreignlanguage, teachers consistently assume that explicit teaching of the L2 system,whether grammatical or phonological, will facilitate both what is learned andhow quickly it is learned.

The importance of metalinguistic awareness is at the heart of SLA researchrelated to focus on form (FonF). Most FonF research has been done on acquisitionof grammar and vocabulary rather than phonology, but key researchers likeDoughty and Williams (1998) assume that FonF principles apply to all types oflanguage form, including phonology and discourse. Long and Robinson (1998),in a review of FonF research at that time, distinguish three primary approaches tolanguage form in the L2 classroom. The first, the non-interventionist approach,is most strongly associated with Stephen Krashen. In this approach, classroompractice does not focus on form because L2 learning is thought to mimic L1learning, where language form is acquired through attention to meaningfullanguage use. Form is believed to follow naturally without explicit attention.There is increasing evidence that there are qualitative differences between L1 andL2 (especially adult L2) acquisition, and that the non-interventionist approachdoes not seem to reflect the reality of acquisition. Nowhere is this more obviousthan with phonology, where the existence of maturational constraints meansthat only a rare few show native-like abilities.

The second approach, called focus on formS by Long and Robinson (1998: 16),emphasises a synthetic syllabus in which grammatical or other structural ele-ments are predetermined. This approach reflects traditional language teaching,in which a predetermined syllabus defines important language forms that arethe focus of instruction. The forms can be such things as grammar points, vo-cabulary items, or language functions, but all versions of this approach havein common that ‘pedagogical materials are designed to present and practicea series of linguistic items’ which ‘have no independent reason for existence’.A focus on formS approach is a standard practice in pronunciation teaching,but its effectiveness is by no means clear. Derwing and Rossiter (2003), in onerecent study, found that learners who received explicit instruction on vowel andconsonant errors improved their accuracy over time but their overall compre-hensibility did not improve.

The last main approach, the focus on form, or FonF, is distinct from a focuson formS and refers to ‘how focal attentional resources are allocated’ (Long &Robinson, 1998: 23). It employs a meaning-based, interaction approach in whichattention to language form is provided through communicative feedback that‘draws learners’ attention to mismatches between input and output. . . and caninduce the noticing of the kinds of forms’ that would otherwise not be easilynoticed (p. 23). This focus on form is not always provided by a teacher, but ‘oftenconsists of an occasional shift of attention to linguistic code features. . . triggeredby perceived problems with comprehension or production’ (p. 23). Jenkins(2000) reports examples of such shifts of attention in her research on how pro-nunciation errors affect comprehension in NNS–NNS interaction. There appearsto be some evidence that FonF in these interactions led to convergence on morecomprehensible pronunciation.

Phonological Awareness and Speech Comprehensibility 265

An important concept in the FonF research is that of noticing. Schmidt (1990)has proposed the ‘noticing hypothesis’, which claims that conscious awareness(noticing) is essential for the development of L2 (and, presumably, L1) profi-ciency. Schmidt has argued that ‘what learners notice in input is what becomesintake for learning’ (Schmidt, 1995: 20). Schmidt and Frota (1986) have also em-phasised that ‘noticing the gap’ – learners’ awareness of the disparity betweenthe input and their current interlanguage (the developing system of knowledgeof the target language) – enhances learning. While it is not clear that this kind ofnoticing is always amenable to formal instruction (Lightbown, 2000), studies onform-focused instruction find that learners who have such instruction usuallyoutperform those who have meaning-only instruction (Spada, 1997).

Because the subjects in this study self-reported no previous formal instruc-tion in pronunciation, we conclude that their phonological awareness developedthrough a combination of experience with spoken English and formal instruc-tion in areas of language form (e.g. syntax and lexicon) other than phonology. Wehypothesise in this study that learners who show greater phonological aware-ness do so because they are better at noticing patterns in phonological input. Anextension of this hypothesis is that those L2 speakers who are better at notic-ing phonological patterns will also be judged as more comprehensible to NSlisteners.

A second area related to our hypotheses is that of metalinguistic task per-formance. White and Ranta (2002: 261) say that ‘performance of metalinguistictasks. . . has been found to co-vary with proficiency in the L2 and with levelsof L2 aptitude’. Metalinguistic ability in the L2 is often promoted by classroomattention to the formal system of the L2. English as a foreign language (EFL)instruction mostly revolves around vocabulary, syntax and morphology, withrelatively little attention to phonology, as suggested by surveys of practicingteachers in North America, Britain and Australia (Burgess & Spencer, 2000;Macdonald, 2003; Murphy, 1997). However, it is likely that as a consequenceof instruction, learners develop a variable amount of metalinguistic knowledgeabout pronunciation, what we are calling phonological awareness – that is, con-scious knowledge of the sounds, syllable structure, phonotactics and prosody ofthe target language. We used only subjects who self-reported no formal instruc-tion in English pronunciation, and we thus assume that differences in phono-logical awareness reflect the normal variation in any population. Clearly, ‘met-alinguistic performance involves both implicit and explicit knowledge’ (White& Ranta, 2002: 263), but as our tasks involved little direct instruction, we as-sume that the study reflects implicit metalinguistic knowledge. It appears thatcertain learners have higher metalinguistic abilities even without direct instruc-tion although direct instruction also has facilitating effects on metalinguisticperformance (White & Ranta, 2002).

A final related area is aptitude. Aptitude has often been shown by researchto correlate well with success in classroom language learning (Robinson, 2005).Our study examines phonological awareness in the absence of an experimentaltreatment (i.e. instruction), and the study can therefore be seen as tapping into el-ements usually considered to be part of aptitude, especially pattern recognitionand processing speed (considered to reflect abilities involved in Schmidt andFrota’s [1986] idea of ‘noticing the gap’), in addition to ‘phonological working


memory capacity’ and ‘phonological working memory speed’ (Robinson, 2005:51). Although Robinson indicates support for the idea that success in class-room learning correlates with measures of aptitude, there is little evidence thataptitude tests (which heavily focus on phonological awareness and patternrecognition) correlate with speech comprehensibility, although Robinson alsoreports on studies in which ability to learn from recasts appears to be related tophonetic sensitivity (p. 56), suggesting that aptitude may play a role in speechcomprehensibility.

Our study was designed to determine if phonological awareness is related tospeech comprehensibility in a group of adult EFL speakers. Specifically, we areexamining whether subjects with higher scores on a battery of tasks designed tomeasure phonological awareness also were rated as being more comprehensible.A strong positive correlation between phonological awareness and comprehen-sibility would appear to support Schmidt’s (1990) noticing hypothesis. An EFLspeaker with superior phonological awareness presumably notices the ‘gap’between her production of a word and what the word actually requires, even-tually leading to more comprehensible pronunciation. Indeed, an assumptionof much language teaching, including the teaching of pronunciation, is that ex-plicit knowledge of the language system is helpful, and perhaps even necessary.On the other hand, a weak or negative correlation between the two variableswould suggest that phonological awareness does not significantly contribute tocomprehensibility.

In addition to phonological awareness, a second variable, which may be im-portant, is phonological short term memory. Phonological memory is the abilityto recall the sequences of phonemes in words and non-words. Obviously, allnew words in both L1 and L2 begin as non-words until they are stored in longterm memory with their meanings. Gathercole and Baddeley (1990) suggestedthat a non-word repetition task is a purer measure of phonological short termmemory because it is less influenced by vocabulary stored in long term mem-ory. Costanza et al. (1991) reported that phonological short term memory playsa significant role in foreign language vocabulary acquisition. The role of phono-logical short term memory in the development of phonological awareness andintelligible L2 pronunciation is not known. In the present study, two non-wordrepetition tasks were used to measure phonological short term memory of EFLparticipants. We also explore the connection between phonological short termmemory, phonological awareness and comprehensibility to examine whetherphonological short term memory may play a role in articulatory mastery.

Method

ParticipantsSeventeen adult EFL speakers completed a series of tasks that were designed

to tap a range of phonological awareness skills (described in the ‘Tasks’ sec-tion). There were seven males and ten females in the group with a mean ageof 28 years (range: 20–37). On average, they had studied English in a class-room for 11 years (range: 4–17) and had lived in an English-speaking countryfor 15 months (range: 3–60). Chinese was the first language for 10 participants,


Korean for three, Russian for two and Serbian and Romanian for one each. Be-cause many participants could not identify with certainty the particular dialectof the language they spoke, this information was not collected. All participantswere college students who scored a minimum of 500 on the TOEFL (Test of En-glish as a Foreign Language) or its equivalent for the computer-based TOEFL.These scores are a minimum requirement for admission to Iowa State University.The EFL participants were paid for their participation in the study.

The comprehensibility ratings, described under ‘Tasks’ below, were per-formed by 12 native speakers of English. All were adult graduate or advancedundergraduate students majoring in Teaching English as a Second Languageand/or Linguistics.

TasksThere is no single standard set of tasks that measures phonological aware-

ness. Typically, six types of skills are included in the domain of phonologicalawareness. Tasks used in this study to tap each of the six skill areas are describedbelow. Fourteen separate tasks were used to evaluate phonological awareness.

(1) Phonological blending is the ability to blend phonemes into syllables andsyllables into words. A phoneme-blending task in which the participantsheard a sequence of isolated sounds with a short pause between them – e.g./k æ t/ – was used to measure the phoneme blending ability of participants.The participants responded by typing a word that they thought consistedof the sequence of sounds they heard. In order to be counted as correct, thespelling had to be correct (e.g. /b æ k/ required the response, back).

(2) Phonological manipulation is the ability to add, delete, substitute, or rear-range phonemes or groups of phonemes within a word or a phrase. Threetasks were used to measure this skill area in the present study:

(a) Initial consonant deletion. Participants were presented with a picturewith its label written next to it. All words began with a consonant.Participants were asked to pronounce the word without the first con-sonant. Without the first consonant, all words changed to non-words(e.g. frog pronounced as rog).

(b) Final consonant deletion. This was similar to the initial consonant deletiontask described above, except that the participants pronounced the wordwithout the final consonant.

(c) Spoonerisms. The task was to exchange the initial sounds of words intwo-word phrases (e.g. white fish and no tails), so that it would resultin phrases with two different English words (fight wish and toe nails).The transformed phrases were not required to be meaningful as longas they consisted of real English words.

(3) Phonological segmentation is the ability to decompose a word intophonemes and syllables. Two tasks – number of sounds in words andnumber of syllables in words – were used to test this ability.

(a) In the first task, participants were presented with written single-syllablewords one at a time and were asked to sound them out before enteringthe number of sounds present in each word.


(b) In the second task, the participants sounded out one- to four-syllablewords one at a time before entering the number of syllables for eachword.

(4) Phonological sequencing ability allows a speaker to identify the positionof a specified phoneme or a sequence of phonemes within a word.

(a) In the phoneme position identification task, the participant heard aword and, following a brief pause, a vowel or consonant produced inisolation. The participant was then to orally identify the sound thatcame after the reference sound. The responses were audio-recorded.

(b) In the cluster identification task, the participant typed the word-initialor word-final consonant cluster upon hearing each word spoken. Thewords had either word-initial or word-final clusters but not both.

(5) Rhyming and alliteration abilities include identification of rhyming andalliterating words as well as the ability to produce rhyming and alliteratingwords.

(a) Identification of rhyme and identification of alliteration. Participants werepresented with a set of three words on the computer screen, two ofwhich rhymed (in the rhyming task) or alliterated (in the alliterationtask). The participants checked off the two words that they thoughtrhymed or alliterated.

(b) Rhyming fluency and alliteration fluency. The participants were givencommon single-syllable words, one for rhyming and one for alliterationseparately, and asked to produce as many rhyming or alliterating wordsrespectively as they could within 30 seconds. Responses were audio-recorded.

(6) Non-word reading. Non-words are sequences of phonemes that conformto the phonotactic constraints of the language but without any semanticcontent. Reading a non-word would require converting text into a sequenceof sounds and syllables and then applying a prosodic structure to thesequence before pronouncing it. The participants were presented with 10written non-words, one at a time and were asked to read them. Responseswere audio-recorded and evaluated for two elements, each of which wasscored separately: (a) the correct number of syllables and (b) the correctplacement of the main stress relative to how a native speaker of English (thesecond author) with extensive background in Linguistics and English asSecond Language pronounced these words. Some of the non-words usedon this task as well as the other two tasks described below were taken fromCarter et al. (2002) and Gathercole et al. (1994).

Phonological short term memory tasks

(1) Picture – non-word association. The participants were presented with a set of10 line drawings for which English had no words. Each picture had a non-word label underneath it and the participants heard the spoken form ofthe non-word twice. After they had heard the labels for all the 10 pictures,the pictures were rearranged in a pseudorandom order on the screen and


the labels were shown at the bottom of the screen in pseudorandom order.The participants were asked to match the labels with the pictures by typingthe labels in the space provided below the pictures.

(2) Non-word repetition. The participants immediately repeated a non-wordthey heard. The participants’ responses were audio-recorded and wereevaluated for two separate areas: (a) the correct number of syllables and(b) the correct placement of primary stress relative to the spoken test word.

ProceduresA computer program was written in Visual Basic 5, which presented each

of the tasks on a computer screen. The program recorded and tallied correctand incorrect responses for those tasks that required a typed response. Theprogram also audio-recorded responses for tasks that required an oral response.A research assistant listened to the audio recordings and tallied the correct andincorrect responses.

For each of the tasks, there was a practice period preceding the presentation oftest items. During the practice trials, prompts and explanations were freely givento maximise correct responses. This was to ensure that participants understoodthe requirements of the task. However, errors were not explicitly corrected.During the test, participants followed standard written instructions presentedon the screen without any additional prompts, explanations, or feedback. Thescreen layout and written instructions on tests were identical to those on thepractice trials. Practice trials consisted of five items and the test tasks typicallyincluded 10 items. The exceptions were rhyming and alliteration fluency taskswhere participants produced as many rhyming and alliterating words as theycould within 30 seconds. No practice items were used as test items.

Comprehensibility ratingsEach EFL participant in the study read one of the several short passages and

narrated a sequence of pictures that depicted a story. Each text was 120 wordslong. The number of words that fell within the 2000 most frequent words inEnglish ranged from 86 to 91%, as measured by the Compleat Lexical Tutor(http://132.208.224.131/). The participants were allowed to read the passagessilently just prior to reading them aloud and to examine the pictures silentlyfor a short time before describing them. From each EFL speaker’s reading andnarration, three read sentences and three picture description sentences wereextracted. The first and the last sentences read and spoken by a participant wereexcluded from selection to allow for warm-up and cool-down. In addition, in-complete sentences, sentences with false starts and/or phrase repetitions andvery short sentences (those with less than four words) were excluded. A com-puter program, written in Visual Basic 5, presented the sentences to 12 ratersindividually through headphones. The raters listened to the read and spokensentences alternately; the 51 (17 × 3) read and 51 spoken sentences were, how-ever, presented in a pseudorandom order to each rater, so that no two ratersheard the 51 read and 51 spoken sentences in the same order.

The raters judged the comprehensibility of how each sentence was said on a9-point Likert scale (1 = extremely difficult to understand; 9 = extremely easy to


understand). The program recorded the responses and saved them in a text file.At the start of the listening session, each rater listened to the same two read andtwo spoken sentences, one at a time. The raters were told that the first sentencein each group was representative of sentences that were easy to understand andthe second, an example of a sentence that was hard to understand. This was doneto inform raters of the range of performance they were likely to encounter so thatthey would make use of the entire scale – 1 through to 9 – when rating sentences.

Inter-rater reliability

Table 1 Inter-rater reliability measures for 12 raters for three sentences each read andspoken by 17 EFL participants. (ICC: intraclass correlation coefficient)

Sentences Mean Mean Mean Cronbach’s Mean ICC ICCrating minimum maximum alpha corre- (single (average

rating rating lation measures) measures)

Reading 1 6.09 4.82 7.29 0.950 0.627 0.614 0.950

Reading 2 5.87 4.71 6.88 0.953 0.642 0.627 0.953

Reading 3 5.47 4.29 6.82 0.935 0.581 0.546 0.935

Speaking 1 5.33 4.12 6.24 0.946 0.621 0.595 0.946

Speaking 2 5.08 3.94 5.94 0.905 0.447 0.443 0.905

Speaking 3 4.77 3.65 5.65 0.845 0.312 0.312 0.845

Table 1 presents statistics concerning the inter-rater reliability of 12 raters.The mean ratings as well as mean minimum and mean maximum ratings, forthe most part, occupy the middle of the rating scale, indicating that ratersgenerally avoided giving extremely low or extremely high ratings to sentencesthey listened to. The EFL participants in this group were all college students whohad to meet certain minimum English proficiency standards to be admitted totheir degree programmes. On the other hand, no participant had native-likeproficiency in spoken English. This probably explains why there were few veryhigh and very low ratings.

To measure inter-rater reliability, in addition to the various mean ratings,intraclass correlation coefficients (ICC) for consistency were computed usingthe two-way mixed effects model where rater effects were random and ratingeffects were fixed. The Cronbach’s alpha values reported in Table 1 are very high,indicating that there was a high degree of consistency among raters. The samplesthat were highly rated by one rater also received high ratings by other raters andvice versa. However, the ICC (single measures) values are lower, indicating therewas less absolute agreement among raters for individual samples. Overall, theconsistency among raters is satisfactory. This is in line with the results reportedby Derwing and Munro (1997) and Munro and Derwing (1995), who have alsoreported that untrained listeners produce highly reliable ratings.

ResultsTable 2 presents summary data for 17 measures obtained in the study. Mean

proportions reported in Table 2 were computed by dividing the number ofcorrect responses by total items in each task for each participant and then com-puting the mean of these proportions. For alliteration and rhyme fluency tasks,


Table 2 Mean, maximum and minimum proportions and standard deviations for 17measurements obtained from EFL participants

Tasks Mean St. Dev. Max Min

1. Alliteration 0.776 0.164 1.00 0.40

2. Alliteration fluency 0.641 0.184 0.90 0.20

3. Rhyming 0.876 0.135 1.00 0.50

4. Rhyme fluency 0.331 0.193 0.60 0.10

5. Spoonerisms 0.518 0.194 0.90 0.30

6. Number of sounds in words 0.724 0.211 1.00 0.20

7. Number of syllables in words 0.618 0.198 1.00 0.40

8. Initial consonant deletion 0.906 0.144 1.00 0.60

9. Final consonant deletion 0.894 0.100 1.00 0.70

10. Phoneme blending 0.547 0.194 0.90 0.20

11. Phoneme position identification 0.741 0.191 1.00 0.40

12. Cluster identification 0.812 0.196 1.00 0.40

13. Non-word reading (number of syllables) 0.900 0.079 1.00 0.70

14. Non-word reading (stress placement) 0.788 0.136 1.00 0.60

15. Picture–non-word association 0.718 0.253 1.00 0.30

16. Non-word repetition (number of syllables) 0.882 0.221 1.00 0.10

17. Non-word repetition (stress placement) 0.876 0.231 1.00 0.10

which did not have a set number of items, the denominator was 10. Rhymefluency data for four participants, and initial and final consonant deletion datafor one participant each were unavailable due to technical problems with thecomputer. The performance of the group was poorest in rhyme fluency. Thismight be, in part, due to the loss of data for four participants. Except for rhymefluency, participants as a group tended to score in the 50 to 90% range in othertasks. Performance of individual participants, however, varied considerably ondifferent tasks with some participants scoring as low as 20% on some phonolog-ical awareness tasks, while at least one participant scoring 100% on all but fourtasks (alliteration fluency, rhyme fluency, spoonerisms, and phoneme blending).Clearly, most of the participants possessed a significant amount of phonologicalawareness. The performance of the group on phonological memory tasks (tasks15–17 in Table 3) followed a similar pattern.

Table 3 presents composite means for phonological awareness and phonolog-ical short term memory tasks for each participant along with his/her rated com-prehensibility scores. Composite means are in the upper half of the range for bothphonological awareness and phonological memory tasks (except for one partic-ipant who scored 0.4 on phonological memory). In contrast, comprehensibilityratings generally fall in the low to middle portions of the range with only twoparticipants scoring above 7 on the 9-point scale. Classroom learning of L2 ap-pears to ensure a certain level of phonological awareness. Generally, those whoscored in the upper ranges of phonological awareness and tasks were also rated


Tab

le3

Mea

npr

opor

tion

san

dst

and

ard

dev

iati

ons

forp

hono

logi

cala

war

enes

san

dph

onol

ogic

alsh

ort-

term

mem

ory

task

san

din

telli

gibi

lity

rati

ngs

for

17E

FLpa

rtic

ipan

ts


higher in comprehensibility. In contrast, no clear pattern of association appearsto exist between comprehensibility ratings and phonological short term memory.

Table 4 shows a matrix of correlation coefficients for the 17 tasks used in thisstudy. The Pearson’s correlation coefficient between mean rated comprehensi-bility scores and mean composite phonological awareness scores was significant(r [17] = 0.491, p < 0.05) indicating that the two measures are related. A Pearson’sr in the neighbourhood of 0.5 is considered large when measuring the strengthof association between two separate but related variables (Cohen, 1996). In fact,according to Cohen, correlations much larger than 0.5 are ‘measuring the samething’ rather than two different variables. A simple linear regression analy-sis with composite phonological awareness scores as the independent variableand the rated comprehensibility scores as the dependent variable showed that,conservatively, 19% of the variance in rated comprehensibility scores was ac-counted for by phonological awareness scores (R2 = 0.241; adjusted R2 = 0.190).In the present study, phonological awareness was a factor in comprehensibility,although it is clearly not the whole story.

The correlation between phonological short term memory and speech com-prehensibility was non-significant (r [17] = 0.199, p > 0.05). The phonologicalshort term memory scores are clustered at the top of the range, which suggeststhe presence of a ceiling effect. The non-significant correlation between speechcomprehensibility and phonological short term memory may, in part, be the re-sult of this ceiling effect. Phonological short term memory, however, was highlycorrelated with some aspects of phonological awareness (r [17] = 0.502, p <

0.05). Two phonological awareness tasks – alliteration fluency (r [17] = 0.628, p <

0.01) and initial sound deletion (r [17] = 0.827, p < 0.01) – correlated highly withphonological short term memory. The very high correlations suggest that thesetwo tasks and phonological short term memory probably tap the same pool ofcognitive resources. Other phonological awareness tasks were non-significantlycorrelated with phonological short term memory. The three purported phono-logical short term memory tests do not appear to measure the same underlyingconstruct, given that Cronbach’s alpha coefficient was only 0.53 for the threetasks. Nunnaly (1978) and George and Mallory (2003) have stated that an alphavalue of 0.7 or higher is an adequate measure of internal reliability for tests.

Among the phonological awareness tasks, rhyme and alliteration identifica-tion tasks (r [17] = 0.511, p < 0.05), spoonerisms and the number of syllablestasks (r [17] = 0.577, p < 0.05), rhyme identification and cluster identificationtasks (r [17] = 0.483, p < 0.05), the number of sounds and cluster identifica-tion tasks (r [17] = 0.506, p < 0.05), and cluster identification and non-wordreading with correct number of syllables tasks (r [17] = 0.563, p < 0.05) weresignificantly correlated. The number of syllables task correlated highly withsound blending (r [17] = 0.499, p < 0.05), sound position identification (r [17] =0.544, p < 0.05) and non-word reading with correct stress placement (r [17] =0.565, p < 0.05). However, correlations are not perfect or even near perfect,indicating that each task also required unique knowledge not shared with othertasks. Other correlations among the phonological awareness tasks did not reachstatistical significance. Overall, however, the numerous significant correlationssuggest that the different phonological awareness tasks share a common knowl-edge base. Cronbach’s alpha was 0.73 for the 14 phonological awareness tasks,indicating an adequate level of internal reliability for these tests.


Tab

le4

Cor

rela

tion

mat

rix

for

phon

olog

ical

awar

enes

sta

sks,

phon

olog

ical

mem

ory

and

rate

din

telli

gibi

lity

scor

es

1.A

llite

rati

on;2

.Alli

tera

tion

flue

ncy;

3.R

hym

ing;

4.R

hym

efl

uenc

y;5.

Spoo

neri

sms;

6.N

umbe

rof

soun

ds

inw

ord

s;7.

Num

ber

ofsy

llabl

esin

wor

ds;

8.In

itia

lco

nson

ant

del

etio

n;9.

Fina

lcon

sona

ntd

elet

ion;

10.P

hone

me

blen

din

g;11

.Pho

nem

epo

siti

onid

enti

ficat

ion;

12.C

lust

erid

enti

ficat

ion;

13.N

on-w

ord

read

ing

(num

ber

ofsy

llabl

es);

14.N

on-w

ord

read

ing

(str

ess

plac

emen

t);1

5.Ph

onol

ogic

alm

emor

y(p

ictu

re-n

on-w

ord

asso

ciat

ion,

non-

wor

dre

peti

tion

[num

ber

ofsy

llabl

es],

and

non-

wor

dre

peti

tion

[str

ess

plac

emen

t])1

6.C

ompo

site

phon

olog

ical

awar

enes

ssc

ores

;17.

Rat

edin

telli

gibi

lity

scor

es.

∗ p<

0.05

;∗∗ p

<0.

01


There was no significant correlation between the number of years Englishwas studied and composite phonological awareness scores (r [17] = −0.092, p >

0.7) or between years English was studied and rated comprehensibility scores(r [17] = −0.147, p > 0.5). This is not surprising considering the quality andquantity of classroom instruction varies considerably as do individual learn-ers’ aptitude and effort. There were also no significant correlations betweenmonths living in an English-speaking country and phonological awareness(r [17] = 0.221, p > 0.3) and rated comprehensibility scores (r [17] = −0.240,p > 0.3). Generally, living in an L2 environment has a positive effect on com-prehensibility, especially in the initial months and years (Piske et al., 2001).However, in the present sample, all but three participants had lived in anEnglish-speaking country for one year or less. Ten of the 14 participants hadlived in an English-speaking country for between 7 and 9 months making thegroup relatively similar with regard to L2 exposure.

Discussion and ConclusionThe results indicate that phonological awareness may be a factor in the com-

prehensibility of EFL speakers. Greater amounts of explicit knowledge of phono-logical patterns and rules, as measured by the phonological awareness tasks,may have facilitated greater speech comprehensibility. The findings providesome support for Schmidt’s noticing hypothesis (1990; Schmidt & Frota, 1986) –EFL speakers with superior explicit knowledge of phonological structures andpatterns of English were generally rated as more intelligible speakers. However,given that phonological awareness accounted for only 19% of the variance inrated comprehensibility, it appears that ‘noticing’ phonological structures andpatterns in L2 input is not the only aspect related to highly intelligible speech.Schmidt’s contention that additional, beneficial learning can occur when L2learners consciously analyse and understand language structures and patternsappears to find support in this study. Given the incomplete knowledge of thesubjects’ instructional backgrounds, we cannot be certain where the differencesin phonological awareness developed.

The results of this study leave open an obvious question. It is not clear whetherteaching phonological awareness will help L2 speakers become more compre-hensible. Most FonF research indicates that pedagogical intervention that callsattention to formal properties of the L2 encourages greater levels of accuracy, thatwhen pedagogical strategies are used to highlight formal properties of language,learners benefit and may learn the language at a faster rate than would otherwisebe possible (Long & Robinson, 1998). Although most FonF research has exam-ined syntactic and lexical form, there is no reason to believe that phonologicalform would not be amenable to the same kind of awareness-building strategiesin the classroom (Doughty & Williams, 1998). Indeed, it is a basic assumption ofmost classrooms that explicit pedagogical interventions are helpful. What is notclear is whether traditional pronunciation teaching techniques are effective inpromoting noticing. It is likely that pronunciation pedagogy is full of ‘focus onforms’ (Long & Robinson, 1998) practices, many of which may not be effectivein producing greater comprehensibility.

It is not uncommon to find adult L2 learners who, after many years of expo-sure to the language, have reached a plateau that is well short of being sufficient


to meet their comprehensibility needs. It is possible that a significant cause ofthis plateau is the hit-or-miss approach to phonological instruction that appearsto be common throughout the world, at least in the teaching of English. Mul-tiple studies have indicated that teachers simply do not teach pronunciation,partly because they do not feel that they know how (Burgess & Spencer, 2000;MacDonald, 2003; Murphy, 1997). It may be that the FonF approach that seems sopromising in the teaching of other language forms has not been tried sufficientlywith phonological awareness.

The findings do not support a strong version of Krashen’s (1981, 1982, 1994)theory that explicit knowledge of language structures and rules are detrimentalto L2 performance because they interfere with the implicitly acquired systemof knowledge, which, according to Krashen, is the source of L2 competence. Inthe present study, speakers who had explicit phonological knowledge were notat a disadvantage. Instead, they appear to have benefited from that knowledge.If explicit and implicit forms of knowledge are indeed additive, then form-focused instruction in a milieu of meaningful exposure to language shouldaccrue maximum benefit.

The present study, which is based on a correlational analysis, cannot be used tocategorically conclude that greater phonological awareness causes greater com-prehensibility. What it does suggest is that such awareness may be an importantfactor predicting whether an L2 speaker is likely to be easier or more difficultto understand. A logical next step would be an experimental study to exam-ine whether such awareness raising actually helps. Instructing L2 learners inphonological awareness and comparing pre- and post-instruction speech sam-ples may answer more precisely whether comprehensibility can be improvedthrough promotion of awareness. It is also an open question as to whether suchawareness building requires a teacher at all. Instruction in most of the phono-logical awareness skill areas measured in this study can be presented through acomputer interface.

CorrespondenceAny correspondence should be directed to Mr H.S. Venkatagiri, 2130 Pearson

Hall, Iowa State University, Ames, Iowa 50011, USA ([email protected]).

ReferencesBurgess, J. and Spencer, S. (2000) Phonology and pronunciation in integrated language

teaching and teacher education. System 28, 191–215.Carter, A., Dillon, C. and Pisoni, D. (2002) Imitation of nonwords by hearing impaired

children with cochlear implants: Suprasegmental analyses. Clinical Linguistics and Pho-netics 16, 619–638.

Cohen, B.H. (1996) Explaining Psychological Statistics. Pacific Grove, CA: Brooks/ColePublishing Company.

Costanza, P., Valentine, T. and Baddeley, A. (1991) Phonological short-term memory andforeign-language vocabulary learning. Journal of Memory and Language 30, 331–347.

Derwing, T. and Munro, M. (1997) Accent, intelligibility, and comprehensibility: Evidencefrom four L1s. Studies in Second Language Acquisition 20, 1–16.

Derwing, T. and Rossiter, M. (2003) The effects of pronunciation instruction on the ac-curacy, fluency, and complexity of L2 accented speech. Applied Language Learning 13,1–17.


Doughty, C. and Williams, J. (1998) Pedagogical choices in focus on form. In C. Doughtyand J. Williams (eds) Focus on Form in Classroom Second Language Acquisition (pp. 197–262). New York, NY: Cambridge University Press.

Gathercole, S.E. and Baddeley, A. (1990) The role of phonological memory in vocabularyacquisition: A study of young children learning new names. British Journal of Psychology81, 439–454.

Gathercole, S.E., Willis, C. and Emslie, H. (1994) The children’s test of nonword repetition:A test of phonological working memory. Memory 2, 103–127.

George, D. and Mallery, P. (2003) SPSS for Windows Step by Step: A Simple Guide andReference. 11.0 Update (4th edn). Boston, MA: Allyn & Bacon.

Giambo, D. and McKinney, J. (2004) The effects of a phonological awareness interventionon the oral English proficiency of Spanish-speaking kindergarten children. TESOLQuarterly 38, 95–117.

Jenkins, J. (2000) The Phonology of English as an International Language. Oxford: OxfordUniversity Press.

Krashen, S. (1981) Second Language Acquisition and Second Language Learning. Oxford:Oxford University Press.

Krashen, S. (1982) Principles and Practice in Second Language Acquisition. New York: Perg-amon Press.

Krashen, S. (1994) The input hypothesis and its rivals. In N. Ellis (ed.) Implicit and ExplicitLearning of Languages (pp. 45–77). London: Academic Press.

Lightbown, P. (2000) Classroom SLA research and second language acquisition. AppliedLinguistics 21, 431–462.

Long, M. and Robinson, P. (1998) Focus on form: Theory, research and practice. InC. Doughty and J. Williams (eds) Focus on Form in Classroom Second Language Ac-quisition (pp. 15–41). New York: Cambridge University Press.

Macdonald, S. (2003) Pronunciation – views and practices of reluctant teachers. Prospect17 (3). http://www.nceltr.mq.edu.au/prospect/17/pros17 3.html.

Munro, M.J. and Derwing, T.M. (1995) Foreign accent, comprehensibility and intelligi-bility in the speech of second language learners. Language Learning 45, 73–97.

Murphy, J. (1997) Phonology courses offered by MATESOL programs in the US. TESOLQuarterly 31, 741–764.

Nunnaly, J. (1978) Psychometric Theory. New York: McGraw-Hill.Piske, T., MacKay, I. and Flege, J. (2001) Factors affecting degree of foreign accent in an

L2: A review. Journal of Phonetics 29, 191–215.Pousada, A. (1994) The multilingualism of Joseph Conrad. English Studies 75, 335–349.Robinson, P. (2005) Aptitude and second language acquisition. Annual Review of Applied

Linguistics 25, 46–73.Schmidt, R. (1990) The role of consciousness in second language learning. Applied Lin-

guistics 11, 129–158.Schmidt, R. (1995) Consciousness and foreign language learning: A tutorial on the role

of attention and awareness in learning. In R. Schmidt (ed.) Attention and Awareness inForeign Language Learning (Tech. Rep. No. 9, pp. 1–63). University of Hawaii at Manoa,Second Language Teaching and Curriculum Center.

Schmidt, R. and Frota, S. (1986) Developing basic conversational ability in a secondlanguage: A case study of an adult learner of Portuguese. In R. Day (ed.) Talking toLearn: Conversation in Second Language Learning (pp. 237–326). Rowley, MA: NewburyHouse.

Spada, N. (1997) Form-focussed instruction and second language acquisition: A reviewof classroom and laboratory research. Language Teaching 30, 73–87.

Tunmer, W., Pratt, C. and Herriman, M. (1984) Metalinguistic Awareness in Children: Theory,Research and Implications. Berlin: Springer-Verlag.

White, J. and Ranta, L. (2002) Examining the interface between metalinguistic task per-formance and oral production in a second language. Language Awareness 11, 259–290.

phonological awareness and speech...

Documents