Read Writ (2010) 23:1021–1053
DOI 10.1007/s11145-009-9194-6
Reading and spelling acquisition in Thai children
Heather Winskel Æ Kanyarat Iemwanthong
Published online: 24 July 2009
Ó Springer Science+Business Media B.V. 2009
Abstract Thai, a tonal language, has its own distinctive alphabetic orthography.
The study investigates reading and spelling development in Thai children, with an
aim of examining the grain size that is predominantly used when reading and
spelling. Furthermore, word and nonword lists were developed to examine the
acquisition of the complex system of vowels and tone rules in Thai. Reading and
spelling of words and nonwords were assessed in 60 Thai children ranging in age
from 7 to 9 years 8 months from Grade(s) 1, 2, and 3. A lexicality effect was found
for both reading and spelling. Spelling lagged behind reading in the Grade 1 children. Development rapidly increased between the youngest Grade 1 children and the
older Grade 2 and 3 children. For word reading there were significantly more lexical
errors than phonological errors. Beginning readers appear to predominantly use a
larger lexico-syllabic grain size to read Thai.
Keywords Development Grain size Reading Spelling Thai
Lexicality effect
Introduction
The basic aim of learning to read is a process of matching the visual symbols of
the particular orthography with the phonology of the language. The characteristics
H. Winskel (&)
School of Psychology, University of Western Sydney, PO Box 1767,
Penrith South, NSW 1767, Australia
e-mail:
[email protected]
K. Iemwanthong
Center for Research in Speech and Language Processing (CRSLP), Chulalongkorn University,
Phaya Thai Road, Pathumwan, Bangkok 10330, Thailand
123
1022
H. Winskel, K. Iemwanthong
of the orthography and the language that it maps onto, pose different challenges to
beginning readers, and influences the developmental pathways observed in reading
development. The psycholinguistic grain size theory formulated by Ziegler and
Goswami (2005) suggests that the way the phonology of a language is represented
by its orthography, affects the psycholinguistic grain size that is primarily
accessed by beginning readers. The relative consistency of mapping between
phonology and orthography is a crucial contributing factor to the reading
acquisition process and the relative ease with which beginning readers acquire an
alphabetic orthography. In transparent orthographies such as German, acquisition
is rapid and children appear to have ready access to small grain size units,
grapheme-phoneme correspondences, whereas in other less consistent orthographies such as English, children can take several years to gain a similar level of
competence, and appear to use a mixed or flexible grain size strategy (Goswami,
2000, 2003; Seymour, Aro, & Erskine, 2003; Goswami, Ziegler, Dalton, &
Schneider, 2003).
Another characteristic that can affect reading and spelling development is the
degree of balance between the mapping of phonemes to graphemes and graphemes
to phonemes in different orthographies. For example, when there is consistent
mapping between phonemes and graphemes but there are multigrapheme to
phoneme correspondences, then spelling development generally lags behind reading
(e.g. French, Hebrew, Portuguese, and Persian; Geva, Wade-Woolley, & Shany,
1993; Rahbari, Sénéchal, & Arab-Moghaddam, 2007; Pinheiro, 1995; SprengerCharolles, Siegel, & Bonnett, 1998). Thus, predictions can be made about reading
and spelling development based on the characteristics of the orthography and the
language it maps onto.
Thai, a tonal language, forms an interesting case study, as it has its own unique
orthography, which has several interesting, distinctive characteristics worthy of
further investigation. Thai has an alphabetic script, which also has shared properties
with syllabic scripts, as it has implicit vowels for some consonants. In addition, the
syllable is a highly salient unit in the spoken language, as it has clear syllable
boundaries and words are predominantly monosyllabic. Thai consonants are written
in a linear order, but vowels can be written non-linearly above, below, or to either
side of the consonant, and commonly combine across the syllable to produce a
single vowel or dipthong. These hybrid characteristics of Thai give us an
opportunity to investigate the influence that features commonly associated with
alphabetic scripts (e.g. linearity, phonemic prominence) and syllabic scripts (e.g.
non-linearity, syllable prominence) have on reading development (Vaid &
Padakannaya, 2004). Furthermore, lexical tone is a neglected area in reading
research, but is important to investigate as it plays a crucial role in tonal languages
throughout the world (e.g. Mandarin Chinese, Cantonese, and Vietnamese). Little is
known about how beginning readers of Thai learn to read or write Thai, and how
development is influenced by the characteristics of the orthography. The current
study investigates reading and spelling development in Thai children, with an aim of
examining the grain size that is predominantly used when reading and spelling.
Furthermore, the tasks are designed to focus on vowels and tones when reading and
spelling Thai, so that the developmental pathways involved in learning these
123
Reading and spelling acquisition
1023
subsystems can be traced. Reading and spelling of words and nonwords was
assessed in 60 Thai children ranging in age from 7 to 9 years 8 months from Grades
1, 2, and 3. First, more detailed information about the Thai language and its
orthography will be reviewed, followed by relevant background literature, and the
research aims and questions of the current study.
Characteristics of Thai language and its orthography
Thai is a tonal language that has an alphabetic orthography with 44 basic
consonants plus 4 archaic consonants1 for 21 consonant sounds. Thai initial
consonants are listed in Appendix 1. Thai is predominantly monosyllabic but it
also has polysyllabic words, which have been borrowed mainly from Khmer,
Pali, or Sanskrit. Spoken Thai has clearly defined syllable boundaries (Hudak,
1990). Possible syllable structures are V, VC, CV, CCV, CVC and CCVC
[e.g. vowel (V), vowel-consonant (VC)]. In Thai script there are no spaces
between words; instead, spaces in Thai text may indicate the end of a clause or
sentence.
Thai has a high degree of consistency in mapping between phonemes and
graphemes but there are multigrapheme to phoneme correspondences for some
consonants (e.g. for /ph/, /th/, /kh/, /s/, /t h/; refer to the list of initial consonants in
Appendix 2). In addition, there is a change in grapheme-phoneme correspondences
of consonants when they occur in final position. Thai has only nine final consonant
phonemes, /p/, /t/, /k/, / /, /m/, /n/, / /, /w/, /t h/, and other consonants occurring in
/a:ha:r/ food becomes
final position have to take one of these sounds (e.g.
/a:ha:n/). There are 15 initial consonant clusters, which all start with stop consonants
and are followed by laterals, trills, or bilabials;
kr ,
kl ,
kw ,
khr ,
khl ,
khw ,
khr ,
khl ,
khw ,
tr ,
pr ,
pl ,
or
phl ,
phr .
In addition, there are orthographic class-change clusters, in which the first consonant
of the cluster, or is silent, and is used to change the class of consonant to a high or
/
middle class expression with a corresponding change in tone, for example,2
na:w0/ becomes
/na:w42/ cold or
/ja:k2/ becomes
/ja:k1/ want. Thai
does have additional irregularities, which include silent consonants and vowels that
in
week is not
are not pronounced, for example the final consonant
expressed, which in this case is explicitly marked by the diacritic .
Another characteristic of Thai is that many word forms have multiple functions,
which change depending on their position in a clause or sentence. For example,
paj0, ma:0, or daj2 can act as different function words such as prepositions or verbs.
1
There are 4 archaic letters;
, which are rarely used.
2
Tones are marked in the Thai examples cited in this paper as follows; 0 = mid, 1 = low, 2 = falling,
3 = high, 4 = rising. This system is based on the system that was developed at the Linguistics Research
Unit (LRU) of Chulalongkorn University (Luksaneeyanawin, 2004). IPA transcription is used for the
transcription of all other Thai text.
123
1024
H. Winskel, K. Iemwanthong
When reading languages with many multifunction words, we can expect sentence
context and top down processes to play a crucial role.
Vowels
Vowels are placed vertically above, or below, or to either side of the consonant, and
can combine to produce a large additional number of vowels, as illustrated in
Table 1. Vowel length is linguistically significant and Thai has both short and long
vowels. In addition, vowels are not always explicitly specified. Instead, inherent
vowels can occur with the consonant; an a is usually found in words of Sanskrit,
Pali, or Khmer origin, whereas an o is found in native Thai words. However, these
type of vowels are not used in the present study. Thai vowel expressions can be
classified into the following categories based on the combination of letters that
produce the vowel or dipthong: (1) single vowel expressions, (2) simple vowel
combinations that consist of two vowel letters, and (3) complex vowel combinations
that result from the combination of three or more vowel letters (see Table 1). In the
simple and complex vowel combinations, the reader has to combine vowel
orthographic units across the syllable to form a single vowel phoneme or dipthong.
There are also five commonly used vowels that precede the consonant and yet
operate or combine across the syllable, as illustrated in Table 1 (i) Single vowel
expressions. Thai vowels are complex graphemically in their expression, but have
consistent grapheme to phoneme mapping.
Tones
Tone forms an integral element of the syllable in Thai, and serves an essential
function in distinguishing meanings of syllables and words with identical
phonological structure. There are five tones in Thai conceptualised as high, mid,
falling, rising, and low. However, there are only four tone markers;
which
occur above the consonant. The tone determination of a syllable is complex, as it is
influenced by a combination of the class of initial consonant, the type of syllable
(open or closed), the tone marker, and the length of the vowel. In Table 2, it can be
seen that the different tone markers interact with the class of the consonant to
produce particular tone realizations. Typically, the zero marked tone marker
produces a mid level tone, the tone marker, a low level tone, and a falling tone.
However, it can be seen that there are exceptions or irregularities to the rule (marked
in bold). The low tone marker produces a low level tone in both high and mid class
consonants, but with a low class consonant it produces a falling tone. A high class
consonant with mid level zero marking produces a rising tone, which again is an
exception to the rule. A falling tone in combination with high and mid class
consonants produces a falling tone, however, with a low class consonant a high tone
is produced. Furthermore, the final ending of the syllable also interacts with vowel
length and again there are inconsistencies or irregularities in tone expression
(marked in bold). We can see from this review that the expression of tone in Thai
123
Reading and spelling acquisition
1025
Table 1 Thai vowel expressions classified in terms of vowel combinations (adapted from Schoknecht 2000)
123
1026
H. Winskel, K. Iemwanthong
Table 2 Lexical tone is determined by an interaction between tone markers, class of consonant and
syllable ending
Class of initial consonant
High class
consonant
Mid class
consonant
Low class
consonant
Rising
Mid
Mid
Low
Low
Falling
Falling
Falling
High
Any ending except k, p,
or t, or short vowel
Rising
Mid
Mid
k, p, or t after a long vowel
Low
Low
Falling
Short vowel, or k, p, or t
after short vowel
Low
Low
High
Tone marker
Ending
The irregular tone realizations are marked in bold
involves complex rules that can be inconsistent, and that combine and operate at the
syllable level.
The prominence of grain size used when learning to read
and spell in Indo-European orthographies
There has been much debate about what grain size beginner readers and spellers
use across different orthographies when learning to read and spell; whether
beginners use whole words or phonemes or intermediate sub-syllabic units such as
onsets and rimes, body and coda, or syllables (e.g. Duncan, Seymour, & Hill,
2000; Kim, 2007; Treiman & Zukowski, 1996; Ziegler & Goswami, 2005). The
consistency of the orthography affects the grain size used by children learning to
read different orthographies. In English, a notoriously irregular orthography,
children initially appear to adopt a whole word lexical approach in learning to
read (Frith, 1985). Seymour and Elder (1986) and Stuart and Coltheart (1988)
found that reading errors made by English beginning readers involved reading one
word for another and substituting visually similar words for the target word, which
gives evidence for the use of a lexical (whole-word) reading strategy. In contrast,
children learning to read German, a relatively transparent orthography, produce
predominantly nonwords that share letter-sound correspondences with the target
sounds. This indicates that German beginning readers are using a sublexical,
phonemic level of processing (Wimmer & Hummer, 1990). Similarly, Spencer
and Hanley (2003) found that English-speaking children’s reading errors were
predominantly real-word substitutions, whereas children learning to read Welsh,
123
Reading and spelling acquisition
1027
a relatively consistent orthography, produced more phonologically based
reading errors.
Similar to reading most models of spelling have assumed that people rely on two
procedures when engaging in spelling: a lexical look-up procedure and a nonlexical
procedure that uses phoneme–grapheme rules (e.g. Perry, Ziegler, & Coltheart,
2002). For spelling in German, it has been argued that similar to reading, the
nonlexical phonemic level takes precedence (Wimmer & Hummer, 1990). In
English it has also been found that small grain size is important for beginning
spellers (Caravolas, Hulme, & Snowling, 2001; Cassar, Treiman, Moats, Pollo, &
Kessler, 2005). However, other researchers have argued that sublexical units other
than phonemes, such as rimes and consonant–vowel clusters, are also important in
spelling (e.g. Bruck & Treiman, 1990; Treiman, Kessler, & Bick 2002; Treiman &
Zukowski, 1988). Perry and Ziegler (2004) concluded that phoneme–grapheme
units are predominantly used by skilled readers when spelling English words, but
larger grain sizes are also used.
In transparent orthographies, it has been widely found that word recognition
skills develop rapidly with relatively few errors by the end of first year of formal
instruction. In contrast, in English, due to its relatively irregular or non-transparent
orthography, children can take several years to gain a similar level of competence
(Goswami, 2000, 2003). Seymour et al. (2003) found that reading accuracy after
1 year of learning to read was greater than 90% in consistent orthographies such as
Dutch, Finnish, German, Greek, Italian, Norwegian, Swedish and Spanish, whereas
children reading English showed only 34% accuracy for word reading. Similar
results were found for Greek children, who read an average of 90% of real words
correctly and 89% nonwords by the end of Grade 1 (Porpodas, Pantelis, & Hantziou,
1990). This contrasts with children learning to read English, who at the end of Grade
2 on average could read correctly 70% of the real words and only 45% of the
nonwords (Frith, Wimmer, & Landerl, 1998). Reduced levels of accuracy in reading
have been found to be dependent on the relative degree of consistency of
orthographies, for example Danish (71%), French (79%), and Portuguese (73%)
lagged behind the more transparent orthographies, but were more accurate than
English (Seymour et al., 2003). Eventually, in order to become a competent or
skilled reader in any alphabetic orthography, children need to progress beyond
processing individual letters in a word to automatically processing groups of letters
using well developed orthographic skills (Ehri, 1992; Ziegler & Goswami, 2006).
Recent research on the grain size used when learning to read
and spell in Asian orthographies
Recent research on Asian languages has highlighted the syllable as an important
additional processing unit when learning to read and write. Research investigating
children acquiring Kannada, a semi-syllabic Indo-Dravidian script, indicates that the
optimal unit for beginners is the syllable, although more proficient readers and
spellers can also manipulate phonemes (Padakannaya, Rekka, Vaid, & Joshi, 2002).
Research on spelling in Malaysian children (Rickard Liow & Lee, 2004), reading
123
1028
H. Winskel, K. Iemwanthong
and spelling in Indonesian children (Winskel & Widjaja, 2007), and reading in
Korean children (Cho & McBride-Chang, 2005; Kim, 2009), has also highlighted the
syllable as a prominent processing unit.
As discussed previously, Thai vowels can precede the consonant in writing but
follow it in speech; hence there is a mismatch between how the word is spoken and
how it is written. There are two types of misaligned vowels, one that operates within
the syllable (e.g.
:t o:m/ watermelon is spoken as : mo:/), and a more
:ml / insect
severely misaligned vowel that operates across the syllable (e.g.
is spoken /m(a)l : /). This misalignment of vowels allows us to examine whether the
phoneme or syllable is the more prominent processing unit when reading and
spelling Thai. In a recent study, Winskel (2009) collected eye movement data of
adults reading words with and without misaligned vowels in sentences in
comparison with aligned control words (the written and spoken order corresponds,
e.g.
/mamua / mango), reaction times of adults rapidly naming the three
different types of words (the two types of misaligned words and the aligned control
words), and reading and spelling comparable words in children. Results revealed a
processing cost due to the more severely misaligned words where the vowel
operates across the syllable, and gives support for the prominence of the syllabic
level rather than the phonemic level of segmentation for reading and spelling in
Thai adults and children. It was also found that Grade 1 children could read more
words (Grade 1 82%, Grade 2 96%) than they could spell (Grade 1 70%, Grade 2
94%), which reflects the multigrapheme-phoneme imbalance. However, in this
study nonwords were not utilized for the reading and spelling tasks.
Reading and spelling of nonwords
Reading and spelling of nonwords is a particularly useful assessment tool to use
with children learning to read and spell, as it requires children to assemble or decode
word pronunciations, and thus assesses children’s phonological knowledge and
ability to map sublexical units onto graphemes. In transparent orthographies such as
German, children use a similar sublexical approach in reading familiar words as
they do when reading nonwords (Wimmer & Goswami, 1994). Furthermore, the
performance of German first grade children in nonword reading correlated highly
with their reading of familiar words (r = .85), which gives additional evidence that
a similar approach is used by German beginning readers to read both nonwords and
familiar words (Wimmer & Hummer, 1990). It appears that, due to the transparency
of the orthography, German beginning readers have ready access to graphemephoneme correspondence rules, and after only 6 months of instruction were able to
read and write nonwords with considerable success. In contrast, English beginning
readers experience difficulty reading nonwords (and have a corresponding high
refusal rate), as they appear to be relying predominantly on a whole word or lexical
strategy, and do not yet have sufficiently well developed alphabetic knowledge and
skills for reading the nonwords (Wimmer & Goswami, 1994). English children
appear to utilize reading by analogy strategies for rimes to aid word recognition, as
they are a relatively consistent unit in English orthography. Hence, it appears that
123
Reading and spelling acquisition
1029
English children develop a variety of recoding or grain size strategies due to the
irregularity or inconsistency of the orthography (Wimmer & Goswami, 1994;
Ziegler & Goswami, 2005).
The relationship between reading and spelling
A commonly held view is that there are shared representations that underlie both the
reading and spelling acquisition process (Curtin, Manis, & Seidenberg, 2001;
Holmes & Carruthers, 1998). Similar predictors for reading and spelling have been
found in various orthographies, for example in both Turkish and German, relatively
transparent orthographies, letter knowledge and phonemic awareness have been
found to be significant predictors of both spelling and reading (Öney & Durgunoğlu,
1997; Wimmer & Hummer, 1990).
In transparent orthographies such as Italian, Spanish, or Greek, there is an
approximately balanced bidirectional regular mapping between graphemes and
phonemes and vice versa (Borzone de Manrique & Signorini, 1994; Cossu,
Gugliotta, & Marshall, 1995; Porpodas et al., 1990). In contrast, English has a
bidirectionally inconsistent orthography; it is orthographically inconsistent both in
terms of phoneme-grapheme and grapheme-phoneme correspondences (although it
is more inconsistent, in terms of spelling than reading; Ziegler, Stone, & Jacobs,
1997). It has been found in English that children can read more words than they can
spell (Treiman, 1997), whereas in the more consistent orthographies such as Italian
and Spanish, children can spell most of the words that they can read, and can even
spell words that they can not read (Borzone de Manrique & Signorini, 1994;
Thorstad, 1991). In other languages the mapping relationship can be less balanced,
for example when there is consistent mapping between phonemes to graphemes but
there are multigrapheme to phoneme correspondences. Hence, phonemes can be
spelled in multiple ways, which is one of the main reasons why spelling lags behind
reading in languages such as French, Hebrew, Portuguese, and Persian (Geva et al.,
1993; Pinheiro, 1995; Rahbari et al., 2007; Sprenger-Charolles et al., 1998).
Reading and spelling error analysis contributes to our understanding of the
different cognitive strategies that children utilise when learning to read and spell,
and how they acquire the specific rules of the written language they are learning
(Abu-Rabia & Taha, 2004; Worthy, 1990). In general, spelling is considered to
reveal more about a child’s phonological knowledge than reading, as a greater
awareness of orthographic units is required in spelling than in the reading process
(Alcock & Ngorosho, 2003; Lennox & Siegel, 1993). In order to spell effectively,
children need to be able to retrieve well developed orthographic lexical representations (Abu-Rabia & Taha, 2004).
The current study
In the current study, word and nonword reading and spelling are utilized to
investigate how reading and spelling develops in Thai children and to gain insight
123
1030
H. Winskel, K. Iemwanthong
into what particular challenges Thai orthography poses to beginning readers.
Furthermore, the tasks are designed to focus on reading and spelling of vowels
and tones in Thai, which allows us to examine the developmental pathways
involved in the acquisition of these subsystems. A number of predictions can be
made based on previous research in conjunction with the characteristics of Thai
language and its orthography:
The relative degree of prominence of grain size used in reading and spelling
Based on prior research (Winskel, 2009), we expect a larger grain size, the
syllable, to play a prominent role in word recognition in Thai, which in turn will
be reflected in the errors made. As monosyllabic words were used in the current
study, the syllable level also corresponds to the lexical level. On this basis we can
expect more lexico-syllabic errors than phonological errors when reading and
spelling words in Thai. If Thai children also primarily use a larger lexical or
syllabic grain size when reading nonwords, then this should result in difficulty in
reading the nonwords, as they are attempting to use a grain size that is not directly
compatible with the requirements of the nonword reading task. This should result
in a lexicality effect whereby real words are read more effectively than the
corresponding nonwords. In contrast, if children predominantly read alphabetically
using grapheme-phoneme correspondence rules, then reading errors should
primarily be phonological and consist of nonwords beginning with the first letter
of the target word, as occurs in German, Welsh, or Indonesian (Wimmer &
Hummer, 1990; Wimmer & Goswami, 1994; Spencer & Hanley, 2003; Winskel &
Widjaja, 2007). Furthermore, nonword reading would not be expected to
substantially lag behind word reading.
The relationship between reading and spelling
Previous research indicates that in languages in which phonemes can be spelled in
multiple ways, spelling is likely to lag behind reading (Geva et al., 1993; Rahbari,
Sénéchal, & Arab-Moghaddam, 2007; Sprenger-Charolles et al., 1998). As Thai has
multigraphemic consonants as well as graphemically complex vowel and tone
expressions dependent on well developed, stored orthographic knowledge, we can
predict that spelling will lag behind reading in beginning readers. This trend has also
been found in a previous study on Thai (Winskel, 2009).
Vowel development
Due to the complex graphemic representation of Thai vowels, we can expect
beginning readers to have particular difficulty with the complex vowel combinations, as they require the memorization of a complex combination or pattern of
vowels. We can expect this to be reflected in the number of correct responses and
substitution errors made by children, and in their substitution of less complex vowel
123
Reading and spelling acquisition
1031
expressions for more complex ones (Table 1). Hence, we can predict that children
will give more correct responses for single vowel expressions than simple vowel
combinations, and in turn more correct responses for simple vowel combinations
than complex vowel combinations. In relation to errors, we can expect the children
to substitute less complex vowel expressions for more complex ones. Hindi, like
Thai, has a complex system of vowels placed non-linearly above, below, and to the
left or right of the consonant. It has been found that Hindi children have specific
problems in reading and spelling vowels, short and long vowels in particular (Gupta,
2004).
Tone development
As the rules associated with the expression of tone in Thai are complex and can be
inconsistent or irregular, we can expect tone-related errors to be potentially
problematic for children learning to read. For example, in reading and spelling, the
incorrect interpretation or selection of homophonous consonant or tone marker can
result in the incorrect tone realization. Based on prior research, we can also expect
that orthographically visually similar letters to be a source of confusion and
consonant clusters to be potentially problematic, as has been found in other
languages (Sprenger-Charolles & Siegel, 1997; Stuart, 2005; Treiman & Weatherston, 1992).
Methods
Participants
A total of 60 Thai children ranging in age from 7 to 9 years 8 months were recruited
from a public school in a town located in the outskirts of Bangkok. Twenty children
were recruited from each of the following grades: Grades 1, 2, and 3. Grade 1
children (9 boys and 11 girls) ranged in age from 7 to 7 years 8 months with a mean
age of 7 years 2 months (SD = .20); Grade 2 children (9 boys and 11 girls) ranged
in age from 8 to 8 years 10 months with a mean age of 8 years 3 months
(SD = .30); Grade 3 children (12 boys and 8 girls) ranged in age from 8 years
10 months to 9 years 8 months with a mean age of 9 years 3 months (SD = .30).
All children were native speakers of Thai and had no known reading or learning
difficulties as evaluated by their teachers.
Thai children typically begin school in Grade 1 when they are 6 or 7 years of age.
Prior to that, the majority of children attend Kindergarten where they are taught the
letters of the alphabet; the consonants and single vowel letters are taught first.
Initially, they are taught how to read real words or syllables composed of a consonant
and simple vowel (e.g.
/ma:0/ come) without final consonants, then easy words
with final consonants are introduced. At the next stage, both monosyllabic and
bisyllabic words are typically presented to the children. Gradually, children are
taught to read words with more complex combinations of vowels (e.g.
or ).
123
1032
H. Winskel, K. Iemwanthong
Simple sentences are subsequently introduced to the children. When children begin
to learn to read Thai they are given spaced text, but then unspaced text is introduced
in Grade 2 (around 7 or 8 years of age).
In Thailand the school year starts in May. Children in the current study were
tested in September of the school year, so children had been in their current grade
for approximately 4 months. Parental permission was obtained, and standard ethical
guidelines, both, were adhered to.
Stimuli and procedure
The test stimuli consisted of two lists of monosyllabic words to be read or written by
the children. Words were selected on the basis that they were very common familiar
monosyllabic words and also tested children’s knowledge of the vowels and the tone
rules in Thai. The words also occurred in the children’s school readers. The first list
consisted of 28 vowel-related words and the second list consisted of 29 tone-related
words (see Appendices 2 and 3).
The vowel word list was designed to assess children’s knowledge of the different
vowels and vowel combinations that exist in Thai. The vowel word list consisted of
(1) eight single vowels either preceding, following, or occurring above or below the
consonant, (2) six simple vowel combinations consisting of any two vowels listed in
(1), and (3) fourteen complex vowel combinations consisting of more than two
vowels (see Table 1).
The tone word list was designed to assess children’s knowledge of the range of
tone rules in Thai. In the tone list, eight of the words had class-change clusters in
which the first consonant of the cluster, or is silent and is used to change the class
of consonant to a high or middle class expression with a corresponding change in
tone (e.g.
/na:w4/ cold). The other words consisted of representative words
with initial consonant selected from each of the three different consonant classes
combined with an open or closed syllable, and either a tone marker or zero marked
tone marker and short or long vowel.
Four tasks were administered to the children: (a) reading words, (b) reading
nonwords, (c) spelling words, and (d) spelling nonwords for both the vowel and
tone word lists. Nonwords were created by changing the first consonant of the
corresponding monosyllabic word, for example the word
:an/ month
:an/ (both test stimuli were included in either the
became the nonword
word or nonword task). The consonant that was substituted was selected from the
same class of consonant, so that the tone realization remained the same. As
homophonous consonants are a potential source of phonological error, particularly
for spelling and tone realization, care was also taken to select a similar number of
commonly used homophonous consonants for the word and nonword lists; vowel
words had ten homophonous consonants and vowel nonwords had eleven; for the
tone list there were thirteen homophonous consonants for the word list and
fourteen for the nonword list. It is important to note when judging correctness of
response in the nonword spelling task that any of the homophonous consonants
were acceptable as long as the tone was not affected, whereas in the spelling word
123
Reading and spelling acquisition
1033
task the exact homophone had to be selected. This means that the criterion for
judging correctness of responses in the spelling word task were more stringent
than in the nonword spelling task.
Children were tested on the reading and spelling tasks individually in a quiet
room. The order of presentation of the reading and spelling tasks as well as the
presentation of the real words and nonwords was counterbalanced across children,
as they were asked to read and spell the same words and corresponding nonwords.
In addition, the order of presentation of the real words and nonwords was
counterbalanced. Children were given practice items to familiarize them with each
of the tasks. Children read or spelled the real words separately from the nonword
reading or spelling task, and the vowel list was tested prior to the tone list. For the
reading task, the words or nonwords were displayed on a laptop screen in a fixed
random order. Children read the words and nonwords aloud. For the nonwords,
children were told that the words were not real but could be read. Children’s
responses were audio-recorded, transcribed and later analyzed for errors. All
mispronunciations were noted by a research assistant, a native speaker who was
unaware of the purpose of the study. In the spelling task, children were asked to
spell the words dictated by the experimenter. Children were all praised for their
efforts, regardless of whether they gave the correct response or not.
Scoring of responses and error analysis
The number of correct responses for all tasks was recorded. In addition, errors were
classified as either phonological or lexical errors based on Defior, Martos, and
Cary’s (2002) classification system. Phonological errors involved producing another
nonword which is incorrect in some way as it involved a deletion, substitution, or
inversion of some part of the target nonword (e.g. for
/khian4/ write the child
responded with
/sian4/). Lexical errors involved producing a real word for the
/khian4/ write the child responded with
/
word or nonword (e.g. for
khiaw4/ green).
An additional more detailed error analysis was conducted. Errors were further
classified into the following categories:
1.
2.
3.
Unrelated word or nonword. This error response was when children made
either a lexical or phonological response that was completely unrelated to the
target word or nonword (e.g. for
a 0/ town children produced
:n0/ money). These types of errors occurred predominantly in the
youngest, Grade 1 children.
Initial consonant only. This error response was when a child either gave a
lexical or phonological response but only used the initial consonant of the target
word or nonword in the response they made (e.g. for
/plaw1/ empty they
produced /paj0/ go). These types of errors also occurred predominantly in the
Grade 1 children.
Segmentation error. Segmentation errors were when children incorrectly
segmented the word. In general, this occurred with words with consonant
123
1034
4.
H. Winskel, K. Iemwanthong
clusters including the class-change clusters (e.g. for
/plaw1/ empty children
/pe:0-la:2/). These types of errors
responded with a bisyllabic response
were only observed in the reading task.
Minus one errors. These types of errors consisted of the correct monosyllabic
lexical or phonological response except for one error, either an incorrect
consonant, vowel, or tone marker (e.g. instead of
/khiaw4/ green children
/khiaw0/ or for
/plaw1/ empty read or wrote
/
either read or wrote
plaw0/), which is similar to the ‘‘minus one errors’’ cited by Sprenger-Charolles
et al. (1998) and Fernandes, Ventura, Querido, and Morais (2008).
Vowel errors. Reading and spelling errors specific to vowels in the vowel word
list were further categorised into the following categories.
1.
2.
3.
Substitution for simpler vowel expression. In this case the child substituted a
simpler vowel for a more complex vowel combination (e.g. instead of
/
/pla:0/ fish, or for
/t h a2/ believe
plaw1/ empty children read or wrote
/t h 2/ name).
they produced
Vowel length error. In these type of examples the child made an error in
relation to the vowel length (e.g. for
0/ curly children read or wrote
/
aw0/ shade or for /phu/ rotten produced /phu:/).
Incorrect vowel. In these examples the child made a mistake when either
reading or writing one of the vowels or vowel components. Typically, they
would confuse vowels that are graphemically similar (e.g. instead of
/haj4/
/pr : 0/ brush children produced
jar children read or wrote /ho:4/, or for
/pre: 0/), or for the complex vowel combinations children made a mistake
/gl a0/ salt children produced
in one of the vowel components (e.g. for
/gl :0/).
Tone errors. Reading and spelling errors in the tone word list for the minus one
errors were further classified as either adhering to the same tone profile as the target
/fa:k2/ bank children
word or using a different tone to the target word (e.g. for
/fa:k1/ deposit with a different tone profile or
either read or spelled the word as
with the same tone profile as
/pa:k2/, or for
/yaw4/ produced
/ aw4/
/yaw0/). The reason
lonely with the same tone profile or with a different profile
for this analysis was to assess the awareness of tone when reading and spelling
words by the children in the different grades. It is expected that older children will
have developed a greater awareness of tone, and hence produce more same tone
error responses than different tone responses.
Results
Comparison of performance across the different tasks and grades
In order to compare performance in the word and nonword reading and spelling
tasks in the different age groups (see Fig. 1), an analysis of variance was conducted.
123
Reading and spelling acquisition
1035
100
Grade 1
Grade 2
Grade 3
90
80
% correct
70
60
50
40
30
20
10
0
Reading words
Reading nonwords
Spelling words
Spelling nonwords
Fig. 1 The percentage of correct responses for reading and spelling of words and nonwords across grade
The effect of literacy task (reading, spelling) by word type (word, nonword) on correct
responses produced by the children from the three different grades (Grades 1, 2, 3)
was analysed using a repeated measures analysis of variance (ANOVA). Literacy task
and word type were within-subjects factors and grade level was a between-subjects
factor. There was a significant main effect of literacy task, F(1, 57) = 4.24, P = .044,
gp2 = .069, as reading performance was superior to spelling performance. There was
a significant main effect of word type, F(1, 57) = 455.50, P \ .001, gp2 = .889, as
reading or spelling of words was superior to nonwords. There was also a significant
main effect of grade, F(2, 57) = 62.99, P \ .001, gp2 = .689 (refer to Fig. 1). There
was a significant interaction effect between literacy task and grade, F(2, 57) = 5.53,
P = .006, gp2 = .162, and between word type and grade, F(2, 57) = 12.77,
P \ .001, gp2 = .309. There were no other significant effects. Post-hoc comparisons
were conducted for grade using a Bonferroni adjusted a of .017 to avoid familywise
error (Howell, 2002). This revealed that the Grade 1 children scored significantly
lower than Grades 2 and 3 children (P \ .001), which were not significantly
different.
Planned contrasts were conducted. They revealed that both word reading and word
spelling had significantly higher correct responses than nonword reading and
nonword spelling, respectively, for Grade 1 (reading t(59) = 9.72, P \ .001; spelling t(59) = 6.92, P \ .001), Grade 2 (reading t(59) = 10.82, P \ .001; spelling
t(59) = 14.57, P \ .001), and Grade 3 (reading t(59) = 9.14, P \ .001; spelling
t(59) = 8.92, P \ .001) children. For the Grade 1 children, the number of correct
responses for word reading was significantly higher than word spelling (t(19) = 4.50,
P \ .001), and nonword reading was significantly higher than nonword spelling
(t(19) = 2.25, P = .04). However, for both the Grades 2 and 3 children, word reading
and word spelling, and nonword reading and nonword spelling were not significantly
different.
There was a high correlation between word and nonword reading for the Grade 1
children (r = .92), which declined with age, although still significant in the older
children (Grade 2 r = .78; Grade 3 r = .60). There was also a high correlation
123
1036
H. Winskel, K. Iemwanthong
between spelling words and spelling nonwords (Grade 1 r = .77; Grade 2 r = .87;
Grade 3 r = .77). High correlations suggest that similar strategies are being utilized
for reading words and nonwords, and for spelling words and nonwords (Wimmer &
Hummer, 1990; Wimmer & Goswami, 1994).
Lexical and phonological error responses
Reading. In order to examine the effect of word type (word, nonword) on error type
(lexical, phonological) produced by the children from the three different grades
(Grades 1, 2, 3) for reading was analysed using a repeated measures analysis of
variance (ANOVA) was conducted (see Fig. 2). Word type and error type were
within-subjects factors and grade was a between-subjects factor. There was a
significant main effect of word type, F(1, 57) = 287.49, P \ .001, gp2 = .835, with
more errors made for nonword reading than word reading. There was a significant
main effect of error type, F(1, 57) = 5.34, P = .024, gp2 = .086, as there were
more lexical errors made than phonological errors. There was also a significant main
effect of grade, F(2, 57) = 43.21, P \ .001, gp2 = .603. In addition, there was a
significant interaction effect between word type and grade, F(2, 57) = 8.08,
P \ .001, gp2 = .221. There were no other significant effects. Post-hoc comparisons
were conducted for grade. For the Grade 1 children, more lexical errors were
produced than phonological errors for word reading (t(19) = 2.57, P = .019), but
there was no significant difference for the Grade 2 and 3 children. For nonword
reading for all grades, there was no significant difference between the number of
lexical and phonological errors made.
Spelling. The effect of word type (word, nonword) on error type (lexical,
phonological) produced by the children from the three different grades (Grades 1, 2,
3) for spelling was analysed using a repeated measures analysis of variance
80
Correct
Lexical error
Phonological error
70
% response
60
50
40
30
20
10
0
Reading words
Reading nonwords Spelling words
Spelling nonwords
Fig. 2 The proportion of correct responses, lexical errors and phonological errors made to the reading
and spelling tasks
123
Reading and spelling acquisition
1037
(ANOVA; see Fig. 2). Word type and error type were within-subjects factors and
grade was a between-subjects factor. There was a significant main effect of word
type, F(1, 57) = 222.01, P \ .001, gp2 = .796, with more errors made for nonword
spelling than word spelling. There was a significant main effect of error type,
F(1, 57) = 22.34, P \ .001, gp2 = .282, as there were more phonological errors
made than lexical errors. There was also a significant main effect of grade,
F(2, 57) = 78.94, P \ .001, gp2 = .735. There was a significant interaction effect
between word type and grade, F(2, 57) = 8.41, P \ .001, gp2 = .228, and between
error type and grade, F(2, 5 7) = 9.15, P \ .001, gp2 = .243. There were no other
significant effects. Post-hoc comparisons were conducted for grade. For Grade 1
children, there were more phonological errors than lexical errors for spelling words
and nonwords (t(19) = 4.12, P \ .001; t(19) = 3.86, P \ .001, respectively). For
the older children, the Grade 2 and Grade 3 children, the effect was not so apparent
as there was only a significant difference for spelling words (Grade 2 t(19) = 3.20,
P = 005; Grade 3t(19) = 2.18, P = .04), but not for nonwords.
In summary, there was a lexicality effect for both reading and spelling, as reading
and spelling of words was more accurate than the corresponding nonword reading
and spelling tasks in all age groups. Furthermore, it was found that performance in
general was better on reading than spelling. This was more pronounced in the Grade
1 children than in the older children. There was rapid development in literacy skills
between Grade 1 and the older Grade 2 and 3 children. For reading it was found that
there were more lexical error responses made than phonological errors, however, the
reverse trend was apparent for spelling.
Developmental pathways in reading and spelling
Reading. Unrelated and initial consonant only errors predominantly occurred in the
youngest Grade 1 children with a rapid reduction in these types of errors in the
older Grades 2 and 3 children (refer to Fig. 3). These types of errors indicate that
the children are having difficulty reading. Furthermore, these types of errors in
Grade 1 children were predominantly real word errors rather than phonological
errors for word reading and nonword reading for both the unrelated errors (reading:
words 84% lexical, 16% phonological; nonwords 97% lexical, 13% phonological)
and initial consonant errors (reading: words 67% lexical, 33% phonological;
nonwords 80% lexical, 20% phonological). So when reading, the youngest Thai
children commonly responded with either an unrelated word or with a word that
had the same initial consonant only as the target word. Refusals were not apparent
in the Thai children; however, this could be a product of the role of the child, the
Thai education system, and the emphasis in Thai culture on obedience and
compliance.
A common error response in all grades, but particularly the older children, was to
produce minus one errors (a correct syllabic response except for one error). This
type of error was primarily a phonological error due to the use of the incorrect
consonant, vowel, tone marker (e.g. instead of
/khiaw4/ green children
produced ‘‘
/khiaw0/’’), which in turn lead to the incorrect tone realisation.
123
1038
H. Winskel, K. Iemwanthong
400
unrelated
initial consonant only
segmentation error
minus one errors
350
frequency
300
250
200
150
100
50
0
word
noword
reading reading
word nonword word
spelling spelling reading
GRADE 1
noword
reading
word nonword word
spelling spelling reading
GRADE 2
noword
reading
word nonword
spelling spelling
GRADE 3
Fig. 3 Developmental patterns in error type across task and grade
Segmentation errors were mainly made to the words that had complex consonant
/plaw1/
clusters with the preposed vowels occurring prior to the consonant (e.g.
/pe:-0la:2/), or with a silent class-change consonant (e.g.
empty was read as
/ aw4/ lonely was read as
/he0- a:0/).
Spelling. Similar to the reading task the unrelated and initial consonant only
errors were predominantly found in Grade 1 children (refer to Fig. 3). Due to the
characteristics and demands of the task, there were more phonological than lexical
errors made to the spelling task than the reading task. A common error response,
similar to the reading task, was to produce a minus one error with either the
incorrect consonant, vowel or tone marker. This was primarily due to children using
the incorrect homophonous consonant or tone marker. The use of homophonous
consonants in nonword spelling was acceptable as long as the tone of the word was
not changed, however, when spelling words the exact homophonous consonant for
that particular word had to be used to be deemed correct. This may account for the
larger number of phonological errors found in spelling words than spelling
nonwords.
Vowel development
It was predicted that there would be more correct responses for the single vowel
than the simple vowel combination, and in turn more correct responses for the
simple vowel combination than the complex vowel combination. In order to
123
Reading and spelling acquisition
1039
examine the effect of vowel type (single, simple combination, complex combination) on proportion of correct responses produced by the children from the three
different grades (Grades 1, 2, 3) a repeated measures analysis of variance (ANOVA)
was conducted. Vowel type was a within-subjects factor and grade was a betweensubjects factor. There was a significant main effect of vowel type, F(2, 57) = 57.54,
P \ .001, gp2 = .502, and a significant main effect of grade, F(2, 57) = 52.54,
P \ .001, gp2 = .648. In addition, there was a significant interaction effect between
vowel type and grade, F(2, 57) = 14.40, P \ .001, gp2 = .336.
Bonferroni post hoc analysis at .05 level revealed that Grade 1 responses were
significantly lower than Grade 2 which in turn were lower than Grade 3. There was a
significant interaction between vowel type and grade, F(2, 57) = 14.40, P \ .001,
gp2 = .336. The proportion of single vowel responses were significantly higher
than the simple vowel combinations (t(59) = 2.42, P = .02), which in turn were
significantly higher than the complex vowel combinations (t(59) = 7.85, P \ .001).
For the Grade 1 children the proportion of single vowel responses were significantly
higher than the simple vowel combinations (t(19) = 3.61, P = .002), which in turn
were significantly higher than the complex vowel combinations (t(19) = 6.79,
P \ .001). For Grades 2 and 3 only the simple vowel combinations were significantly
higher than the complex vowel combinations (t(19) = 4.76, P \ .001; t(19) = 3.33,
P = .004), as the single vowel and simple vowel combinations were not significantly
different (Table 3).
For spelling the effect of vowel type (single, simple combination, complex
combination) on proportion of correct responses produced by the children from the
three different grades (Grades 1, 2, 3) was analysed using a repeated measures
analysis of variance (ANOVA). Vowel type was a within-subjects factor and grade
was a between-subjects factor. There was a significant main effect of vowel
type, F(2, 57) = 39.62, P \ .001, gp2 = .410, and a significant main effect of
grade, F(2, 57) = 123.62, P \ .001, gp2 = .812. Bonferroni post hoc analysis at .05
level revealed that Grade 1 was significantly lower than Grade 2, which in turn was
lower than Grade 3 (P \ .001). There was a significant interaction between vowel
type and grade, F(2, 57) = 7.51, P \ .001, gp2 = .208.
The proportion of single vowel responses were significantly higher than the
simple vowel combinations (t(59) = 4.06, P \ .001), which in turn were significantly higher than the complex vowel combinations (t(59) = 4.09, P \ .001). For
the Grade 1 children the proportion of single vowel responses were significantly
Table 3 The proportion of correct responses to the different vowel types for reading and spelling
Grade 1
Grade 2
Grade 3
Reading
Spelling
Reading
Spelling
Reading
Spelling
Single vowel
0.66 (.23)
0.48 (.12)
0.89 (.19)
0.94 (.12)
0.88 (.14)
0.93 (.09)
Simple vowel combination
0.46 (.24)
0.37 (.23)
0.89 (.15)
0.83 (.17)
0.90 (.13)
0.88 (.15)
Complex vowel combination
0.15 (.21)
0.11 (.16)
0.74 (.21)
0.81 (.19)
0.79 (.16)
0.81 (.21)
123
1040
H. Winskel, K. Iemwanthong
higher than the simple vowel combinations (t(19) = 2.20, P = .04), which in turn
were significantly higher than the complex vowel combinations (t(19) = 4.97,
P \ .001). For Grades 2 and 3 only the single vowel responses were significantly
higher than the simple vowel combinations (t(19) = 2.74, P \ .01; t(19) = 2.44,
P = .03), as the simple vowel and complex vowel combinations were not
significantly different.
From error analysis, it was also found that the Grade 1 children in particular had
a tendency to substitute simpler vowels for more complex vowel combinations
(Grade 1 19%, Grade 2 10%, Grade 3 9%). In addition, there were incorrect vowel
errors (Grade 1 6%, Grade 2 18%, Grade 3 14%) and vowel length errors (Grade 1
5%, Grade 2 4%, Grade 3 7%) made by all children. Incorrect vowel errors were due
to selecting visually similar vowels (e.g. /o/ for /aj/), or in the complex vowel
was produced
combinations one of the component vowels was incorrect (e.g.
instead of ).
Tone development
For reading and spelling of the tone words and nonwords, the minus one errors were
further classified as either adhering to the same tone profile as the target word or
using a different tone to the target word. Children when reading words tended to
respond with a different tone to the target word rather than retaining the same tone
profile. There were significantly more different than same tone responses for all
grades (Grade 1 t(19) = 5.55, P \ .001; Grade 2 t(19) = 3.36, P = .003; Grade 3
t(19) = 6.02, P \ .001). Similar significant results were found for nonword reading
(Grade 1 t(19) = 6.2, P \ .001; Grade 2 t(19) = 7.63, P \ .001; Grade 3:
t(19) = 10.11, P \ .001).
For spelling words there were only significantly more different than same tone
responses for the Grade 1 children (t(19) = 5.65, P \ .001), but not for the Grade 2
or 3 children. However, for the nonword spelling task, similar to the reading task,
there were significantly more different than same tone responses for all grades
(Grade 1 t(19) = 5.89, P \ .001; Grade 2 t(19) = 4.13, P = .001; Grade 3
t(19) = 5.57, P \ .001). For spelling words in the older children, the same and
different tone responses were not significantly different, which could be interpreted
as indicating that the older children have an enhanced awareness of tone for spelling
in comparison to the younger children (Table 4).
Discussion
In the current study, we examined the development of reading and spelling of words
and nonwords in Thai children from Grade(s) 1, 2, and 3. We were interested in the
particular challenges that this distinctive orthography poses to beginning readers.
One of the major aims was to investigate the prominent grain size used when
reading and spelling words and nonwords in Thai, an orthography which shares
common characteristics with both alphabetic and syllabic scripts. A lexicality effect,
123
Reading and spelling acquisition
Table 4 The same and different error tone responses made to the reading and spelling tasks
Reading
Spelling
Words
Same tone
Nonwords
Different tone
Words
Nonwords
Same tone
Different tone
Same tone
Different tone
Same tone
Different tone
Grade 1
2.40 (1.96)
6.15 (3.51)
2.70 (2.36)
10.40 (5.39)
2.90 (2.22)
7.00 (3.34)
3.30 (2.00)
9.05 (5.77)
Grade 2
0.80 (.95)
2.25 (1.92)
1.55 (1.43)
7.75 (4.06)
1.15 (1.39)
1.90 (2.36)
4.15 (1.53)
7.40 (3.84)
Grade 3
0.80 (.89)
2.70 (1.30)
1.90 (1.92)
9.50 (2.84)
1.05 (1.43)
2.00 (2.64)
2.70 (2.43)
6.30 (4.01)
1041
123
1042
H. Winskel, K. Iemwanthong
a word-advantage was apparent for both reading and spelling, as children performed
substantially better on the word reading and spelling tasks than on the corresponding
nonword reading and spelling tasks. After 4 months of school, the Grade 1 children
achieved 42% correct for reading words and 24% for reading the corresponding
nonwords, and 32% correct for word spelling and only 17% for nonword spelling. It
was noticeable that reading and spelling performance rapidly increased between the
youngest Grade 1 children and the Grades 2 and 3 children, which concurs with
results found in a previous study conducted on Thai (Winskel, 2009). However,
even for the oldest children, the Grade 3 children, nonword reading and spelling still
significantly lagged behind word reading and spelling (86% correct for word
reading, 59% for nonword reading, 87% for word spelling, and 62% for nonword
spelling). These results are more comparable with results found for children learning
to read relatively inconsistent orthographies such as English, rather than more
transparent orthographies such as German. Children learning to read German do not
have difficulty reading nonwords as they have ready access to grapheme-phoneme
conversion rules, which is compatible with the requirements of the nonword reading
task in German, whereas English children have difficulty reading nonwords due to
the inconsistency of the orthography as well as apparently being overly reliant on
using a larger grain size, whole word or lexical strategy, and ‘‘having a rather underdeveloped alphabetic strategy’’ to read the nonwords (Wimmer & Goswami, 1994,
p. 100).
The lexicality effect, the better performance by the Thai children on reading
and spelling of words than nonwords, suggests that a larger grain size (lexical or
syllabic) is being utilised when reading and spelling Thai. If we examine the
proportion of lexical and phonological errors made to the different tasks, we can
make additional inferences about the grain size being used. For word reading
across all the children, there were significantly more lexical errors (59%) than
phonological errors (41%). For reading nonwords, there were still a relatively
high but slightly reduced proportion of lexical errors (lexical: 53% and
phonological: 47%). In the Grade 1 children, the effects were more pronounced
with 61% lexical errors and 39% phonological errors for word reading; 56%
lexical errors and 44% phonological errors for nonword reading. The relatively
high proportion of lexical errors in conjunction with the relatively poor
performance on nonword reading and the high correlation between word and
nonword reading, suggests that Thai children, particularly the youngest Grade 1
children, were predominantly using a larger grain size (syllabic or lexical) to
read both words and nonwords.
For spelling, in contrast to reading, there were significantly more phonological
errors (words 64% and nonwords 57%) than lexical errors (words 36% and
nonwords 43%). This trend was more pronounced in the Grade 1 children than the
older children, as there were 64% phonological errors and 36% lexical errors for
spelling words and 63% phonological errors and 37% lexical errors for spelling
nonwords. The relatively higher proportion of phonological errors in the spelling
than reading tasks reflects the particular demands of the task and the characteristics
of Thai orthography; the multigrapheme-phoneme mapping and the complex
graphemic expression of the vowel and tone subsystems. Effective spelling is
123
Reading and spelling acquisition
1043
dependent on the retrieval of stored orthographic representations; hence well
developed orthographic knowledge for spelling is required, which is particularly
under-developed in the youngest, inexperienced beginning readers (Rahbari &
Sénéchal, 2009). Furthermore, in order to become a proficient speller, additional
orthography-specific contextual and linguistic knowledge (besides those for
reading) is required, which includes grammatical and morphological knowledge
(Alcock & Ngorosho, 2003; Cossu et al., 1995; Rahbari & Sénéchal, 2009; Rickard
Liow & Lee, 2004).
The rapid increase in children’s spelling performance in Grade 2 reflects
children’s developing orthographic skills and knowledge. The high correlation
between word and nonword spelling in conjunction with the relative proportion of
phonological and lexical errors made in both tasks, suggests that similar strategies
are being utilized by Thai children when spelling words and nonwords. However,
results are somewhat contradictory for spelling, as there was a lexicality effect but
in addition more phonological than lexical errors were made by the Grade 1
children. In the older Grade 2 and 3 children, there were more phonological than
lexical errors for spelling words but not for spelling nonwords. This difference
between the phonological errors for spelling words and nonwords in the older
children could be due to the effect of homophonous consonants. In the nonword
spelling task any of the homophonous consonants were deemed as correct as long as
the tone was not affected, whereas in the spelling task the exact homophone needed
to be selected.
In transparent regular orthographies, word recognition skills develop rapidly with
relatively few errors for both word and nonword reading and spelling by the end of
the first year of formal instruction (Seymour et al., 2003). However, similar levels of
competence for children learning to read less regular orthographies, such as English,
Danish, or French takes a prolonged time to develop. In comparison to these other
orthographies, Thai reading and spelling development is also relatively delayed, as
although children scored 86% correct by Grade 2 for reading and spelling words,
they still lagged behind in nonword reading and spelling with only approximately
60% correct in Grade 2 and Grade 3. Thai children are quite competent in word
reading and spelling by Grade 2, but nonword reading and spelling is still
problematic even for the Grade 3 children. This is particularly surprising
considering that the only difference between the monosyllabic words and nonwords
was the initial consonant.
The prediction that Thai children would perform better on reading than on
spelling was supported. It was found that the youngest Thai children were more
accurate in reading than spelling words, and in turn read accurately more nonwords
than they could spell. This is similar to results found in a previous study on Thai
(Winskel, 2009), as it was also found that Grade 1 children could read correctly
more words (Grade 1 82%, Grade 2 96%) than they could spell (Grade 1 70%,
Grade 2 94%). Children in that study had been in their current grade for 9 months,
whereas in the present study children had been in their current grade for only
4 months. The lower performance on spelling and the relatively high proportion of
phonological errors than lexical errors on the spelling tasks, in particular the
spelling of words, reflects the relative imbalance between phoneme-grapheme (PG)
123
1044
H. Winskel, K. Iemwanthong
and grapheme-phoneme (GP) mapping and the complex graphemic expressions for
vowels and tones in Thai. Notably, in both these studies, the Grade 1 children
lagged substantially behind in word spelling in comparison with word reading
ability, but by Grade 2 this disparity had dissipated. It appears that difficulties
associated with learning to spell Thai words were to a large extent overcome by
Grade 2. The differences in performance in the two studies on Thai are presumably
due to the length of time (in months) children had spent in their current grade and
the types of word stimuli used to assess the children’s reading and spelling ability.
In the current study, the words selected were designed to assess children’s
knowledge of the complex vowel and tone subsystems.
More detailed error analysis can contribute to our understanding of the different
cognitive strategies that children utilize and the particular challenges that the
orthography poses to beginning readers, as well as the developmental pathways
involved in acquiring the specific rules of the written language they are learning
(Abu-Rabia & Taha, 2004; Worthy, 1990). The youngest children, the Grade 1
children, made a high proportion of unrelated and initial consonant errors, which
reflects their poorly developed phonological and orthographic skills, and inability at
this stage to decode or encode the words and nonwords. These errors were also
predominantly lexical errors for reading, which supports the notion that these
children are using predominantly a lexical strategy when reading. The youngest
children when reading also had a tendency to incorrectly segment the monosyllabic
words with consonant clusters, including the class-clusters, into two syllables. These
types of errors, however, rapidly declined in the older children as their phonological
and orthographic skills developed.
Children acquired the single vowels prior to the simple vowel combinations,
and the more complex vowel combinations were acquired later (as reflected by the
number of correct responses). Additional evidence supporting this order of
acquisition was that Grade 1 children, in particular, frequently substituted simpler
vowels for more complex vowel combinations. This confirms that beginning
readers do find the complex Thai vowels challenging when learning to read and
write Thai. In addition, children had difficulty as reflected in the errors produced,
with vowel length and visually similar vowels, which concurs with results found in
Hindi (Gupta, 2004). However, by Grade 2, children had largely acquired the
complex system of vowels and made relatively few vowel-related errors. The
graphemic expression of Thai vowels is complex, but there is consistent mapping
between phonemes and graphemes. A common error response in all children was
to produce the correct response apart from one error, termed ‘‘minus one errors’’.
These errors were primarily due to the incorrect selection of consonant, vowel, or
tone marker.
Tone errors were still apparent in the older children, which were largely due to
selection of the incorrect homophonous consonant or tone marker, which affected
the tone realisation of the syllable. As expected the complex rules associated with
the expression of tone proved problematic for Thai children. Furthermore, children
in general tended to produce ‘‘minus one errors’’ with an incorrect rather than
correct tone profile for reading, which suggests that tone expression takes a
secondary role to the expression of consonants and vowels when reading. In relation
123
Reading and spelling acquisition
1045
to spelling words, this effect was only apparent in Grade 1 children but not Grade
2 or 3 children, which suggests that the older children have an enhanced
awareness of tone particularly when spelling real words in comparison to the
younger children.
As predicted, children tended to make errors on the words involving consonant
clusters; when reading they either tended to not pronounce the second consonant in
the cluster or segment the word incorrectly, or when spelling they tended to drop the
second consonant. This difficulty with consonant clusters is similar to results found
in other studies (Sprenger-Charolles & Siegel, 1997; Stuart, 2005; Treiman &
Weatherston, 1992; Treiman, 1993). Specific to Thai, the younger children had
difficulty with the orthographic class-change clusters, in which the first consonant of
the cluster, or is silent. When reading children tended to pronounce the silent
/ aw4/ was pronounced
consonant and segment the word inappropriately (e.g.
/he:4- a:0/) and when spelling, these silent class-change consonants were
as
often omitted.
We were particularly interested in the errors that persist in the older children as
this gives us additional clues about the specific challenges Thai orthography poses
to the child. The persistent errors found in the current study for Thai children
predominantly involved the incorrect selection or usage of homophonous consonants and the incorrect expression of tone. Older children’s errors typically had the
correct vowel expression but the consonant and/or tone was incorrect. The tone
rules for Thai are complex and irregular, and hence these persistent errors confirm
that tone rules are problematic for beginning readers. However, it is important to
consider that in the tasks used in the current study, children were required to read
and spell single words in isolation, whereas when reading words more naturally in
continuous text, readers use contextual cues and top-down processing to aid in the
recognition of words.
As Thai also does not have interword spaces and has multifunction words that
change function dependent on context, we can expect Thai readers to be fairly
reliant when reading continuous text on top–down processes to both segment the
text into words and for word identification. Thai readers may not need to have
detailed knowledge of the tone rules and use bottom–up processes for identifying
lexical tone on a syllable when reading continuous text, as instead, they can use
contextual cues to assign the correct tone to the word or syllable. In Hebrew, it has
been found that the cost of reading words without vowels (as occurs in adult text) in
isolation disappears once readers are presented with sentences or continuous text, as
they are then able to use morphological information that is available in full
sentences (Frost, 2006). This could also be the case for reading lexical tone by Thai
readers.
There is a commonly held view that there are shared representations that underlie
both the reading and spelling acquisition process (Curtin et al., 2001; Holmes &
Carruthers, 1998), however, maybe we need to modify this claim at least in relation
to orthographies that are more irregular or unbalanced in terms of PG and GP
correspondence rules. The difference in proportions of lexical and phonological
errors made in the spelling and reading tasks suggests that there are qualitatively
different strategies and processes being utilized for reading and spelling by
123
1046
H. Winskel, K. Iemwanthong
beginning readers of Thai, which changes with age as orthographic and phonological knowledge develops. In Persian orthography, which is also more inconsistent in
terms of phoneme-grapheme correspondences than grapheme-phoneme correspondences, it has been found that different processes are involved in reading and
spelling. Persian adults relied more on lexical processes to read words, whereas
phonological processes were more dominant for spelling words (Rahbari &
Sénéchal, 2009), which is similar to results found for Thai.
Different orthographies have unique linguistic characteristics that pose different
challenges to beginning readers and affect the reading and spelling developmental
process (Abu-Rabia & Taha, 2004). Thai has its own distinctive orthography, which
has complex graphemic expressions, particularly in relation to non-linear vowel
combinations and for the expression of tone that operates at the syllable level, and
requires the memorisation of a complex combination of vowels and diacritics. In the
current study, it was found that in comparison to more transparent orthographies
such as German, Welsh, or Indonesian, a relatively high proportion of lexical errors
were produced by the Thai children, particularly by the youngest Grade 1 children.
Nonword reading and spelling proved problematic to even the oldest children, the
Grade 3 children. It appears that children are predominantly using a larger lexical or
syllabic grain size to read both words and nonwords. This is particularly apparent in
the youngest children, but declines in the older children as their reading skills
develop. Spelling was found to lag behind reading in the youngest children, which
reflects the PG mapping imbalance, the multigrapheme to phoneme mapping of
consonants in conjunction with the complex graphemic representation of Thai
vowel and tone expressions. However, these problems were largely overcome by
Grade 2 with relatively few of these error types persisting. The main errors that
persisted were related to homophonous consonants and the expression of tone.
Larger grain size, lexico-syllabic as well as phonemic grain size appears to be
important when reading. This reflects the hybrid characteristics of the orthography.
In order to become a skilled reader of Thai, it appears that a mixed strategy and
ability to use both larger and smaller phonological units is advantageous. In future
studies, it is important to extend this line of research to investigate in greater depth
how the relative imbalance between PG and GP mapping in different orthographies
as well as orthographic-specific characteristics, affects the developmental patterns
and strategies used by children learning to read and spell typologically distinct
orthographies.
Appendix 1
See Table 5.
123
Reading and spelling acquisition
1047
Table 5 Thai initial consonants (IPA symbols are in parenthesis)
Place of articulation
Bilabial
Labio-
Alveola
dental
r
Palatal
Velar
Glottal
Manner of articulation
Voiceless
aspirated
(ph)
(tçh)
(th)
Stop
(kh)
Voiceless
unaspirated
(p)
(t)
(b)
(d)
(m)
(n)
(tç)
(k)
(?)
Voiced
Nasal
( )
Fricative
(f)
(h)
(s)
Lateral
(l)
Trill
(r)
Approximant
(w)
(j)
123
1048
H. Winskel, K. Iemwanthong
Appendix 2
See List 1.
List 1 The words and
nonwords for the vowel reading
task
Words
boat
month
nine
tired
believe
write
brush
empty
week
new
town
candle
string
salt
123
Nonwords
Reading and spelling acquisition
1049
List 1 continued
try
money
change
green
ox
add
curly
jar
web
remember
rotten
snake
bobbin
lean
Appendix 3
See List 2.
123
1050
List 2 The words and
nonwords for the tone reading
task
H. Winskel, K. Iemwanthong
Words
Words
two
picture
work
slave
bank
love
joke
thing
bottle
one
six
before
fry
old
stop
aunt
lonely
is
cold
run
taste
part
keep
grass
want
must
horse
house
water
123
Nonwords
Nonwords
Reading and spelling acquisition
1051
References
Abu-Rabia, S., & Taha, H. (2004). Reading and spelling error analysis of native Arabic dyslexic readers.
Reading and Writing: An interdisciplinary Journal, 17, 651–689.
Alcock, K. J., & Ngorosho, D. (2003). Learning to spell a regularly spelled language is not a trivial task—
patterns of errors in Kiswahili. Reading and Writing: An Interdisciplinary Journal, 16, 635–666.
Borzone de Manrique, A. M., & Signorini, A. (1994). Phonological awareness, spelling and reading
abilities in Spanish-speaking children. British Journal of Educational Psychology, 64, 429–439.
Bruck, M., & Treiman, R. (1990). Phonological awareness and spelling in normal children and dyslexics:
The case of initial consonant clusters. Journal of Experimental Child Psychology, 50, 156–178.
Caravolas, M., Hulme, C., & Snowling, M. (2001). The foundations of spelling ability: Evidence from a
3-year longitudinal study. Journal of Memory and Language, 45, 751–774.
Cassar, M., Treiman, R., Moats, L., Pollo, T. C., & Kessler, B. (2005). How do the spellings of children
with dyslexia compare with those of nondyslexic children? Reading and Writing: An Interdisciplinary Journal, 18, 27–49.
Cho, J. R., & McBride-Chang, C. (2005). Correlates of Korean Hangul acquisition among kindergarteners
and second graders. Scientific Studies of Reading, 9, 3–16.
Cossu, G., Gugliotta, M., & Marshall, J. C. (1995). Acquisition of reading and written spelling in a
transparent orthography: Two non-parallel processes? Reading and Writing: An interdisciplinary
Journal, 7, 9–22.
Curtin, S., Manis, F. R., & Seidenberg, M. S. (2001). Parallels between the reading and spelling deficits of
two subgroups of developmental dyslexics. Reading and Writing: An Interdisciplinary Journal, 14,
515–547.
Defior, S., Martos, F., & Cary, L. (2002). Differences in reading acquisition development in two shallow
orthographies: Portuguese and Spanish. Applied Psycholinguistics, 23, 135–148.
Duncan, L. G., Seymour, M. P. H. K., & Hill, S. (2000). A small-to-large unit progression in
metaphonological awareness and reading? The Quarterly Journal of Experimental Psychology, 53A,
1081–1104.
Ehri, L. C. (1992). Reconceptualising the development of sight word reading and its relation to recoding.
In P. B. Gough, L. C. Ehri, & R. Treiman (Eds.), Reading acquisition (pp. 107–143). Hillsdale, NJ:
Erlbaum.
Fernandes, S., Ventura, P., Querido, L., & Morais, J. (2008). Reading and spelling acquisition in
European Portuguese: A preliminary study. Reading and Writing: An interdisciplinary Journal, 21,
805–821.
Frith, U. (1985). Beneath the surface of developmental dyslexia. In K. Patterson, J. Marshall, &
M. Coltheart (Eds.), Surface dyslexia, neuropsychological and cognitive studies of phonological
reading (pp. 301–330). London: Erlbaum.
Frith, U., Wimmer, H., & Landerl, K. (1998). Differences in phonological recoding in German and
English-speaking children. Scientific Studies in Reading, 2, 31–54.
Frost, R. (2006). Becoming literate in Hebrew: the grain size hypothesis and Semitic orthographic
systems. Developmental Science, 5, 439–440.
Geva, E., Wade-Woolley, L., & Shany, M. (1993). The concurrent development of spelling and decoding
in two different orthographies. Journal of Reading Behaviour, 25, 383–406.
Goswami, U. (2000). Phonological representations, reading development and dyslexia: Towards a crosslinguistic theoretical framework. Dyslexia, 6, 133–151.
Goswami, U. (2003). Why the theories about developmental dyslexia require developmental designs.
Trends in Cognitive Sciences, 7, 534–540.
Goswami, U., Ziegler, J. C., Dalton, L., & Schneider, W. (2003). Nonword reading across orthographies:
How flexible is the choice of reading units? Applied Psycholinguistics, 24, 235–247.
Gupta, A. (2004). Reading difficulties of Hindi-speaking children with developmental dyslexia. Reading
and Writing: An Interdisciplinary Journal, 17, 79–99.
Holmes, V. M., & Carruthers, J. (1998). The relation between reading and spelling in skilled adult
readers. Journal of Memory and Language, 39, 264–289.
Howell, D. C. (2002). Statistical methods for psychology. Pacific Grove, CA: Thompson Learning.
Hudak, T. J. (1990). Thai. In B. Comrie (Ed.), The major languages of East and South-East Asia
(pp. 757–773). London: Routledge.
123
1052
H. Winskel, K. Iemwanthong
Kim, Y.-S. (2007). Phonological skills and literacy skills in Korean: An examination of the unique role of
body-coda units. Applied Psycholinguistics, 1, 69–94.
Kim, Y.-S. (2009). Crosslinguistic influence on phonological awareness for Korean-English bilingual
children. Reading and Writing: An interdisciplinary Journal. doi: 10.1007/s11145-008-9131-0 (in
press).
Lennox, C., & Siegel, L. S. (1993). Visual and phonological spelling errors in subtypes of children with
learning disabilities. Applied Psycholinguistics, 14, 473–488.
Luksaneeyanawin, S. (2004). The Thai one million word database. Bangkok: Chulaglongkorn University,
Center for Research in Speech and Language Processing (CRSLP).
Öney, B., & Durgunoğlu, A. Y. (1997). Beginning to read in Turkish: A phonologically transparent
orthography. Applied Psycholinguistics, 18, 1–15.
Padakannaya, P., Rekka, D., Vaid, J., & Joshi, M. (2002). Simultaneous acquisition of literacy skills in
English and Kannada. A longitudinal study. Poster presented at the 13th World Congress of the
International Association of Applied Linguistics, Singapore.
Perry, C., & Ziegler, J. C. (2004). Beyond the two-strategy model of skilled spelling: Effects of
consistency, grain size, and orthographic redundancy. The Quarterly Journal of Experimental
Psychology A, Human Experimental Psychology, 57A, 325–356.
Perry, C., Ziegler, J. C., & Coltheart, M. (2002). A dissociation between orthographic awareness and
spelling production. Applied Psycholinguistics, 23, 43–73.
Pinheiro, A. M. V. (1995). Reading and spelling development in Brazillian Portuguese. Reading and
Writing: An Interdisciplinary Journal, 7, 111–138.
Porpodas, C. D., Pantelis, S. N., & Hantziou, E. (1990). Phonological and lexical encoding processes in
beginning readers: Effects of age and word characteristics. Reading and Writing: An Interdisciplinary Journal, 2, 197–208.
Rahbari, N., & Sénéchal, M. (2009). Lexical and nonlexical processes in the skilled reading and spelling
of Persian. Reading and Writing: An interdisciplinary journal. doi: 10.1007/s11145-008-9122-1
(in press).
Rahbari, N., Sénéchal, M., & Arab-Moghaddam, N. (2007). The role of orthographic and phonological
processing skills in reading and spelling of monolingual Persian children. Reading and Writing:
An interdisciplinary Journal, 20, 511–533.
Rickard Liow, S. J., & Lee, L. C. (2004). Metalinguistic awareness and semi-syllabic scripts: Children’s
spelling errors in Malay. Reading and Writing: An Interdisciplinary Journal, 17, 7–26.
Schoknecht, C. (2000). Appendix: Three systems of Thai transcription. In D. Burnham, S. Luksaneeyanawin, C. Davis, & M. Lafourcade (Eds.), Interdisciplinary approaches to language processing:
The international conference on human and machine processing of language and speech (pp. 329–
336). Bangkok: National Electronic and Computer Technology Center.
Seymour, P. H. K., & Elder, L. (1986). Beginning reading with phonology. Cognitive Neuropsychology,
3, 1–36.
Seymour, P. H. K., Aro, M., & Erskine, J. M. (2003). Foundation literacy acquisition in European
orthographies. British Journal of Psychology, 94, 143–174.
Spencer, L. H., & Hanley, J. R. (2003). Effects of orthographic transparency on reading and phoneme
awareness in children learning to read in Wales. British Journal of Psychology, 94, 1–28.
Sprenger-Charolles, L., & Siegel, L. S. (1997). A longitudinal study of the effects of syllabic structure on
the development of reading and spelling skills in French. Applied Psycholinguistics, 18, 485–505.
Sprenger-Charolles, L., Siegel, L. S., & Bonnett, B. (1998). Reading and spelling acquisition in French:
The role of phonological mediation and orthographic factors. Journal of Experimental Child
Psychology, 68, 134–165.
Stuart, M. (2005). Phonemic analysis and reading development: some current issues. Journal of Research
in Reading, 28, 39–49.
Stuart, M., & Coltheart, M. (1988). Does reading develop in a sequence of stages? Cognition, 30, 139–
181.
Thorstad, G. (1991). The effect of orthography on the acquisition of literacy skills. British Journal of
Psychology, 82, 527–537.
Treiman, R. (1993). Beginning to spell: A study of first grade children. Oxford: Oxford University Press.
Treiman, R. (1997). Beginning to spell in English. In C. Hulme & R. M. Joshi (Eds.), Reading and
spelling: development and disorders (pp. 371–393). Mahwah, NJ: Lawrence Erlbaum Associates.
Treiman, R., & Weatherston, P. (1992). Effects of linguistic structure on children’s ability to isolate initial
consonants. Journal of Educational Psychology, 84, 174–181.
123
Reading and spelling acquisition
1053
Treiman, R., & Zukowski, A. (1988). Units in reading and spelling. Journal of Memory and Language,
27, 466–477.
Treiman, R., & Zukowski, A. (1996). Children’s sensitivity to syllables, onsets, rimes and phonemes.
Journal of Experimental Child Psychology, 61, 193–215.
Treiman, R., Kessler, B., & Bick, S. (2002). Context sensitivity in the spelling of English vowels. Journal
of Memory and Language, 47, 448–468.
Vaid, J., & Padakannaya, P. (2004). Reading and writing in semi-syllabic scripts: An introduction.
Reading and Writing: An Interdisciplinary Journal, 17, 1–6.
Wimmer, H., & Goswami, U. (1994). The influence of orthographic consistency on reading development:
Word recognition in English and German children. Cognition, 51, 91–103.
Wimmer, H., & Hummer, P. (1990). How German speaking first graders read and spell: Doubts on the
importance of the logographic stage. Applied Psycholinguistics, 11, 349–368.
Winskel, H. (2009). Reading in Thai: The case of misaligned vowels. Reading and Writing. An
Interdisciplinary Journal, 22, 1–24.
Winskel, H., & Widjaja, V. (2007). Phonological awareness, letter knowledge and literacy development
in Indonesian beginner readers and spellers. Applied Psycholinguistics, 28, 21–43.
Worthy, M. J. (1990). Spelling errors of normal and disabled students on achievement levels one through
four: Instructional implications. Annals of Dyslexia, 40, 138–151.
Ziegler, J. C., & Goswami, U. (2005). Reading acquisition, developmental dyslexia, and skilled reading
across languages: A psycholinguistic gram size theory. Psychological Bulletin, 131, 3–29.
Ziegler, J. C., & Goswami, U. (2006). Becoming literate in different languages: similar problems,
different solutions. Developmental Science, 9, 429–453.
Ziegler, J. C., Stone, G. O., & Jacobs, A. M. (1997). Phonology can help or hurt the perception of print.
Journal of Experimental Psychology: Human Perception and Performance, 23, 845–860.
123