Reading in an alphasyllabary Nag & Snowling - pre · READING IN AN ALPHASYLLABARY 5 reading requires the child to use grain ... Japanese Hiragana ... we assessed whether performance

READING IN AN ALPHASYLLABARY 1

Running Head: READING IN AN ALPHASYLLABARY

Reading in an Alphasyllabary: Implications for a Language-Universal Theory of

Learning to Read.

Sonali Nag

University of York

and

The Promise Foundation, Bangalore

Margaret J. Snowling

University of York

Pre-print version.


Abstract

We report the associations between phonological awareness and orthographic

knowledge in readers of alphasyllabic Kannada. Less fluent 9-12 year olds with

lower orthographic knowledge were at floor on phoneme tasks but more fluent

readers, with greater orthographic knowledge showed significant phonemic

awareness. Orthographic knowledge, phoneme awareness and RAN were

independent predictors of reading rate, and together with syllable awareness

predicted individual differences in reading accuracy. Taken together, we suggest

that increasing alphasyllabic literacy promotes a dual representation at the syllable

and phoneme level, and that the analytic processes involved in acquiring

orthographic knowledge and mappings with phonology are a universal aspect of

reading development across languages.

Keywords: syllable awareness, phoneme awareness, RAN, orthographic knowledge,

non-alphabetic scripts, Indian language


Reading in an alphasyllabary: implications for a language-universal theory of

learning to read.

An important agenda for cross-linguistic research is to clarify which aspects of

reading development are universal as distinct from language-specific (Share, 2008).

However, compared to the voluminous literature on reading development in

alphabetic languages, studies of non-alphabetic scripts are still relatively rare, with

the possible exception of Chinese (Hanley, 2005; Perfetti & Liu, 2005). Here we

report findings of a study of individual differences in children’s reading of Kannada,

an alphasyllabary with typological features seen in several South Asian scripts.

A commonly held view of early phonological development and alphabetic literacy

has been that phonological awareness proceeds from large units, namely syllables,

to awareness of small units or phonemes (Carroll, Snowling, Hulme & Stevenson,

2003; Treiman & Breaux, 1982). It has been proposed that in order to abstract the

principles of the alphabetic system, children need to develop both phoneme

awareness and letter knowledge (Byrne 1998; Share, 1995). However, an

alternative view is that phoneme awareness is a consequence rather than a

precursor of learning to read (Castles & Coltheart, 2004; Goswami & Bryant, 1990;

Morais, Carey, Alegria & Bertelson, 1979).

Much of the debate regarding the causal relations between phoneme awareness

and reading skills draws on evidence from readers of alphabetic languages (Hulme,

Snowling, Caravolas and Carroll 2005; Mann & Wimmer, 2002). Within such

languages, the rate of development of phoneme awareness and of decoding skills

varies with the consistency of the orthography, readers of more consistent

orthographies gaining competence more quickly than readers of English (Seymour,

Aro & Erskine , 2005). Thus, phoneme awareness plays a time-limited role as a


predictor of individual differences in consistent orthographies (de Jong & van der

Leij, 2003) whereas in English, it continues to predict reading skill throughout

development (Bruck, 1992; McDougall, Hulme, Ellis & Monk, 1994; Muter, Hulme,

Snowling & Stevenson, 2004). Such findings favour the idea of reciprocal causation

whereby phoneme awareness, letter knowledge and reading skills interact in the

process of learning to read and phoneme awareness develops rapidly in readers

who primarily encounter consistent grapheme-phoneme relationships.

The role of phonological skills in learning to read in non-alphabetic languages is

less clear. In Chinese, each symbol maps to a syllable pointing to the possibility that

phoneme awareness may not play a critical role in logographic literacy. However,

studies have shown that there is a phase of Chinese word learning that entails

phonological analysis (Ho & Bryant, 1997; Lin et al., 2010) and good and poor

readers differ in their performance on tasks tapping both syllable and phoneme level

skills (McBride-Chang, Bialystock, Chong & Li, 2004; Shu, Peng & McBride-Chang,

2008). A proviso is that these studies involved Chinese readers who had been

taught an alphabetic script, Pinyin, and the findings about the development of

phoneme awareness are less clear in the traditional Chinese script and where

instruction does not include reference to phonetic information (Hanley, 2005).

Drawing together findings from different languages, Ziegler and Goswami (2005)

proposed that reading development is shaped by three factors: the availability (or

accessibility) of different phonological units prior to reading (the phonological

structure of the ambient language, (Caravolas & Bruck,1993)), the consistency of the

orthography-phonology mappings and the granularity of the orthography, specifically

whether mappings are at the level of smaller or larger units (or both as in English).

Within this ‘Psycholinguistic Grain Size’ framework, the successful development of


reading requires the child to use grain sizes in the symbol system of their language

that allow an optimal mapping to the phonology of their language. It follows that

languages in which there are mappings to orthographic units at more than one unit

size (e.g., single letters and letter strings in English) present more of a challenge to

the learner than orthographies that mainly contain small or large units (e.g., Finnish

with predominantly phoneme level units; Japanese Hiragana with predominantly

syllable level units).

Against this back-drop, we report findings from a study of an Indian language,

Kannada, in which we examined the associations between orthographic knowledge

and different phonological units and their concurrent relationship with reading

accuracy and fluency. Orthographic Knowledge refers to knowledge of individual

symbol units and the legal order, both spatial and sequential, of these symbols1. In

addition, we assessed whether performance on a measure of rapid automatized

naming (RAN) would also predict individual differences in reading skills. RAN is a

simple task in which participants name a series of familiar symbols (e.g., letters;

digits) as fast as possible. The time to complete such tasks is a robust predictor of

reading in highly consistent alphabetic orthographies (de Jong & van der Leij, 2003;

Moll, Fussenegger, Willburger & Landerl, 2009; Nikolopoulos, Goulandris, Hulme &

Snowling, 2006), less consistent alphabetic orthographies (Bowey, 2005) and the

logographic Chinese (Liao, Georgiou & Parrila, 2008; McBride-Chang & Ho, 2000),

and it has been hypothesized that RAN taps a cognitive universal associated with

reading - the ability to rapidly access verbal (name) codes from visual (symbol)

information (Lervåg & Hulme, 2009; McCrory , Mechelli, Frith & Price, 2005).

Kannada orthography and phonology


We present Kannada as an example of a language in which both the phonological

units of the language are salient and hence accessible and the consistency of the

mappings between orthography and phonology is high. In standard Kannada, most

words are bi-syllabic and tri-syllabic in length and monosyllabic words are rare.

Words typically end with open syllables and consonant clusters are common in all

word positions.

In most alphabetic languages, the factors of consistency and granularity are

confounded in that consistent orthographies tend also to be low in granularity (often

consisting of reliable, small sized orthographic representations). However in

Kannada, the mappings between orthographic and phonological units vary in size

despite its consistency and hence, can be described as a language with mixed

granularity. To be specific, the primary mapping of phonology in Kannada is at the

level of the orthographic syllable (syllabic) but the symbols of the language also

embody phoneme markers (hence alphasyllabic). These orthographic units with

syllable-phoneme encodings are called akshara and are a characteristic of the

Kannada writing system and all other Indian alphasyllabaries.

The akshara can be divided according to the amount of phonemic information

they encode (see Figure 1). The primary vowels (the /V/ syllables) and the

consonants with the vowel /a/ inherent in the symbol (the /Ca/ syllables) do not

reveal distinct phonemic markers. The CV akshara with vowels other than /a/ (the

CV syllables) encode the consonant and the vowel as two distinct phonemic

markers. At the next level of complexity are the CCV akshara: one phonemic marker

each for the first consonant, the second consonant and the vowel. Technically it is

possible for the writing system to construct larger clusters of phonemic markers (e.g.

CCCV syllables) though the need for representing such sound units is rare in


Kannada. The syllabic akshara (the /V/ and /Ca/ syllables) are taught first followed

by the akshara with embedded phonemes (the CV and CCV syllables), with

instruction methods differing in the extent to which children’s attention is explicitly

drawn to these markers.

The availability of distinct phonemic representation in the Kannada orthography

has one important consequence – the availability of more than 400 distinct symbols,

and learning of this extensive symbol set continues well into the primary school

years (Nag, 2007). A further important feature is the dual mappings between

akshara and phonemes. Consonants and vowels in the Kannada writing system

have two symbol representations; the primary form is a full symbol and the

secondary form is a diacritic2 symbol mapping to the same sound. In the minimal

word pair uma-suma, the phoneme /u/ is written in the primary form in the first word

but the secondary form in the second word: GªÀÄ - ¸ÀÄªÀÄ. Among consonants, the

primary form is the frequently used symbol, with the diacritic form limited to

representing the second consonant in consonant clusters. Among vowels the

diacritic form is the often used form, representing the vowels that follow consonants

in either CV or CCV/CCCV syllables. The primary form of the vowel is reserved for

vowels in word initial positions and for the vowel as a separate morpheme. Thus,

while there is a consistent and regular association of individual symbols to sounds in

Kannada, there are two-symbols for each sound (the primary and the secondary

symbol). These complexities pose challenges to children learning to read and it has

been reported that akshara characteristics such as the position of phoneme markers

carry a processing cost (Karanth, Mathew & Kurien, 2004; Vaid & Gupta, 2002).

While the consistency of the Kannada writing system favours rapid acquisition, poor


readers are slow in their mastery of the large number of orthographic units (Gupta,

2004; Nag, 2007; Ramaa, 1993; Purushottama, 1994).

Turning to phonological development, syllable awareness appears to be more

critical than phoneme awareness in alphasyllabic reading. Reduced phonemic

awareness has been reported among primary school readers of Kannada (Nag-

Arulmani, 2003; Prakash, Rekha, Nigam & Karanth, 1993). There is, however,

evidence that even among primary school children, good readers are better than

poor readers not just in syllabic but also in phonemic processing (Nag-Arulmani,

2003; Ramaa, 1993). Further, early knowledge of the akshara with phonemic

markers (the CV and CCV syllables) is a predictor of later phoneme processing skills

(Nag, 2007). Such findings suggest that alphasyllabic literacy research may be able

to offer fresh insights into the debate regarding the causal role of phonological units

of different sizes (e.g., syllables and phonemes), in learning to read.

We were interested in three inter-related research questions: What is the

association between reading level, syllable and phoneme processing and

orthographic knowledge about the alphasyllabary? To what extent do syllable and

phoneme level processing and orthographic knowledge explain variance in Kannada

reading accuracy and fluency? And does RAN, an important predictor in consistent

alphabetic orthographies, explain variations in reading attainments in the consistent

Kannada alphasyllabary? Based on findings reported in cross-linguistic studies

(Prakash et al., 1993; McBride-Chang et al., 2004; Seymour et al., 2003; Gupta,

2004) we made several predictions about the associations between levels of reading,

phonological awareness and orthographic knowledge and the predictors of individual

differences in reading fluency.


First, because phonological development is known to proceed from large to small

units and the emphasis in the participating schools in our sample was on teaching of

whole akshara (the orthographic syllables) rather than the phonemic markers within

these symbols, we anticipated that syllable awareness would be better than

phonemic awareness. However, because of the availability of phonemic information

in the akshara system, we predicted that increasing knowledge about the akshara

symbols would be associated with greater phonemic awareness; the more akshara

that are known, the more likely it would be that the child had abstracted the

phonemic markers within the symbols.

Second, despite the relatively slow acquisition of phoneme level skills, we

expected measures of both syllable and phoneme awareness to predict individual

differences in reading accuracy given the availability of both large and small sized

mappings in the writing system. Third, we predicted that to the extent that there are

similarities between consistent alphabetic orthographies and the consistent Kannada

alphasyllabary, RAN would be a predictor of Kannada reading skill, providing an

independent contribution to individual differences over and above the role of

phonological awareness and orthographic knowledge.

In summary, given the alphasyllabic nature of the writing system we expected

relative differences in syllable and phoneme processing and despite consistency in

the mappings, we expected to find the association between phoneme awareness

and reading fluency to be more robust over time than reported for other consistent

but contained orthographies in which its role is limited to the very early stages of

acquisition (such as German and Dutch). We first report levels of syllable and

phoneme awareness in three groups of children differing in level of reading fluency


and then examine the role of orthographic knowledge, phonological awareness and

rapid naming as predictors of individual differences in reading accuracy and fluency.

Method

Participants

The investigation was in a representative selection of schools that had earlier participated

in a survey of the reading skills of 411 children in grades 1 to 3 (TPF-NIAS, 2004-07). One

hundred and twelve children took part in the study (Mean age = 9,8; SD = 1.05; 67

boys, 45 girls), drawn from Grades 4, 5 and 6 in 12 schools. Consent for children’s

participation in the study was received from the education authority in accordance

with the norms followed by the government school system in the country. The

sample had a fair representation of the SES groups in the region; 47%, 50.9% and

1.8% of children belonged to the lower, middle and upper-middle socio-economic

status groups respectively. Sixty-six percent of the children were either bi- or

multilingual.

The children began formal Kannada literacy instruction in Grade 1, at around age

5. Schools in the study followed a prescribed Kannada language curriculum and

hence the literacy instruction was similar across participating schools. Akshara

instruction focussed on whole symbols rather than on phonemic markers, and rote

memory of a sight vocabulary was encouraged. The progression of Kannada literacy

instruction was based on prescribed text books for each school year. School policy

required English to be introduced in Grade 3. English instruction was sporadic with

only 12% of the Grade 4-6 children in this study accurately spelling and/or reading

words like ‘boy’ and ‘school’. This 12% did not belong to any particular band of

phonological attainments or Kannada reading attainments reported here.


Tests and Materials

General cognitive ability. This was assessed using Raven’s Standard Progressive

Matrices. General ability is reported in percentile scores based on Indian norms

(Deshpande & Patwardhan, 2006).

Literacy skills. Akshara knowledge. Children read nine symbols comprising no

specific phoneme markers (two primary vowels, seven CV akshara with the inherent

vowel /a/), five symbols with two phoneme markers (CV akshara with vowels other

than /a/) and six symbols with three phoneme markers (CCV akshara). Cronbach

alpha .82.

Reading accuracy. Each child was asked to read a word list, a nonword list and a

set of six texts. Correct reading of an item was given an accuracy score of 1. The

word list comprised three bi-syllabic, six tri-syllabic and one polysyllabic word. The

nonword list comprised ten bi-syllabic, four tri-syllabic and one polysyllabic nonword.

The total number of words in the six texts was 80 (alpha =.97).

Reading speed. The reading speed test was based on Grade 3 – 4 level texts of 26,

14, 19 and 21 words each. Reading rate was calculated by subtracting the errors

from the total number of words in a passage and dividing by time taken (alpha = .94).

Phonological awareness. An item pool of 73 bi-syllabic nonwords was assembled

for use in tests of phonological awareness at the syllable and phoneme levels. The

nonwords were developed from a list of common words by replacing one syllable in a

word with another syllable from another word. The syllable and phoneme tasks were

matched for task demands.


Syllable deletion. In this 20 item test, ten items required deletion of the initial syllable

(si in si-ri) and ten items deletion of the final syllable (ka in shal-ka).

Syllable substitution. In this 15 item test, children had to make five initial

substitutions (gnaa-pa to sne-pa) and ten final substitutions (paab-da to paab-che).

Phoneme deletion. In this 20 item test, ten items each required phoneme deletion in

the initial (s in snera) and final position (i in mushti).

Phoneme substitution. In this 15 item test, five items required phoneme substitutions

in the initial position (kruya with /s/ to become sruya) and ten items required

manipulation of the phoneme in the final position (swari with /u/ to become swaru)3.

Reliability was calculated separately for each task: syllable deletion (alpha = .66),

syllable substitution (.73), phoneme deletion (.91) and phoneme substitution (.88).

Rapid Automatized Naming (RAN). A set of numbers with bi-syllabic Kannada

names (1, 3, 4, 5, 6, 7, 8) were presented in semi-random order printed on a A4

sheet in a sans-serif font with font size of 36. There were a total of 45 numbers in

five rows of nine items each. Each child was asked to rapidly name the numbers.

Rate of processing was calculated by subtracting the number of items read out

wrongly from the total number of items divided by the time taken. Studies using a

similar RAN task have reported reliability to range between .7 and .9 (e.g., Wagner,

Torgesen & Rashotte, 1999).

Procedure

Each child was seen individually for approximately half an hour daily over four

days (the testing battery included tests of oral language, visual memory and reading

comprehension that are not reported here). The first author and three research


assistants who were native speakers of Kannada administered a battery of tests.

Research assistants received two days of orientation and on-line support.

Results

The protocols of nine children were incomplete because they did not come on

one of the four testing days. Data screening revealed that eight children had

exceptionally low reading accuracy and were at floor on several tasks; removal of

these children improved the distributions of variables. We present data on 95

children (57 M, 38 F); aged between 8;11 and 12 years (M = 10, SD = .99); scores

on the Raven’s Matrices ranged from the 25th to the 95th percentile (M = 70.16, SD =

20.31). Table 1 shows performance on measures of literacy, phonological

processing and rapid naming for these participants. Overall children found tasks

requiring syllable awareness easier than those requiring awareness of phonemes.

Concurrent associations between reading level, phonological processing and

orthographic knowledge

We were interested to examine the pattern of phonological processing and

level of orthographic knowledge across children with increasing reading levels, as

measured by the reading speed task. We did this by splitting the sample based on

reading rate. Children with the bottom 33% of scores were considered the group

with the lowest reading rate (Level 1), the middle 33% the next (Level 2) and the top

33% as the highest in reading rate (Level 3). We split the group in this way rather

than by grade because on both accuracy and reading rate there was considerable

overlap in scores between Grades 4 and 5, and complete overlap between Grades 5

and 6.


Table 2 gives the descriptive summaries of the groups with the three Reading

Levels. Level 3 was better than Levels 1 and 2 on the reading measures of accuracy

and speed, akshara knowledge and RAN, as was Level 2 when compared to Level

1. Across all three groups, syllable scores were consistently higher than the

phoneme scores and irrespective of the type of phonological unit (syllable, phoneme)

and manipulation (deletion, substitution); Level 3 was significantly better than both

Levels 1 and 2.

For the next set of analyses we summed the deletion and substitution scores

for each phonological unit. Figure 2 shows the relative attainments in syllable and

phoneme processing in comparison to akshara knowledge at each Reading Level.

Turning to the inter-correlations among measures, associations between syllable

awareness and akshara knowledge was significant at Level 1 (r = .51, p < .001) but

not significant in the other two groups (Level 2, r = .35, ns; Level 3, r = .27, ns). The

pattern was reversed for phoneme awareness. In the groups with better reading

rates, the associations between measures of phoneme awareness and akshara

knowledge were between .48 (Level 2) and .68 (Level 3), both significant at the .001

level. For the group with the slowest reading rate (Level 1) the association between

phoneme awareness and akshara knowledge, r = .33, was not significant.

Concurrent predictors of word recognition

For these analyses we examined the data from all 95 children together. Table 3

shows zero-order correlations among the variables in the upper quadrant and partial

correlations controlling for age and non-verbal ability in the lower quadrant. As seen

in the lower quadrant, there were strong inter-correlations among the measures of

reading and akshara knowledge and in turn, these measures correlated moderately


with measures of syllable and phoneme awareness and rapid naming. There were

moderate to strong inter-correlations among most measures of phonological

awareness (.37 - .72), with the syllable deletion task showing weak correlations with

the two phoneme tasks; substitution (.26) and deletion (.33). The correlations

between phonological awareness and RAN were weak to moderate (.20 - .52).

We next assessed the concurrent predictors of reading accuracy using a series of

parallel hierarchical regressions. For these analyses we formed composite

measures of syllable awareness and phoneme awareness (by summing z scores for

each component task and dividing by 2). In each analysis we first entered General

Ability as a control variable (age did not correlate with word recognition and was

excluded from the analysis). On step 2 we entered akshara knowledge followed on

step 3 by either syllable awareness, phoneme awareness or RAN scores (Table 4).

After General Ability was controlled, akshara knowledge explained a highly

significant 50.1% of the variance in reading accuracy. On the next step, each of the

three phonological variables (syllable, phoneme and RAN) made a further significant

contribution accounting for between 2.2 and 3.7% of the variance. There was a high

degree of commonality between these three measures; when they were entered

together at step 3 they accounted for 5.8% of the variance in reading accuracy. In a

final model, we assessed the proportion of variance in reading accuracy accounted

for by RAN over and above the variance explained by akshara knowledge and

phonological awareness. RAN explained an additional significant 1.9% of this

variance (F (1, 89) = 4.45, p < .05).

Our next analyses followed the same procedure to assess the concurrent

predictors of reading rate. Again, we entered General Ability and akshara


knowledge on steps 1 and 2, followed on step 3 by entering syllable awareness,

phoneme awareness or RAN scores. Together the three phonological variables

(syllable, phoneme and RAN) accounted for a large and significant 21.8% of

variance in reading rate, as shown in the right hand columns of Table 4. When each

of the variables was entered separately, the contribution of syllable awareness was

not significant but the contributions of both phoneme awareness and RAN were

highly significant (13.1 and 16.5% of variance). In a final model, RAN explained

8.3% of the variance in reading rate over and above the variance explained by

akshara knowledge and phoneme awareness (F (1, 89) = 14.78, p < .001).

Discussion

This study investigated the concurrent relationships among reading and reading-

related skills among 9-12 year-old children who were learning to read in the Kannada

orthography, an alphasyllabary. Although Kannada has a consistent orthography, an

extensive symbol set embodying multiple mappings between orthographic and

phonological units has to be mastered. Our first research question was about the

associations between syllable and phoneme level processing, orthographic

knowledge and different levels of reading. We found skilled readers were better than

less skilled readers in both syllable and phoneme awareness; this finding extend

those reported for younger readers of Kannada (Nag, 2007; Prakash et al., 1993).

Consistent with findings of studies with younger children reading alphabetic and

other alphasyllabic orthographies , performance on syllable processing tasks was

better than on phoneme processing tasks, and performance on these tasks was

modulated by the reading level of the child. Our findings, however, stand in contrast

to those from alphabetic languages in two ways. First, we found syllable awareness

to remain an important predictor of reading accuracy through the middle school


years and second, we found that the ability to manipulate phonemes was much less

well developed in these older children; other consistent but alphabetic orthographies

report performance on phoneme tasks to typically reach ceiling (e.g. de Jong & van

der Leij, 2003).

Given the salience of the simple syllable in the spoken form of Kannada (Karanth,

2006), its predominance in the written script and its emphasis in reading instruction

for the sample in the study, it would seem probable that these factors operating

together contribute to the relative stability of syllable awareness as a predictor of

reading accuracy skills. However, since more able readers performed better on

phoneme awareness tasks, our interpretation is that better readers, who are more

familiar with the akshara, are more likely to be able to attend to the internal details of

these symbols to uncover their phonemic constituents. This perhaps triggers a

process of reciprocal interaction such that increased orthographic knowledge

precipitates the development of explicit phoneme awareness. Indeed, although the

present data are cross-sectional and we are careful not to assume causality, our

findings suggest that better readers, who are more familiar with the akshara, are

more likely to show improved phonemic awareness. We propose that this analytic

process is a universal aspect of reading development. Just as Byrne (1998) argued

that children must abstract the alphabetic principle to become proficient readers,

here we propose, children must infer the ‘alphasyllabic principle’; the awareness that

akshara map to phoneme as well as syllable units and both may be used to decode

print. Consistent with this, we have shown that alphasyllabic competence (akshara

knowledge, syllable awareness, phoneme awareness) strongly predicts reading

accuracy in the Kannada orthography. However, reading fluency depends upon

phoneme awareness (associated with the use of phoneme markers) and the skills


tapped by RAN. In line with Lervåg and Hulme (2009) we propose that RAN is an

index of cross-modal learning which taps how well children can retrieve the

phonological forms linked to the orthographic representations in their sight

vocabulary. In short, the distinguishing feature of Kannada is that mappings co-exist

at the level of the orthographic syllable and the phonemes within the same symbol

units. Children who have this insight are better readers.

A further important finding of this study is that, despite the consistency of the

Kannada orthography, the 9-12 year-old children in this sample still showed

significant variation in akshara knowledge and in reading accuracy. This contrasts

markedly with findings from children learning both consistent and inconsistent

alphabetic orthographies who have, by this age, typically mastered the name and

sounds of the alphabet set (Seymour et al., 2003); indeed even in English, reading

accuracy asymptotes by around this age (Francis, Shaywitz, Karla, Shaywitz and

Fletcher, 1996). These findings are consistent with the view that the extensiveness

of the Kannada orthography presents a challenge to learners even when the system

of symbol-sound mappings is regular and consistent.

We now turn to consider the theoretical implications of our findings within the

framework of the Psycholinguistic Grain Size theory (Ziegler & Goswami, 2005).

Specifically, we ask whether the current findings confirm three main predictions: that

the ease of learning to read is determined by the availability of phonological units in

the spoken language, the consistency of the mappings between phonological and

orthographic units and the granularity of these mappings.

A script like the Kannada alphasyllabary, unlike the phonemic alphabetic

systems, is efficient in solving the availability problem. The writing system has a


layer of orthographic units at the level of the syllables that is immediately accessible

for beginning readers, allowing the learning of the sub-lexical layer of phoneme and

phoneme markers to emerge later. We found that children reading Kannada had a

more developed sense of syllable awareness than phoneme awareness, though

better readers were also better in their performance on the phoneme level tasks.

The second prediction of the Psycholinguistic Grain Size theory is that the extent

of consistency in sound-symbol mappings determines reading acquisition. While the

mappings between orthography and phonology are consistent in Kannada, they are

not one-to-one as in alphabetic languages but comprise correspondences of one

phoneme to two symbols. This form of consistency increases the number of

symbols to be learned. Elsewhere we have proposed that it is particularly the

mastery of the low frequency symbols that set the pace of development in reading

and spelling (Nag, 2007; Nag, Treiman & Snowling, 2010). As is evident from this

study, the construct of consistency needs to be modified on at least two counts:

consistency notations may be other than 1:1 mappings and discussion of how

consistency determines reading development is incomplete without reference to the

cognitive demands placed by the extensiveness of a symbol set.

The third construct in the framework is granularity, the size of symbol units that

can reliably map phonology in a language. The conception of granularity appears to

be predominantly based on linearised, alphabetic writing systems in which symbols

are typically sequenced in line (although diacritics are permitted, e.g., French,

Czech). In such alphabetic scripts, grain size refers to single letters (small grain

size) or letter strings (large grain size). In this formulation, smaller grain size

representations are seen as more efficient and these are sufficient to allow reading

to be accurate in consistent orthographies. In contrast, , inconsistent orthographies


use both small and large grain size units to accommodate the phonology of the

language and this renders the reading system less efficient, as is the case in

English. Kannada with its consistent but dual representation at the large and small

unit size (syllable and phoneme) challenges this conception of granularity. We

propose that the advanced akshara knowledge seen among skilled readers of

Kannada includes parallel knowledge about the small and the large unit sizes

embodied in each symbol. While the granularity formulation in the Grain Size theory

would predict that such mixed representations should entail ‘switching costs’ that

slow down reading rate, our findings are to the contrary. Greater Kannada reading

speed was predicted by better akshara knowledge and was arguably indicative of the

ability to abstract the orthographic syllable-phoneme encodings in the symbol set.

Thus, contrary to predictions of Grain Size theory, alphasyllabic strategies, made

possible by the availability of both small and large grain-sized units (mappings to

phonemes and syllables), offered a strategic advantage and reduced rather than

increased processing costs. In summary, the current findings suggest that the

Psycholinguistic Grain Size theory is in need of modification in order to explain data

from alphasyllabic scripts.

A limitation of the present study is that the analyses did not include the bottom

7% of the sample who were at floor on several tasks and had exceptionally low

reading accuracy. We excluded these children given the restricted range in their

data. Another limitation of the present study is that the data are cross-sectional and

therefore cannot address the causal links between reading and the related cognitive

skills that we have assessed. Nonetheless, these findings have implications for a

theory of reading development that addresses both processing universals as well as

typological features of different orthographies. We have found that in Kannada,


learning to read depends upon knowledge of the symbol-sound relationships that

comprise the orthography and the nature of the phonological representations in the

ambient spoken language. The reciprocal influence of these factors is what shapes

the course of reading development and what ultimately will distinguish learning to

read in an alphasyllabary from the development of alphabetic reading.


References

Bowey, J. A. (2005). Predicting individual differences in learning to read. In M. J.

Snowling & C. Hulme (Eds.), The Science of reading: A Handbook (pp. 155-

172). Oxford: Blackwell.

Bruck, M. (1992). Persistence of dyslexics’ phonological awareness deficits, 28, 874-

886.

Byrne, B. (1998). The foundation of literacy. Hove, UK: Psychology Press.

Caravolas, M., & Bruck, M. (1993). The effect of oral and written language input on

children's phonological awareness: a cross-linguistic study. Journal of

Experimental Child Psychology, 55, 1-30.

Carroll, J., Snowling, M. J., Hulme, C., & Stevenson, J. (2003). The development of

phonological awareness in pre-school children. Developmental Psychology,

39, 913-923.

Castles, A., & Coltheart, M. (2004 ). Is there a causal link from phonological

awareness to success in learning to read? Cognition, 91, 77-111.

de Jong, P. F., & van der Leij, A. (2003). Developmental changes in the

manifestation of a phonological deficit in dyslexic children learning to read a

regular orthography. Journal of Educational Psychology, 95, 22-40.

Deshpande, C. G., & Patwardhan, V. (2006). Raven Standard Progressive Matrices:

norms for Indian tribal areas. New Delhi: Manasayan.

Francis, D. J., Shaywitz, S. E.; Karla, K. S.; Shaywitz B. A., & Fletcher, J. M. (1996).

Developmental lag versus deficit models of reading disabilities: a longitudinal,

individual growth curves analysis. Journal of Educational Psychology, 88 (1),

3 – 17.


Goswami, U., & Bryant, P. E. (1990). Phonological skills and learning to read.

London: Erlbaum.

Gupta, A. (2004). Reading difficulties of Hindi-speaking children with

developmental dyslexia. Reading and Writing: An Interdisciplinary Journal, 17,

79-99.

Hanley, J. R. (2005). Learning to read Chinese. In C. Hulme & M. Snowling (Eds.), A

Science of reading: A Handbook. (pp. 272-289). Oxford: Blackwell.

Ho, C. S., & Bryant, P. (1997). Learning to read Chinese beyond the logographic

phase. Reading Research Quarterly, 32, 276-289.

Hulme, C., Snowling, M. J., Caravolas, M., & Carroll, J. (2005). Phonological

awareness is (probably) one cause of success in learning to read. A comment

on Castles & Coltheart. Scientific Studies of Reading (9), 351-365.

Karanth, P. (2006). The Kagunita of Kannada – Learning to read and write an Indian

alphasyllabary. In R. M. Joshi & P. Aaron (Eds), Handbook of Orthography

and Literacy (pp 389-404). Lawrence Erlbaum.

Karanth, P., Mathew A., & Kurien, P. (2004). Orthography and reading speed: Data

from native readers of Kannada. Reading and Writing: An Interdisciplinary

Journal, 17, 101 – 120.

Liao, C.-H., Georgiou, G. K., & Parrila, R. (2008). Rapid naming speed and Chinese

character recognition. Reading and Writing: An Interdisciplinary Journal,

21(3), 231-253.

Lin, D., McBride-Chang, C., Shu, H., Zhang, Y., Li, H., Zhang, J., Aram, D. & Levin, I.

(2010). Small wins big: Analytic Pinyin skills promote Chinese word reading.

Psychological Science, 21(8), 1117-1122.


Lervåg, A., & Hulme, C. (2009). Rapid naming (RAN) taps a basic constraint on the

development of early reading fluency. Psychological Science, 20, 1040-1048.

Mann, V., & Wimmer, H. (2002). Phoneme awareness and pathways to literacy: A

comparison of German and American children. Reading and Writing, 17, 653-

682.

McBride-Chang, C., Bialystok, E., Chong, K. K. Y., & Li, Y. (2004). Levels of

phonological awareness in three cultures. Journal of Experimental Child

Psychology, 89(2), 93-111.

McBride-Chang, C., & Ho, C. S. (2000). Developmental issues in Chinese children's

character acquisition. Journal of Educational Psychology, 92(1), 50-55.

McCrory, E. J., Mechelli, A., Frith, U., & Price, C. J. (2005). More than words: A

common neural basis for reading and naming deficits in developmental

dyslexia? Brain, 128(2), 261-267.

McDougal, S., Hulme, C., Ellis, A. & Monk, A. (1994). Learning to read: the role of

short-term memory and phonological skills, Journal of Experimental Child

Psychology, 58, 112-133.

Moll, K., Fussenegger, B., Willburger, E., & Landerl, K. (2009). RAN is not a

measure of orthographic processing. Evidence from the asymmetric German

orthography. Scientific Studies of Reading, 13(1), 1-25.

Morais, J., Cary, L., Alegria, J., & Bertelson, P. (1979). Does awareness of speech

as a sequence of phones arise spontaneously? Cognition, 7, 323-331.

Muter, V., Hulme, C., Snowling, M. J., & Stevenson, J. (2004). Phonemes, rimes,

vocabulary, and grammatical skills as foundations of early reading


development: Evidence from a longitudinal study. Developmental Psychology,

40, 663-681.

Nag-Arulmani, S. (2003). Reading difficulties in Indian languages. In N. Goulandris

(Ed.), Dyslexia in different languages: cross linguistic comparisons. (pp. 255-

276). London and Philadelphia: Whurr Publishers.

Nag, S. (2007). Early reading in Kannada: the pace of acquisition of orthographic

knowledge and phonemic awareness. Journal of Research in Reading, 30, 1-

17.

Nag, S., Treiman, R., & Snowling, M. J. (2010). Learning to spell in an

alphasyllabary: the case of Kannada. Writing Systems Research, 1(2), 1 -12.

Nikolopoulos, D., Goulandris, A., Hulme, C., & Snowling, M. J. (2006). The Cognitive

Bases of Learning to Read and Spell in Greek: Evidence from a Longitudinal

Study. Journal of Experimental Child Psychology (94), 1-17.

Perfetti, C. A., & Liu, Y. (2005). Orthography to phonology and meaning:

Comparisons across and within writing systems. Reading and Writing, 18(3),

193-210.

Prakash, P., Rekha, D., Nigam, R., & Karanth, P. (1993). Phonological awareness,

orthography and literacy. In R. Scholes (Ed.), Literacy: linguistic and cognitive

perspectives (pp. 55-70). New Jersey: Lawrence Erlbaum.

Purushothama, G. (1994). A framework for testing Kannada reading on the basis of

automaticity, rules of orthography and segmental processing.

Manasagangotri, Mysore: Central Institute of Indian Languages.

Ramaa, S. (1993). Diagnosis and remediation of dyslexia: an empirical study in

Kannada, an Indian language. Mysore: Vidyasagar Printing and Publishing

House.


Seymour, P.H.K., Aro, M., & Erskine, J.M. (2003). Foundation literacy acquisition in

European orthographies. British Journal of Psychology, 94, 143-174.

Share. D.L. (1995). Phonological recoding and self teaching: sine qua non of reading

acquisition, Cognition, 55, 151-218.

Share, D. L. (2008). On the anglocentricities of current reading research and

practice: the perils of overreliance on an ‘outlier’ orthography. Psychological

Bulletin, 134, 584-615

Shu, H., Peng, H., & McBride-Chang, C. (2008). Phonological awareness in young

Chinese children. Developmental Science, 11(1), 171-181.

TPF-NIAS (2004-2007). Language Development Programme, District Quality

Education Programme, Karnataka, India.

Treiman, R., & Breaux, A. M. (1982). Common phoneme and overall similarity

relations among spoken syllables: Their use by children and adults. Journal of

Psycholinguistic Research, 11(6), 569-598.

Vaid, J., & Gupta, A. (2002). Exploring word recognition in a semi-alphabetic

script: the case of Devanagari. Brain and Language, 81, 679-690.

Wagner, R. C., Torgesen, J. K. & Rashotte, C. A. (1999). The Comprehensive Test

of Phonological Processing. Austin, TX: Pro-Ed.

Ziegler, J. C., & Goswami, U. C. (2005). Reading acquisition, developmental dyslexia

and skilled reading across languages: A psycholinguistic grain size theory.

Psychological Bulletin, 131(1), 3-29.


Footnote

1. For researchers of alphabetic writing systems, the term Orthographic

Knowledge usually refers to knowledge of the grapheme sequences that map

to the phonology in the language. Although such knowledge includes

knowledge of letters (individual symbol units), this is often treated as a

separate component called Letter Knowledge. For extensive orthographies,

such as the logographic and alphasyllabary writing systems, knowledge of

symbol units is a significant aspect of orthographic knowledge. In these

writing systems, the terms character knowledge (for the logographies) and

akshara knowledge (for the Indian alphasyllabaries) are used to reflect

knowledge of basic symbol units and their spatial markers.

2. The diacritic is a symbol that can only occur along with another symbol (for

example, the accent in French). In Kannada, specific visuo-spatial rules

govern the position of the diacritic and when used, change the sound value of

the base symbol.

3. The schwa in consonants with inherent vowel has not been considered as a

phoneme.


Figure 1: Types of Orthographic Syllable-Phoneme Encoding in the Kannada

Akshara

Syllables with no phoneme markers: Consonants with the

Inherent Vowel /a/ (the /Ca/ syllables)

PÀ R UÀ WÀ ¥À ¥sÀ § ¨sÀ ªÀÄ

/ka/ /kha/ /ga/ /gha/ /pa/ /pha/ /ba/ /bha/ /ma/

Syllables with two phoneme markers: Consonants with Vowels

other than /a/ (the CV syllables)

V VÃ UÀÄ UÀÆ UÉ UÉÃ UÉÆ UÉÆÃ

/gi/ /gii/ /gu/ /guu/ /ge/ /gee/ /go/ /goo/

Syllables with three phoneme markers: Consonant Clusters (the

CCV syllables)

PÉÌ UÀÄÎ zÀÄÝ ¸Éé ®é ¸ÀÄà vÉ æ ¸ÀÄì

/kke/ /ggu/ /ddu/ /sve/ /lva/ /spu/ /tre/ /ssu/


Figure 2

Percentage of Correct Responses according to Reading Levels on Orthographic Knowledge, Syllable Processing and Phoneme

Processing Measures

Note. Error bars are at 95% Confidence Interval.


Table 1

Descriptive Statistics on Measures of Literacy, Phonological Processing and

Rapid Naming. (N = 95)

M SD Min. – Max.

Reading accuracy (105) 80.43 16.63 12.00 - 103.00

Reading speeda 39.37 16.18 10.50 - 80.00

Akshara knowledge (20) 15.66 3.21 4.00 - 20.00

Syllable deletion (20) 13.56 2.47 5 - 20

Syllable substitution (15) 8.51 3.03 0 - 15

Phoneme deletion (20) 3.63 4.59 0 - 16

Phoneme substitution (15) 2.49 3.42 0 - 12

RAN (digits)b 1.82 .46 .98 - 3.00

Note. a Words per minute. b Items per second.


Table 2

Performance of children at the three Reading Levels on Measures of Literacy, Phonological Processing and Rapid Naming

Level 1

(n =32)

Level 2

(n = 31)

Level 3

(n = 32) F (2, 92) Tukey HSD

M SD Min-Max M SD Min-Max M SD Min-Max

Reading accuracy (105) 65.41 17.81 12-9582.65 7.59

66 - 9693.31 7.51

76 – 103 43.86** 1 < 2 < 3

Reading speeda 22.76 5.86 10.50 –

30.0037.06 4.51

30.97 –

46.1558.19 9.16

46.60 –

80.00

218.68** 1 < 2 < 3

Akshara knowledge (20)13.75 3.42

4 - 2015.58 2.75

9 - 2017.62 2.09

11 - 20 15.26** 1 < 2 < 3

Syllable deletion (20)12.53 2.29

5 - 1613.48 2.50

8 - 1914.66 2.21

11 - 20 6.63** 1= 2, 2= 3, 1 < 3

Syllable substitution (15)7.25 3.21

0 - 138.48 3.02

3 - 149.78 2.32

4 - 15 6.19** 1= 2, 2= 3, 1 < 3

Phoneme deletion(20)1.28 2.19

0 - 82.32 3.66

0 - 147.25 5.04

0 - 16 22.32** 1 = 2, Both < 3

Phoneme substitution (15).66 1.07

0 - 31.90 2.52

0 - 124.91 4.28

0 - 12 17.66** 1 = 2, Both < 3

RAN (digits)b

1.52 .40.98 – 2.25

1.80 .411.05 –

3.002.14 .34

1.25 –

3.00

20.94** 1 < 2 < 3

Note. Level 1 = bottom 33%, Level 2 = middle 33% and Level 3 – top 33% in reading rate.a Words per minute. b Items per second.** p > .001


Table 3

Correlations between Measures of Literacy, Phonological Processing and Rapid Naming. (N = 95)

Accuracy Reading speed Akshara

knowledge

Syllable

substitution

Syllable

deletion

Phoneme

substitution

Phoneme

deletion

RAN

Reading accuracy _ ..69*** 74*** .38*** .49*** .49*** .52*** .42***

Reading speed .72*** _ .52*** .30*** .33*** .55*** .59*** .56***

Akshara knowledge .68*** .49*** _ .43*** .49*** .57*** .54*** .31***

Syllable substitution .34** .24* .39*** _ .59*** .41*** .42*** .26*

Syllable deletion .45*** .29*** .43*** .58*** _ .32*** .37** .17

Phoneme substitution .44*** .51*** .51*** .37*** .26* _ .74** .33**

Phoneme deletion .49*** .55*** .51*** .37*** .33*** .72*** _ .49***

Rapid automatised naming

(RAN)

.46*** .59*** .38*** .27*** .20 .37*** .52*** _

Note. The upper quadrant contains zero-order correlations, the lower quadrant shows partial correlations controlling age and general abilities.

* p < .05; ** p < .01; *** p < .001.


Table 4

Hierarchical Regression Models Predicting Concurrent Reading Accuracy and Reading Rate

(N = 95)

Reading accuracy Reading rate

% R2

Change

F p % R2

Change

F p

Step 1

General abilities 5.3 5.18 < .05 1.6 1.49 ns

Step 2

Akshara knowledge 50.1 103.07 < .001 26.7 34.21 <.001

Step 3

Syllable awareness,

phoneme awareness and

RAN combined

5.8 4.41 < .01 21.8 12.93 <.001

Step 3

RAN 3.7 8.14 < .01 16.5 27.24 <.001

Step 3

Syllable awareness 2.3 4.87 < .05 1.4 1.82 ns

Step 3

Phoneme awareness 2.2 4.75 <.05 13.1 20.39 <.001


Author Note

Sonali Nag, University of York and The Promise Foundation, India; and Margaret J.

Snowling, University of York.

We thank the Department of Education in Chamarajanagara District and members of

the District Quality Education Project at the National Institute of Advanced Studies

for their support in the field; Roopa Kishen, Jayashree Vyasarajan, Mamta Gupta

and Kala B. from The Promise Foundation for data collection and data coding.

Finally, we thank the schools where this study was conducted and the children who

willingly participated in the study.

This work was supported by a research grant to MJS from the British Academy,

preparation of the paper was supported by a Royal Society Newton Fellowship to

SN.

Correspondence: Dr. Sonali Nag, Department of Psychology, University of York,

York, Y010 5DD. Email: [email protected].

Reading in an alphasyllabary Nag & Snowling - pre · READING IN AN ALPHASYLLABARY 5 reading requires the child to use grain ... Japanese Hiragana ... we assessed whether performance

Documents