The insights in this paper about akshara learning (orthographic knowledge) are applicable to early reading in all languages that use the akshara writing system (e.g., Hindi, Bengali). We can say this with confidence for two reasons: First, the logic of the akshara is the same even when the script looks different—e.g., Hindi and Bengali look very different from Kannada but in all three a consonant and a vowel matra combine to form the symbol block called akshara. Second, independent groups have found a similar pattern of akshara learning as reported in this paper—that is, the findings have been replicated in other Indian languages. The paper also reports findings about children’s awareness of the sounds that make up a word (phonological awareness, orthography-phonology mapping). This part of the research findings, however, is only partially understood.

Background:  In 2002, we did not know the exact nature of early skills required for learning to read Kannada. Studies in the 1980s and 90s by G. Purushottam, S. Ramaa, P. Karanth and P. Padakannaya gave us several leads but many areas remained unexplored.  Instead, we knew a lot about learning to read English: the English script is written using letters and letters represent phonemes.  Children who know the names and sounds of letters and are sensitive to the phonemic units that make up a word are at an advantage. They can put these two sets of skills together to decode a word they have never seen before. Knowing about letters is called alphabet knowledge, and being sensitive about the smallest sound in a word is called phoneme awareness. Both these concepts, however, did not seem suitable for Kannada. For one, the ‘akshara’ is quite different from the ‘letter’ (e.g., if you know the name you also know the sound of an akshara but not a letter). Second, when we began our study, all in the research team appeared to prefer the more substantial syllabic sound, over the smaller phonemic sound. It appeared as if it was easier for Kannada speakers to catch the syllabic sounds in a word (many akshara represent sound at this level), as opposed to the phoneme. On reflection, it seemed odd that a sound unit easy for precocious, English-speaking four-year-olds (the phoneme) was difficult to discern for so many of us who were otherwise good readers! Given all these differences, we met the participating students in our study with an open mind.

The survey: We gave children in classes 1 to 3 short language and literacy tasks. Each participating child was met one-on-one by a small team of researchers. We travelled from school to school across the district conducting this research. We did not assess all children in a class. Instead we randomly picked a small number. Such a procedure allowed us to get what, in statistical terms, is called an unbiased, representative sample. In other words, other children from similar schools could be said to be similar to our survey sample and because of this, insights we gained from the sample could be said to apply to not only the children we met but also their peer group who we did not meet. It is this principle of research that would later allow independent groups to make a decision on whether the findings from this study would be relevant in other schools and for other Indian languages.

Another unique aspect of this survey was that it was longitudinal in nature. That is, we returned to meet the same children after fifteen months. The multiple points of data collection (t1 is the first time we met, t2 the second time) allowed us to see how a given skill changed over time. Akshara learning was continuing rapidly but remained incomplete even in class 4. We found that simple akshara like /ka/ and /gi/ were learned earlier than more complex akshara like /swa/ and /pri/.  Perhaps the reason for this is that the more complex akshara contain groupings of sounds, and all the component sounds in this grouping has to be well represented.  For example, in the akshara /swa/, the /s/ sound represents a phoneme, which combines with the next two phonemes, /w/ and /a/, to form a syllable.  In parallel, /swa/ when written down is a more complicated looking akshara than the akshara for just the /sa/ or the /wa/. Children needed more time to learn these multiple layers of complexity.

The design of our research allowed us to also examine whether skills at a certain point in time would predict how children would perform fifteen months later.  We found that early akshara knowledge and syllable awareness was linked with later knowledge of more akshara. Very few children were able to perform the phoneme tasks (that is, most children struggled to work with phoneme units, but a few could). We found that for the children with relatively advanced phonemic sensitivity, there was an early advantage: from the start they were already more discerning about some phoneme sounds and this sensitivity alongside knowing more akshara and being more skilled with noticing syllables predicted their later sensitivity to phonemic units within a word.

Some akshara remained tricky for all children while others were quickly learnt (another line of my research checks what makes some akshara easier to learn than others). Given this pattern of akshara learning, what can this study tell us about children's development as word-readers? 

• First, if a word only has akshara that the child has seen often and recognised successfully before, then she will be able to read the word. Note that these akshara do not necessarily have to be simple. They can also be complex akshara that have come up often in a natural way (e.g., the akshara /thra/ if your name is Chaithra, Nethra or Pavithra).
• Second, words with known akshara should be easy to read even if they are in words seen for the very first time. Some children, however, freeze when seeing an easy but new word. Here, the error does not stem from poor akshara knowledge but perhaps a fear to take risks and make mistakes (a common answer would be, ‘I don’t know’). Successful word-reading is thus also about a positive ‘can do’ attitude with new material.
• Third, there is little point in waiting for children to learn all the akshara before teaching them to read words. There are simply too many akshara and, anyway, many akshara are rarely used. It is advisable to begin word reading right from the start, and pick words that are meaningful and useful, not artificially chosen to be simple. When a complex akshara is encountered in words that are needed by the child, the akshara will be learnt.
• Fourth, be prepared that some akshara learnt in classes one and two will be forgotten by classes four and five. This is because those akshara are infrequently used and naturally fade from memory. One way to avoid complete incomprehension about an akshara is to talk about its parts. Then, if an unexpected akshara is found in a new word, the child can become analytical and attempt to decipher the parts of the akshara.
• Fifth, some akshara will be difficult to split and analyse. This is why building a big vocabulary is important. When the child has a big vocabulary she can guess the word based on the few akshara in a word that she can recognise; she is not stuck because of the akshara she cannot read. Again, a positive ‘can do’ attitude to word-reading helps children attempt words with unfamiliar akshara.
  

  ---

A quick outline of the original paper for interested readers: The first part of the paper (pages 7 to 9) puts down what is known about early skills in English, the nature of the Kannada script and sounds. The second part of the paper describes the children we met (pages 10 and 11), and the different tests we asked them to take to understand their level of skill and knowledge (pages 12 and 13). The third section reports the statistical analysis I conducted (pages 14 to 17) and the fourth part discusses what these analyses reveal (pages 18 to 20).  A parallel story running in this paper is whether children in more efficient schools perform better and if so, does the pattern of skill development look different. We found the pattern was similar but the skills developed faster in better schools. Yet, many akshara remained to be learnt (to follow this thread in the paper start on page 10, then see page 12 for a description of the school indicators checklist we used to judge school functioning, then go to the school-level analysis on page 17 and finish on page 20, where I discuss the findings briefly under the section on practical implications).

---

A highlight of the paper is that the results clearly challenged assumptions made from English. I proposed that a new way of looking at the world’s scripts is needed: the world’s scripts fall along a continuum from contained symbol sets to extensive symbol sets, and the learning demands differ depending on the size of the symbol set (page 19)^[2].

Closing note: This study on ‘pace of acquisition’ shows that the pattern of Kannada skill development is different from the pattern reported in studies on English. Akshara knowledge is different from letter knowledge, syllable awareness is present early and phoneme awareness is slow to develop.

Research such as the one reported in this paper can inform what may work in the classroom. But, equally, researchers need to be alerted of teacher’s successes. If you would like to begin a discussion on how you think children learn the script and decode words, write to the email given below.  

Sonali Nag,

University of Oxford and The Promise Foundation  (promise@vsnl.com)

15^th July, 2018

^[1] This blog piece is a summary of the original paper, Nag, S. (2007). Early reading in Kannada: the pace of acquisition of orthographic knowledge and phonemic awareness. Journal of Research in Reading, 30 (1), 7-22, that can be found at:Nag (2007) JRIR_329

^[2] This description of the world scripts can be found at: http://eli.tiss.edu/index.php/themes/decoding-the-script/