Children entering their teen years are increasingly at home and adept with computer-mediated communication (CMC), and indeed, they and their younger siblings have often been born into homes where CMC technology is a daily feature of their lives. Today's English speaking children in the UK and the US, across the demographic range, have been referred to as “Generation Txt” (Thurlow, 2003), “Digikids” (Marsh, 2005), and “The Net Generation” (Rosen 2007), indicating their comfort with new media as a defining characteristic of the age. Other work in this volume indicates that the descriptions might also apply in other cultures and languages. Developing literacy in this world of multiple text formats requires a wider set of skills than earlier generations required, if children are to grow to fit, and indeed contribute to, the digital world in which they will work and play as adults. One of the most common digital practices young people mention is text messaging, using their mobile phones for SMS communication.
Research on text messaging around the world (see Thurlow and Poff, forthcoming, for a review) has tended to focus either on the uses to which texting has been put, or on the language itself, rather than linking text message language with other literacy measures. It has also focused primarily on teen and adult texters. Our concern has been to demonstrate the ways texting is related to the traditional literacy skills that are at the core of the school curriculum for these preteen children. We review here our work in an ongoing program exploring those relationships.
There is not complete agreement about what “literacy” means, whether it refers to reading and writing printed language (Kress, 2003), or whether it might refer to a wide spectrum of communicative practices (Cope and Kalantzis, 2000). Here, we take the position that literacy is wider than reading and writing in traditional print genres, and think of literacy as the ability to decode information in various orthographic formats, including digital media, to make and take meaning from it, and to encode information into those formats to communicate ideas to others. Our view of literacy does remain close to the written word, however, rather than refer to the wider range of communicative practices that may not explicitly include written language. Much of children's wider literacy experience may be in print or visible text, from comics, graffiti, and road signs to cereal boxes and manhole covers, from computer instructions and instant messaging to mobile phone screens, and some may be multimedia literacy, with visual, spatial, and auditory components, such as computer games. Some may be with books and toys that are commercial outgrowths of stories and information first encountered through visual media of television and DVDs.
Children's exposure to ambient environmental print has been seen as an important adjunct and precursor to traditional school literacy (Weinberger 1996), and the extended experience of today's digikids means that children may be exposed to more words in the environment than ever. Even graffiti may be digital in origin if the perpetrator has access to a “guerrilla projector”.1 Our focus here is on digital literacy, the medium of text messaging, or Short Message Service (SMS), mediated through mobile phones. To disambiguate “text” when it refers to the wider literary context of the term “text,” from “text” when it refers to a text message, the orthographic form for the text message will be “txt” following the text-messaging children's frequent designation, with apology to those who despair for the English language. There is precedent for this usage in otherwise formal register text (Graham 2008, p12). Sending text messages will be referred to as “texting.”
At the beginning of 2007, it was estimated that more than one-third of Earth's population had mobile phone access, with 2.73 billion mobile phone subscriptions (Baron, 2008). In September of 2007, the BBC reported that nearly 5 billion text messages (txts) were sent in Britain alone, some 4,000 per second. At the end of 2007, for the first time, Americans sent more txts on their mobile phones than they made calls (Mindlin 2008). The UK Mobile Life Report (2006) found that 51% of 18–24 year olds sent or received at least 6 txts a day. According to the Ofcom Media Literacy Audit (2006) of over 1500 UK children between 8 and 15, 49% of 8- to 11-year-olds had their own mobile phone, compared with 82% of 12–15 year olds. Our own research in the UK (Plester, Wood, and Bell, 2008) found 78.5% of 11- and 12-year-olds had regular or exclusive use of a mobile phone. In a survey in 2008, part of ongoing research, 93% of 227 UK 9- to 11-year-olds either had their own mobile phone (69%) or could use someone else's freely, and 54% reported that they used the phone for texting more than talking; only 26% reported talking more than texting. No significant gender differences emerged in this survey.
The 8- to 11-year-old group in the Ofcom 2006 report reported making an average of six calls per week, but sending 16 txts in the same time span. The 12- to 15-year-old group made an average of nine calls per week, but sent 31 texts. Texting was the primary use of the mobile phones for both groups, with 82% of the younger group texting, and 93% of the older group. In 2007, a group of young U.S. teens spoke in a public forum (Kids Speak Out) about their digital lives; many mentioned texting and some acknowledged a somewhat excessive enthusiasm for it. When texting has become such a widespread and valued CMC activity among the young, and when the age of receiving a first mobile phone falls significantly year on year (Plester, Wood, and Joshi, 2009), the language practices of this net generation demand investigation. Texting has become a significant part of British preteens' literacy experience.
There are several ways that texting might be seen to affect literacy development. One is increased exposure to text and engagement with written language. A second is related to the phonological awareness that underpins many of the text language abbreviations (textisms) used. A third is related to enjoyment of words, the playful mix of written and spoken language features that characterise text language. We will explore each of these factors and then summarise our research attempting to establish how various factors involved with texting relate to traditional literacy skills. Then we will discuss implications for bringing together children's out of school texting literacy with their in-school new media and traditional literacies.
Exposure to Text
A growing area of research with younger children concerns new media literacies where young children have developed skills well before they reach formal schooling (e.g. Yelland, 2005; Marsh, 2004, 2005). Marsh describes work with a demographic range of UK families of preschoolers identifying a wide variety of new media literacy experiences the children have. These young children often have toy computers and over 80% of the children she studied had toy mobile phones from an average of 12 months of age. Toys act to introduce children to the material artefacts of the world in which they will grow up. But they also have early experience with the genuine technology, participating in computer games with parents or older siblings. They choose and play DVDs and learn to read the covers of DVD cases. Revelle et al. (2007) report a U.S. study with families at risk of poor literacy development, using mobile phones to engage preschoolers in learning the alphabet by sending their parents txts with suggestions for informal teaching in daily activities. The children themselves learned to operate the mobile phones to play and to find and replay video files related to the letters they were learning, and were aware that the phone screens also displayed written words in txts that brought messages to their parents. The children and parents had increased the number of literacy related activities which they undertook by the time the study had concluded.
Children such as these are ready to enter school with an unprecedented amount of new media literacy experience and skills. Many writers (e.g. Yelland, 2005; Jenkins, 2008) have challenged the traditional discourse of school literacy, which privileges the printed word on paper over all other genres of text, encouraging educators to acknowledge the new media literacy expertise already developed by young children when they first enter primary school, and to build on that through the literacy learning in both traditional and digital formats in the contemporary curriculum. With the challenge to traditional discourse on literacy, educators have been quick to point out that critical appraisal need not mean rejection of the traditional (Marsh, 2005; Rowan and Honan, 2005), but that it should be re-presented within the context of children's out of school lives. Today's teachers may be uniquely fitted to making that link between in and out of school digital literacy experiences, having been characterised as Digikids themselves (Graham, 2008).
The typical UK child has frequent contact with CMC, or ICT (Information and Communication Technology), in school, with many primary schools having well-equipped computer learning suites, as well as computers in classrooms, although there is still tension between traditional literacy exponents and those who would integrate traditional with digital texts. There has been research linking digital experience with traditional literacy achievement. Early studies of e-learning in schools seemed positive (e.g. Kulik and Kulik, 1989), although Leu (2002) pointed out that these were often not directly concerned with the development of literacy. Recent analyses are more measured and inconclusive about the effects of educational technology overall (e.g. Abrami et al., 2008). UK classrooms that feature in research cited here are English speaking, but often will contain children whose first language was not English. In our own research we have found no effects of ethnicity or other demographic factors on children's engagement with CMC, and we have worked with classrooms across a wide range of socioeconomic and ethnic profiles.
Contemporary children have exposure to written language in a variety of media both in school and out, with possibly more exposure than ever before, and exposure to print has been robustly associated with ease of learning to read. Cipielewski and Stanovich (1992) demonstrated that American children's reading ability in fifth grade, 10 or 11 years old, roughly the age of most of our participants, was predicted strongly by their measure of text exposure, an Author Recognition Test, after earlier reading ability and orthographic decoding skill were accounted for. Stainthorp (1997) produced a Children's Author Recognition Test as a British equivalent, which was also shown to predict reading. In these measures, children are exposed to a list of genuine children's authors' names mixed randomly with a set of genuine names of people who were not children's authors, and the children had to indicate which they knew as authors. These measures used author recognition over title recognition as the criterion, on the basis that stories may have been encountered in television or film adaptations, but knowledge of the author's name made it more likely that the child had encountered the stories in print. More recently, Ecalle and Magnan (2008) found that an indication of print exposure among French children predicted word reading in younger children, and word recognition, spelling, and vocabulary in older children. They used recognition of titles and magazines as well as authors in their measure, however, and controlled for age and phonological and orthographic skills.
Wolf (2007) discusses fluency in reading text, and the cognitive advantage that ensues when working memory capacity is used less for decoding orthographic representations and freed for reflection on the semantic content of the text. Her argument brings physiological backup to the “Matthew Effect” (Stanovich, 1986), that fluency yields enjoyment and ease of comprehension, which goes on to encourage further exposure, and difficulties decoding hinder fluency, reduce enjoyment and comprehension, and discourage further exposure. It follows that the wider exposure of the Digikids, the Net Generation, to texts in various media, should provide enhanced literacy development at school. But young children's immersion in CMC is a relatively recent phenomenon, and there is little empirical research that has attempted to link it with traditional literacy development. Much discussion has centred on the implications but little on the actual outcomes.
An alternative perspective on the frequency of encounter with written language is raised by Baron (2008). She acknowledges that people's daily experiences are awash with text, a state she refers to as “flooding the scriptorium” (p.198). She makes no comment there about particular sets of people, but the context suggests reference to technologically developed societies, of various languages, since she has studied e-communication in several languages. Aside from becoming comfortable with literacies on every hand, the presence of so many avenues to pursue, there is pressure for instant outcomes and multitasking that is also endemic to Western children's culture. These pressures may render less likely both a motivation to seek precision and correctness in written language, and the reflection and depth of understanding that literacy fluency allows. Carr (2008) writes of developing a “power browsing” reading style in Internet sources that may impair ability to immerse one's self in deep and thoughtful reading. Carr may be referring to older internet users rather than children, in whom “deep and thoughtful reading” may not yet have developed to adult capacity, but the children of our research are old enough to use the internet regularly as an information source, and have informally commented that they do. In 2007, for example, Wikipedia captured 24.3% of all visits to educational resource sites, the most likely to be accessed of over 3,000 sites, and the core of the users were reported to be school children (Tancer, 2007). Darlin (2008) is only partially successful in rebutting Carr's claims, and rests upon the power of technology to save effort in search for information and free our minds for other activity. His argument is reminiscent of the Matthew Effect, but he does not address the quality of thought, only the time spent. Careful empirical research is needed to demonstrate the literacy outcomes for the Net Generation.
We have chosen UK preteen children's mobile phone text messaging (SMS) as the focus for our research because of wide media concern that txt language is a serious threat to traditional literacy, if not to the language itself (Thurlow, 2006), and because very high proportions of preteens have only recently had access to mobile phones. We have chosen to focus on empirical measures here, rather than explore the qualitative material that would give a wider and more comprehensive context within which these children use their phones, because the concern has been expressed strongly in the media that traditional literacy achievements and skills are in jeopardy. Clear measures of achievement and skills in the children we have worked with address those concerns most directly.
Text Message Language
Text messaging has characteristics of both speech and written language. It is spontaneous and loosely structured, as is speech. Although it is not time bound, it is space bound, visually decontextualized and immediately revisable, as is written language (Crystal, 2006a). Texting can be a monologue, since it is not time bound, and so more like writing; it often uses longer units of expression than, for example instant messaging, which is time bound and conversational (Ling and Baron, 2007; Baron, 2008). Texting, however, uses contractions, and colloquial language like speech. This crossing over of communication modes has led 50% of young adults to prefer to text rather than talk (Reid and Reid, 2005), and we have seen that 62.7% of preteen respondents generally preferred txt to talk (Plester, Wood and Bell, 2008).
The enthusiasm of the young for texting has not always been matched by supportive public adult commentary. Thurlow (2006) analyzed over 100 media articles referring to text messaging, identifying several themes of importance to journalists. Most of the themes reflect highly critical, sometimes exaggerated views of texting, often based on anecdote and speculation. Fortunately, there are more positive voices as well. O’Connor (2005) cites a number of educators who value children's text messaging as voluntary engagement in written communication. Crystal (2005, 2006a, 2006b, 2008) has been consistently positive about the robust quality of the English language, and the advantages of creativity, flexibility, and metalinguistic knowledge to accrue from the ludic, enjoyable, and playful use of text language. Typically, as a casual register similar to conversational language, txts may use abbreviations and other nonstandard forms of words. We refer to these text language variants as textisms, and will discuss them in more detail.
Katz and Aakhus (2002) have invited expert-framed academic discussion about the use of technology in communication, to counter the weight of anecdote and opinion. They have asked for data-driven conclusions, and research on texting has ensued, although it has tended to focus on the social uses of texting (e.g. Reid and Reid, 2005), or the linguistic aspects (Thurlow, 2003; Crystal, 2006a, 2008; Ling and Baron, 2007; Baron, 2008; Thurlow, 2009).
Thurlow (2003) analyzed a body of txts produced by British young adults for the types of variant word forms used. He found about 19% of the words in the collected txts to be textisms of one sort or another, but other reports have seen smaller proportions, e.g. about 10% (Crystal, 2008), and less than 5% (Baron, 2008). These figures do not reflect the concern in the media that, for example, Sutherland (2002) showed in describing text messages as “bleak, bald, sad shorthand” and “Linguistically, it's all pig's ear.” Thurlow, however, did not give frequency data for all of the types of textisms they identified, although they, like Ling and Baron (2007), found few emoticons and symbols used. In the investigations we report below, we include frequency data for various types of textism.
If we look at the generating principles behind the types of textisms that Thurlow (2003) identified, we are drawn to another possible positive relationship between texting and traditional literacy. In total there were 12 categories used: shortenings (bro, sis, tues), contractions (txt, plz, hmwrk), G-clippings (swimmin, goin, comin), other clippings (hav, wil, couldn), omitted apostrophes (cant, wont dads), acronyms (BBC, UK), initialisms (ttfn, lol, tb, WUU2), symbols, (@, &, :-o), letter/number homophones (2moro, 18r), misspellings (comming, are [for our], where [for were]), nonconventional spellings (fone, rite, skool), accent stylisation (wiv, elp [help], anuva, wanna, dunno).
Of these types of textisms, several are reliant on the creator and receiver of them having good phonological awareness and alphabetic knowledge. One of the most obvious examples of this is the homophone textism, where the phonemes used when naming the numerical and orthographic symbols are borrowed to generate textisms. These types are familiar to many readers from childhood “rebus” puzzles. Another is nonconventional spellings, where legitimate spelling rules in English are used to generate textism spellings that are generally shorter than the traditionally spelled words they represent. To do this, the user has to be aware of the multiple phoneme-grapheme mappings permissible in English.
Txt abbreviations were first generated to allow more semantic content into the 160 characters allowed per message than traditional spelling would allow. Contractions also generate shorter spellings that pick up sufficient phonemes to allow the reader to understand the intended meaning. Perhaps the type of textism that most clearly demonstrates the keen phonological awareness of users is accent stylisation, where the common pronunciations of casual register English, slang, or “street speak,” are rendered accurately using legitimate phoneme-grapheme conversion rules. Where a strict traditionalist might cringe to hear a child say “Wot ya doin, bro, wanna cum wiv us to da pix?,” a txt that spells it out exactly as it is pronounced [probably including 2 for to] is a clear indication of an accurate ear for speech sounds. G-clippings are another recognition of casual speech sound. The accent stylisations we have coded are specific to UK Midland English, but English spoken in other countries or even parts of the UK would likely have other variants of pronunciation that would be encoded for txts by similar processes.
Phonological awareness has been the focus of reading research for a number of decades now, and such research has demonstrated a consistent association between different forms of phonological awareness and reading attainment (e.g. Adams, 1990; Snowling, 2000). Researchers have shown that children who are given training focused on improving phonological awareness perform significantly better on reading tests than those who do not receive such training (e.g. Blachman, et al., 1994; Bradley and Bryant, 1983). Arguably, the word play that children have in texting offers a rudimentary and informal learning platform from which they can develop sensitivity, confidence, and flexibility with this phonetic language. The sophisticated manipulation of language as achieved through functional practice and active experience gained through texting, may provide a transferable skill concerning the ready application of enhanced phonological knowledge and thus aid standard literacy development. Part of our research goal has been to show how the phonological skills demonstrated through the use of textisms in txts might be related first, to school literacy attainment, and specifically to measures of phonological skills that have been widely used in literacy research.
Play with Language
Equally, release from the constraints of regulated orthographic and spelling conventions in the context of texting carries a sense of play with language, a pursuit that has long been part of human language use, as well as a pervasive ludic human disposition to play when one can (Crystal 2006a, b, 2008; 1998). Crystal celebrates the extraordinary lengths to which people go to have fun with words, citing texting as a modern example. It is also possible that the relatively free play with linguistic materials might echo the effect found by Sylva, Roy, and Painter (1980). They found that children who were allowed free play with physical materials, freed from pressure to follow instructions or copy a model's actions, were more able to use those materials in solving a novel problem than children who had only been instructed or shown. Children may gain both pleasure and confidence in communicating through writing when they can do so in a playful setting where the conventions that normally apply are set aside. Play minimises the consequences of one's actions and allows learning in a less risky situation (Bruner, 1972). That intentional play with language should enhance metalinguistic, including phonological, awareness is not a new notion (Cazden, 1973), and reflects Vygotsky's (1962) view that progress in literacy has a reciprocal relationship with making previously intrinisic knowledge extrinsic, both resting on the process and contributing to it. A more recent iteration of that process may be seen in Karmiloff-Smith's (1992) model of representational redescription. And whatever its possible constructive outcomes, play is still a source of enjoyment.
Engagement with Written Language through Invented and Playful Spelling
A further factor that may link play with language to exposure and engagement with written language is children's invention of spellings when they lack knowledge of conventional word forms. “I chry to dethyn the brayn” as a caption on a 5-year old's drawing of a brain reflects an eagerness to engage with written language, and some knowledge of conventional phoneme-grapheme correspondence rules in English as well as phonological awareness, including that of a juvenile tongue thrust still in evidence in spoken language. Read (1971, 1975) discussed this tendency of children to experiment with writing as precursor to more conventional renderings, as children's phonological awareness and knowledge of correspondence rules became more sophisticated. Richgels (1995) emphasizes that this is a spontaneous creative activity of engagement with the written word, not an explicitly learned set of behaviours. Recently Ouellette and Sénéchal (2008) studied English speaking Canadian 5-year-olds who were asked to try their best to spell words after hearing them repeated and seeing pictures of referents, then given feedback. They were compared with peers who were given explicit phonological awareness training or asked to draw pictures of the referents. The invented spelling group learned more new words than their peers. The authors proposed that the active engagement of the invented spelling group, given age-appropriate feedback, led them to be more able to take an analytical approach to phonological and orthographic knowledge that would help them spell. Preteen texters may have mastered many of the conventions of English spelling, but they also vary in their willingness to recoup that early earnest engagement in a playful mode, and that willingness to play with invented spellings may also be related to older children's literacy skills.
Stephen et al. (2008), having interviewed preschool children about their encounters with digital literacy at home, found that they tended not to associate home use of ICT with learning, much as Maddock (2006) had found with older primary school children. Outside of school literacies are for leisure or play, inside school is for learning. The same children use their skills in both places, however, and research has focused on one area or the other, not relating children's skills in one area to their skills in the other, perpetuating the dissociation the children themselves have indicated. As there has been little research regarding relationships between txt literacy and more traditional literacy achievements, and we would not like to see educational policy formulated on the basis of alarmist media reports and opinion that is anecdotally informed, we have begun a program of empirical research to investigate those relationships.
The first study we present (Plester, Wood and Bell, 2008) investigated relationships between texting and school language achievement among pre-teen mobile phone users. Previous research has focused on teen and older phone users and has tended to focus on social aspects of their use. We asked 65 11- to 12-year-olds in the UK about their mobile phone use, and then related their level of text messaging to their scores on the Cognitive Abilities Test (CAT) in verbal and nonverbal reasoning. The CAT correlates highly with scores on the UK Key Stage II and III English scores (Strand, 2006).2 The children were divided into three groups of mobile phone users in terms of the median number of txts sent per day: High users reported sending three or more per day; low users sent one or two, and there was a No Txt group. This would place the high text users at or above the level of use reported in the Ofcom (2006) audit for 8- to 11-year-olds.
We measured the extent of textism knowledge and use through a translation task, where the children translated one sentence from text register language into standard English, and one from standard English into text language. Children's responses concerning the translation of txt to standard English were then analysed for errors of grammar, spelling, and punctuation, whilst the sentence in translation from standard English to txt was scored for the ratio of textisms to total words. We did not categorise types of textisms used, as the translation task would have prioritised some over other types of abbreviations.
The findings revealed a negative relationship between performance on the CAT and reported txt frequency. Posthoc tests revealed the high-text-use group scored significantly lower in the CAT than the no-text user group, whilst the level of text message use was not found to be associated with the use of textisms in the translation task. Indeed, the ratio of textisms to real words was similar across all groups, about 58%. Furthermore, there was no evidence that texting group had an effect on the errors that children made when they translated from text language to standard English. Further analysis explored any associations there may have been between the results of the text translation task and the children's CAT performance. A significant positive correlation was found between the proportion of textisms that children used and children's verbal reasoning scores. That is, those children who were using a higher proportion of textisms were those with higher scores on the verbal reasoning tasks.
Following the identification of such patterns, further analysis was conducted through hierarchical regression to see whether the ratio of textisms to real words was able to explain individual differences in the verbal CAT scores, whilst controlling for the number of texts sent per day. The number of texts that children were sending each day was not able to account for a significant proportion of a variance in children's verbal scores. Interestingly however, the ratio of textisms to standard words that children used in translation into text language was found to account for a significant 12.4% of the variance in verbal CAT scores.
This first investigation produced mixed results, in that the higher-txt users scored lower than the no-txt users on the CAT overall, a result that would have satisfied those who worried about negative effects. But further analysis showed a positive relationship between the proportion of textisms used and verbal CAT scores, and indeed, that the ratio was a significant predictor of verbal CAT scores. The impact of this first study is limited, because no cause and effect conclusions can be drawn from its design. There is no evidence that high-txt use causes low CAT scores, or that low CAT scores cause high txt use, or that a high proportion of textisms causes high verbal CAT scores, although it is plausible that higher verbal ability might cause greater ability to play with words in creating textisms. We acknowledge that the high proportion of textisms is likely to be an artefact of design, since the children were asked to translate a sentence into txt language, rather than submit actual txts they had sent. Nonetheless, the study provided a platform from which to explore further. We found no gender differences in this study.
The investigation that followed on (Plester et al., 2008, Study 2) made several design changes. We tested children of 10–11, on the basis that the age of receiving mobile phones was decreasing (Mobile Life Report, 2006). They came from two primary schools with quite different socioeconomic catchment areas. We used the Key Stage II English achievement scores on which the children had just been assessed. We wanted to see if the associations previously found would continue with a more direct and specific indication of the kind of writing and spelling abilities the children would demonstrate in school literacy work. We also used more lengthy translation exercises to elicit txts, which would give the children more latitude in choices of textism types, allowing us to begin to measure frequencies of various types. These were, rebus, or letter/number homophones (C U L8R), other phonological reductions (nite, wot, wuz), symbols (& @ +), acronyms (WUU2, what you up to, actually a composite of acronym and homophone, but reduction by initial letter appears first), and accent stylisation, or “youth code” (wanna, gonna, hafta, me bro, da), a phonetic rendering of casual language style. These are a subset of the categories used by Thurlow (2003), used because the number of elicited txts was relatively small.
Figure 1 shows the mean number of textisms of various types used by the children in all their translation exercises. This is not exactly comparable to an analysis per message, as we show below in Figures 2 and 3 from the following study reported, but a rough comparison of proportions of types can be made to the per message usage of various types of textism. The ratio of textisms to words was .50 (sd = .13) in these exercises, not unlike the 58% we found in shorter translations.
The most frequently used textisms were phonological reductions and the rebus/homophone types. The acronym/homophone blend WUU2 was used by a large number of children in one school and by only one in the other, suggesting the development of local “dialects” of textisms. That phonological textisms were used so often suggests that the children were quite adept at playing with the sounds of language, that they could expect a reader to be as well, and that txt language, through its phonological nature, could raise awareness of core literacy skills. But it might also have negative implications for the children's spelling skills, so we examined their Key Stage II spelling scores. We found that use of both phonological reductions and accent stylisation textisms were correlated positively with spelling scores. Those who used the most phonologically based textisms were the best at spelling. A multiple regression analysis was then performed to investigate the amount of variance in spelling ability that could be accounted for by these two factors. These two factors were found to account for 32.9% of the variance in spelling ability. In addition, there was a significant positive correlation identified between spelling performance and the proportion of textisms to real words that children used. Spelling skill was negatively correlated with the number of errors made in translating txt to standard English. In the first study, using the CAT verbal reasoning scores, we had no standardized spelling measure with which to correlate translation errors.
As spelling ability was revealed here to have a beneficial association with the children's ability to interpret txt language, we also investigated children's writing performance with regard to their proportion of textism use. Analysis of variance showed a strong main effect of the level of Key Stage II English measures in the children's use of textisms and interpretation errors. These were relatively able children, scoring either Level 4 or 5 in the Key Stage II English measures, the top two bands of scores. Those children who scored at Key Stage II Level 5 produced a lower number of interpretation errors than those with Level 4 scores. Those children at Level 5, with a better written performance at Key Stage II, were also found to show a greater ratio of textisms to real words in the English to txt translation task and showed a greater use of phonological reductions and acronyms.
Again we found a positive relationship between children's use of text abbreviations and their verbal skills, this time with their Key Stage II English scores. These children also demonstrated metalinguistic knowledge in that they could slip between standard school English when it was required in the Key Stage II tests, and casual register text language when that was appropriate. Informally, the children would often laugh and insist that of course they would not use text language in their school work, and that they thought that was silly.
We have found no support in our research for the negative media and public speculation surrounding mobile phone use, text language and its effects on literacy development. The direction of the relationships identified between knowledge of textisms and language competence showed no suggestion of negative associations. Such findings support the views of writers such as O’Connor (2005) and Crystal (2006a, 2006b, 2008), who has consistently expressed positive views about the resilience of English in discussing types of language used in informal media across the range of computer mediated communication.
With two initial studies showing positive associations between texting and other measures of language use, we carried out a further investigation to try and unpack those associations with the use of more specific language skill measures. The third study we report here (Plester, Wood, and Joshi 2009) also addressed reading skill, adding measures of word and nonword reading and vocabulary to reach other aspects of children's overall literacy competence. We also added measures of short term memory and phonological awareness, so that each of these individual difference factors might be partialled out during data analysis, allowing us to isolate textism knowledge and use by holding constant these other relevant factors.
A third investigation involved 88 UK children from year 6 and year 7, with a mean age of 10.6 years. Children were tested individually on the following measures: The British Picture Vocabulary Scales II, (Dunn, Dunn, Whetton and Burley, 1997) and the Forward and Backward Digit Span Subtests from the British Ability Scales II (Elliot, Smith & McCulloh, 1996). Because of the importance of phonological awareness in literacy skills, children's phonological awareness was measured by the Spoonerisms subtest from the Phonological Assessment Battery (Frederickson, Frith & Reason, 1997) and the Elision subtest from the Comprehensive Test of Phonological Processing (Wagner, Torgeson, & Rashotte, 1999). The Nonword Reading subtest from the Phonological Assessment Battery (Frederickson et al., 1997) was used to provide a measure of alphabetic decoding ability. Furthermore, as texting can be viewed as having features similar to both writing and speaking (Crystal 2006a), the Word Reading and Spelling subtests of the British Ability Scales II (Elliot et al., 1996) were administered.
We also expanded the elicited text message tasks beyond translations. We had been unsuccessful in collecting a large enough body of naturalistic txts from the children in the earlier studies, because they often forgot to save them, or write them down, or to bring the scripts to school. We constructed 10 scenarios, e.g. “It is Tuesday. You just got home from school, and you have so much homework to do that you don't think you will be able to go to the club you usually go to on Tuesday nights, but you know one of the others in the club will be coming by to pick you up. [You decide what kind of club: swimming, judo, tennis, music, Scouts, Guides, the local youth club.]” The children were asked to write the txt they would send if this happened to them, to write it just the way it would look on the mobile phone screen. The txts were coded for types of textism used, the frequency of each type, and the ratio of textism to total words in each message. We anticipated that this more naturalistic method of eliciting txts would give a proportion of textisms that would be closer to spontaneous txts. The children adapted to writing txts enthusiastically rather than keying them as they normally would.
Further analysis calculated correlation coefficients between the measures of text knowledge, the age at which children received their first mobile phone, and the other cognitive and literacy measures. Word reading ability and performance on the Spoonerism test were found to be significantly associated with the age at which children received their first mobile phone and with the proportion of textisms used in their text messages. The younger they had received a phone, the better their performances were, and the more textisms they used. In addition, vocabulary was also found to be positively associated with the proportion of textisms used. However, no associations were identified between the textism measures and the children's performance in the spelling test in this study.
To examine the individual difference factors, we conducted two hierarchical regression analyses, because strengths or weaknesses in these factors can operate with or against one another on outcomes. We controlled for vocabulary, short-term memory, phonological awareness, the amount of time owning a mobile phone, and finally, children's use of textisms in determining the prediction of variance in word reading ability. The children's knowledge of textisms was able to account for a small but significant 2.9% of variance in word reading scores, in addition to that accounted for by vocabulary, phonological awareness, nonword reading (alphabetic decoding), short-term memory and the age at which children got their first mobile phone. A parallel analysis, substituting age at testing for age of first phone, showed that textism knowledge was able to account for 3.7% of variance in word reading scores. Knowledge of textisms, demonstrated in the proportion of textisms to words used, makes its own independent contribution to word reading in this analysis.
The most popular text abbreviations used by the children were found to be contractions, letter/number homophones, nonconventional spellings and accent stylisations, all with phonological groundings. Figure 2 shows the relative frequencies of various types of textism. The strongest of the nonparametric correlation coefficients we calculated between these different forms of textism and the word and spelling measures, was a positive relationship between word reading ability and the use of homophones, a clearly playful use of the sound of letters and numbers. The mean number of words per txt was 14.96 (sd = 5.23) over 10 scenarios, and the mean ratio of textisms to total words was 0.34 (sd = .17). This is somewhat lower than in the translation tasks of earlier studies, and closer to the ratio reported by Thurlow (2003), although still well above the ratios reported by Crystal (2008) and Baron (2008). It seems likely that the children's knowledge that they were participating in a study about texting primed them to use textisms when they could. A gender difference was observed, as girls had a higher ratio of textisms to total words than boys, F(1,76) = 6.38, p < .05. The girls' mean ratio was .38 (.16) and the boys' ratio was .28 (.17), both higher than would be anticipated in naturalistic txts.
From these three investigations, it became clear that children's knowledge of textisms, their ability and choice to use them in the elicited txt exercises, is an important part of the children's literacy profile. This ability to create alternative orthographic forms of known words, and to know when it is appropriate to do so and when it is not, demonstrates linguistic and metalinguistic knowledge. Because knowledge and use of textisms contributes independently to predicting reading scores, beyond vocabulary, short-term memory, orthographic decoding skill, phonological awareness, length of time owning a phone and chronological age, this indicates that something beyond phonological awareness is going on. Texting may indeed contribute to overall literacy experience in several ways. It is clear also that it does not contribute to the demise of preteen children's literacy skills. Although some of the children had spoken other languages than English before they attended school, all had been active in English language schooling for several years.
The final set of results that we report here relates to 63 UK children from Years 4–7, ranging from 8–11 years of age (Wood, Plester & Bowyer, 2008), and represents interim data analysis from an ongoing project, to which we are still recruiting children. We instituted a number of design changes from earlier studies to improve experimental rigor. The first was that this was a cross-lagged longitudinal study, with measures of individual differences and textism use taken at the beginning of the school year, and again at the end of the school year. This design would enable us finally to look at the direction of causality in any associations that we found. We obtained verbal IQ scores (Wechsler Abbreviated Scale of Intelligence, Wechsler, 1999) in order to control for verbal IQ. We again used the Word Reading and Spelling subtests of the British Ability Scales (Elliot, 1996); and Rhyme Detection, Rapid Automated Picture Naming, Spoonerism, Fluency, and Nonword Reading subtests from the Phonological Assessment Battery (Frederickson, Frith and Reason, 1997). A further design change was that we asked the children to record their txts sent over a single weekend at either end of the school year. We tested over 100 children, but only had full data sets from 63, largely by reason of forgotten txt recording.
We first asked if the proportion of textism use might increase from the beginning of the year to the end of the year. We then asked if the textism proportion at the beginning of the year might predict reading, spelling, and phonological awareness scores at the end of the year, when Verbal IQ and other individual differences were accounted for. We also asked if those scores at the beginning of the year might predict textism use at the end of the year.
We found no significant increase in textism use over the year in any age group, although the older children used significantly more textisms than the younger ones. We also found that textism ratio at the beginning of the year was able to predict significantly both reading and spoonerisms at the end of the year significantly, and spelling skills just short of significance. However, none of the literacy measures at the beginning of the year predicted textism use at the end of the year. This shows us that textism use was promoting the development of literacy skills, not the reverse. With this investigation, we have the first indication of directional influence, the positive impact that the use of textisms has upon literacy measures. With earlier studies, we could not be sure that it was not literacy skills influencing textism use.
What could account for the impact of textism use on literacy measures? Three major factors mentioned earlier appear to be involved: first, increased exposure to the written word, which has been associated with greater literacy skills by other researchers (Cipielewski and Stanovich, 1992; Stanovich, 1986; Stanovich and West, 1989). Increased play and enjoyment of verbal material, which is further active engagement with the printed word, such as has been celebrated by Crystal (2006a, b, 2008; 1998), is very apparent in the children's txts.
The most frequently used forms of textisms were again found to be phonological in nature, showing keen phonological awareness: homophones, accent stylisation, unconventional spellings, and contractions. Figure 3 shows the relative frequencies of types of textism at the beginning of the study. The mean number of words per txt was 22.89 (sd = 19.66) at the beginning of the study, and the mean ratio of textism to total words was 0.36 (sd = .25), and these showed no significant change over the year. The spontaneous txts we analyzed here are longer on average than the previous elicited ones, although that may simply be a function of having had 10 scenarios to respond to in the earlier study. The spontaneous txts used a less ambitious variety of textisms than the elicited texts. This may reflect greater distance from experimenters, yielding less priming, because the spontaneous txts were created away from the school context where the testing was done.
In all three charts, Figures 1, 2, and 3, we see that the most frequently used type of textisms were phonologically based, adding strength to the argument that making use of phonological awareness in a playful context is related to the use of that skill in the more formal context of school literacy measures. When we analyzed the naturalistic txts, which became part of the study only after they had been written, we found that the homonyms, mostly U and R and 2, were the dominant form. This suggests that where the children know that txt language conventions demand play with phonological rules, and can enter into that game when required (as in the translation and scenario exercises), their own spontaneous uses, and perhaps their local txt social conventions are much less inventive, although they still relate positively to literacy measures. The children made use of the txt medium to play with phoneme-grapheme correspondence rules, and particularly to use the almost pun-like nature of homophones. To use those conventions playfully, within the safe domain of texting among peers, implies explicit knowledge of the rules and their conventional nature.
Future research directions
We see the need for further research using broader cognitive and literacy measures, across a wider scope of abilities. Further research will explore the extent to which grammatical and morphological knowledge and skills are related to preteens' use of textisms, and the extent to which these children are aware of the language register conventions of various written genres. Without the constraints of standard English conventions and rules, children are able to experiment and play with language, allowing their phonological skills to guide their experimentations without restriction. Children are in control of this text language and are free to adapt these graphical representations to express meaning how they wish. We have only seen the beginning of the ways in which that expressive language is related to the traditional expectations of English language use. We have not used a sociolinguistic approach to children's txt discourse, for instance, looking at the way children's txts may differ in form and function. We would expect to find, as various other researchers have found with older informants, that girls' txts more frequently function to reinforce social relationships, where boys' txts would be more information centered. There may be differences in form and content if the recipient is a parent or other adult. Our focus has been on the relationship between txt language and traditional language skills and we invite further research into children's txt use.
In children whose lives have been saturated with CMC, for some since their infancy, it will be increasingly important to document the crossover linkages between the written language used for enjoyment outside the school context, and the written language development within the school setting. Anecdote and supposition are not sufficient to inform sound educational practice, nor is it reasonable to hold out of school literacies more or less irrelevant to school contexts of literacy. When ICT is a major part of children's in-school educational experience, they bring into school their skill and expertise in those media, gained for some since toddlerhood. We need to be aware of how those funds of experience are related.
We have only worked with children whose basic literacy skills are already part of their repertoire, but in the light of the growing amount of literature about the CMC literacy experiences that children are having outside school, and even before they attend school (e.g. Marsh, 2004, 2005; Yelland, 2005; Maddock, 2006; Revelle et al., 2007; Stephen et al., 2008), it will be important to continue looking at children's texting in younger and younger mobile phone users. In the third investigation reported here, we found that there was a strong positive relationship between the age the children were at testing and the age at which they had received their first phone. The younger they were at testing, the younger they had been when they received their first phone. The youngest age reported was 6 years old, and the youngest age reported in the fourth study was 5 years old. These are children in whom traditional literacy skills are generally not yet firmly established, and their experience with CMC will be formative in their overall literacy development.
We have not yet worked with children with identified dyslexia, or other literacy development difficulties, and the dynamics of these children's use of text messaging may be very different from those of the children with whom we have worked. We have not explored the impact of socioeconomic or personal choice factors, although if exposure to CMC in out of school settings is shown to have an effect on school literacy attainments, a “digital divide” (Norris, 2001) may show in educational attainment. Norris pointed out that there are geographical dissociations between the information rich and the information poor, but that there are also socioeconomic information rich and poor within nations, and individual divisions between those with opportunity to choose to be information rich, and either do or do not partake of the opportunity.
We do not know what effect texting and other CMC experiences will have on the Digikids' ability to succeed in traditional literacy, or on their motivation to do so. We suggest that texting may provide a means to engage boys in literate activities at a time when boys' slipping literacy attainment is a source of concern to parents, educators and governments (e.g. Bleach, 1998; Maynard, 2001; Sanford, 2005; Mills, Martino, and Lingard, 2007). The enthusiastic inclusion of CMC/ICT in educational settings may reverse the privilege of paper and print that has been debated, and children with strong CMC experience outside of the school setting may enjoy the privilege that children with large family libraries once held. But it will be important to investigate how the skills are linked across the in and out of school settings. In one quite different setting, that of spatial knowledge, one study showed that children whose out of school experience gave them extensive three-dimensional spatial knowledge of their environment, had no advantage over their less knowledgeable peers in geography class (Spencer and Easterbrook, 1985). The curriculum may not meet the children where they have already journeyed in their spontaneous learning experience.
We have a number of investigations in progress to extend the findings of the studies reported here. In one, a microgenetic exploration of the effect of texting, we are comparing a group of 8-to 9-year-olds who do not have mobile phones or use them regularly, with a group of like children to whom we have given phones each Friday, programmed for txt only, to use each weekend. They are compared also with a group of peers who do have phones. They are tested regularly to detect consistent changes in reading or spelling skills. Again, we are harvesting naturalistic txts from these groups by asking them to write down the txts they have sent over a weekend and bring them to school on Monday. In another, we are analysing preteens' own discourse about their CMC experience, to compare it with the discourse of parents and teachers, and with the media (Thurlow, 2006).
We would encourage further research in respect of children's texting, exploring other contributions that it makes to literacy beyond exercise in phonological awareness skills. For example, we might look at the effect of texting on children's grammatical skills, because texting relieves the writer of responsibility for using formal grammatical constructions. We might look at the impact of increased experience with the written word upon the enjoyment of other recreational reading and writing, in a variety of CMC and more traditional media. There may be reciprocal relationships that could be tracked in order to make use of them in enhancing children's literacy development as they engage with CMC throughout their educational careers. Our work has only focused on mobile phones and the whole range of out of school digital experience may provide as rich a harvest of associations with in school literacies. We encourage research linking facets of the various literacies children enjoy in and out of school.