Sarah de Jonge & Nenagh Kemp
Sarah de Jonge & Nenagh Kemp
Sarah de Jonge & Nenagh Kemp
Text-message abbreviations and language skills in high school and university students
Sarah De Jonge
Nenagh Kemp
Address for correspondence: Nenagh Kemp, School of Psychology, Locked Bag 30,
University of Tasmania, Hobart 7001, Tasmania, Australia. Phone +61 3 6226 7534. Fax +61
Acknowledgement
We would like to thank all the participants for texting our messages so enthusiastically. We
are also grateful to Heather Hayes and Brett Kessler for identifying the most difficult items
from their pseudoword list for use in our orthographic awareness task, and to Clare Wood for
.
Texting in high school and university students 2
Abstract
This study investigated the use of text message abbreviations (textisms) in Australian
adolescents and young adults, and relations between textism use and literacy abilities. Fifty-
two high school students aged 13-15 years, and 53 undergraduates aged 18-24 years, all users
of predictive texting, translated conventional English sentences into textese using two
methods; writing messages down, and typing them into mobile phones. Participants produced
a variety of textisms, and in both translation methods, adolescents and young adults used
textisms in nearly identical ways. This was true for the proportion and types of textisms used,
textism categories produced, and consistency with which textisms were spelled. The use of
textisms was negatively correlated with scores for reading, nonword reading, spelling and
morphological awareness, but some of these relationships were accounted for by participants’
usual text-messaging frequency. For these age groups, concerns that frequent texting may
Text-message abbreviations and literacy skills in high school and university students
In the past decade, mobile phones have become a ubiquitous means of communicating
with others. Text messaging, or texting, whereby individuals use a mobile phone keypad to
send messages through text rather than voice, is particularly popular among adolescents and
young adults worldwide (e.g., Drouin & Davis, 2009; Ling, 2004), although females tend to
send more numerous, lengthy and complex text messages than males (Ling, 2004; Rosen,
Chang, Erwin, Carrier, & Cheever, 2010). In many countries, including Australia, there is
currently more than one mobile phone for every person (Australia Telecommunications
Report, 2009). A 2004 survey reported that 83% of Australian adolescents owned a mobile
phone, and sent an average of one to five text messages per day (Australian Psychological
Society, 2004). More recent surveys suggest that young adult Australians spend about half an
hour a day reading and writing text messages (James, 2007), and that text-messaging is used
Until recently, text messages were limited to 160 characters (most current phones
automatically concatenate 160-character sections into single, longer messages), giving rise to
an abbreviated form of written language known as textese, which has features in common
Messaging (IM) (Ling & Baron, 2007). Textese is distinctive in its use of textisms;
contractions and nonstandard spellings specifically developed to reduce the length of words
for fast and cost-effective text messaging (Crystal, 2008). Researchers have categorised
textisms as logograms (e.g., c for see or 2 for to), letter/number homophones (e.g., m8 for
mate) and emoticons (e.g., :-) for happy), among others (e.g., Crystal, 2008; Plester, Wood, &
Joshi, 2009; Thurlow & Brown, 2003). Partly because text messaging is a relatively recent
behaviours and conventional language skills, and the evidence that does exist comes largely
from children aged 9-12 years and from young adults. There seems to be little research
examining how the use of textese in younger teenagers (13-15 years) may relate to more
traditional measures of written and spoken language skill. The current study includes students
of this age group, as well as of university age. We aimed to address three apparent gaps in the
texting literature, by examining a) the effect of data collection method on texting behaviour,
b) the effect of text entry method on texting behaviour, and c) the consistency with which
An important aim of this study was to investigate possible differences in text-message data
collection methods. Restrictions on mobile phone use in many schools mean that it can be
difficult to collect naturalistic data from school-age participants. Some researchers (e.g.,
Plester, Wood, & Bell, 2008; Plester et al., 2009) have responded to these restrictions by
having children write down their text message translations on paper, rather than type
messages into a phone. However, collecting data in this way raises questions about accuracy
and ecological validity. We therefore asked our participants to translate sentences from
standard English to text-style messages by writing them down, and also by typing them into
seems likely that standard patterns will emerge over time, and some textisms are already
widely used and recognised, for example 4 for for and b for be (e.g., Crystal, 2008; Drouin &
Davis, 2009). However, many textisms have several plausible abbreviations, and are written
in various ways (e.g., because may be shortened to cause, coz, cus, becos, bcoz, bcus, bcs, bc,
Kemp, 2010; got to go shortened to gotta go, got 2 go, g2g, Varnhagen et al., 2009). It is
Texting in high school and university students 5
unclear how individuals choose between the alternatives. The method of texting used may
influence the textisms produced. Alphanumeric mobile phone keypads contain twelve keys,
with each key representing a number and 3-4 other characters (e.g., 2, a, b, c). Words may be
entered by pressing each key several times until the required letter is displayed (the multi-
press method), or by changing to predictive mode, where each key is pressed once, and the
phone uses a dictionary to predict the most likely word resulting from a particular
combination of key presses (Taylor & Vincent, 2005). In studies with children, predictive
texting has been used by only a minority of participants (Plester et al., 2008, 2009) although
there is some evidence that it is becoming the more popular method for young adults (Kemp,
2010; Ling, 2004). Any textism, whether common or creative, must be programmed into a
phone’s dictionary for the phone to provide that textism as an option. Using the predictive
entry method may thus reduce the number of creative textisms used, and leave intact the use
of less creative textisms, including the omission of punctuation and grammatical features
such as capital letters and apostrophes, since extra key presses are required to include these in
at least some entry systems. This study is, to our knowledge, the first to focus specifically on
predictive texting.
The current study also examined how consistently the participants used the same textism
for a single word each time they wrote that word. Research in conventional orthography has
shown that poorer spellers’ errors are often less consistent than those of better spellers (e.g.,
Dietrich & Brady, 2001). Repeated exposure to written words is thought to build up
orthographic representations in memory (Holmes & Carruthers, 1998), and thus more
addition to mere exposure, individual differences may also play a role in the strength of
orthographic representations, as some good readers are also poor spellers (Frith, 1980). Thus,
Texting in high school and university students 6
general spelling skill may also predict textism consistency. However, the deliberate nature of
textism creation (compared with the presumed accident of spelling errors) makes these
hypotheses tentative. Indeed, if texting represents creativity and play with language (Crystal,
2008), better spellers might make the most of this chance to create a variety of textisms, even
especially in adolescence, may also use textisms for broader social reasons, such as to signal
social identity or group membership, (e.g., Green, 2003; Lewis & Fabos, 2005).
Media attention has focused on concerns that forms of casual, immediate communication
such as text messaging and IM may threaten traditional standards of grammar, spelling and
written expression (Crystal, 2008; Tagliamonte & Denis, 2008; Thurlow, 2006). Despite
concerns that textisms may begin to intrude into conventional writing (e.g., Massengill Shaw,
Carlson, & Waxman, 2007), recent research has shown that both undergraduates (Drouin &
Davis, 2009) and children (Plester & Wood, 2009) perceive the use of textese to be
inappropriate in academic writing. An online study by Rosen et al. (2010) reported that
textisms did intrude into young adults’ formal and informal writing in experimenter-elicited
writing samples, but only rarely (2-3 times each in only 20% of samples). In contrast to the
ideas portrayed by the media, textese can be regarded as a creative use of language which
offers an authentic representation of speech (Tagliamonte & Denis, 2008; Thurlow & Brown,
Reading and spelling in English depend to a great extent on the ability to decode letters
(graphemes) into sounds (phonemes) and vice versa. However, the English spelling system
understand that words with similar-sounding endings require different spellings according to
Texting in high school and university students 7
their structure (e.g., the final /t/ sound of kissed requires -ed because it is a regular past-tense
verb; the final /t/ sound of coast does not). Orthographic awareness helps spellers learn that
certain letter patterns are appropriate in some word contexts but others are not (e.g., word-
final /ʧ/ is generally spelled ch after two vowel letters, as in peach, and tch after one, as in
These different types of metalinguistic awareness may also be important for creating and
deciphering textisms. Some categories of textism emphasise words’ sound structure, often at
the expense of spelling conventions, including non-standard spellings (e.g., fone for phone,
wot for what) and letter/number homophones (e.g., 2day for today, l8 for late). Readers and
writers of such textisms must be able to encode and decode these phonological strings while
ignoring aspects of their conventional spelling. Previous evidence is scant, but the related
skill of phonological awareness was shown to correlate with 10- to 12-year-old children’s use
of textisms in elicited messages (Plester et al., 2009) but not with university students’ textism
use or understanding (Kemp, 2010). Similarly, morphological awareness may be helpful for
writing and reading other categories of textism. For example, knowing the morphological
structure of a frequently abbreviated word such as coming (comin) or anyone (any1) might
make it easier to create or decipher similar abbreviations for words with similar
morphological structure (e.g., goin, some1 for going, someone). Finally, many textisms break
orthographic rules by omitting letters (e.g., txt for text, tgther for together) or by combining
numbers and letters (e.g., 2morrow for tomorrow, 4get for forget). It is unclear whether
having a good knowledge of orthographic conventions would make it more or less difficult to
read or write such textisms. Thus, the current study includes measures of phonological
decoding, and morphological and orthographic awareness, as well as of reading and spelling.
To date, only a small number of empirical studies have investigated relationships between
traditional language and literacy skills and text messaging behaviour, and the results have
Texting in high school and university students 8
varied with the age group studied. Plester et al. (2009) report frequent text messaging to be
positively associated with reading and spelling measures in British children aged 10 to 12
years. However, no similar significant relationship has been observed between the frequency
of text messaging and literacy scores in Australian (Kemp, 2010) or American university
students (Massengill Shaw et al., 2007). In 10- to 12-year-old British children, the use of
textisms has been shown to correlate positively with verbal reasoning and spelling (Plester et
al., 2008), even after controlling for the number of text messages sent per day, and with
reading, phonological awareness and vocabulary (Plester et al., 2009). Plester et al. (2009)
also showed that textism use predicted a small but significant amount of variance in reading
skill even after controls for age, short-term memory, vocabulary, phonological awareness,
and years of phone ownership. Studies with adults show contrasting results. After controlling
for frequency of message-sending, there were no significant correlations between textism use
undergraduates (Kemp, 2010), whereas negative associations between reported textism use
and formal writing skill were observed in young American adults of 18-25 years (Rosen et
al., 2010). These mixed results suggest that relationships between textism use and
conventional literacy skills may be different for young adults than for children.
The aim of this study was thus to investigate textism use among high school and university
categories produced and the consistency with which textisms were spelled, and whether any
of these differed between written and texted messages. We also explored the nature of the
relationships between textism measures and conventional literacy scores, and the differences,
It was predicted that high school and university participants would not differ significantly
in terms of the proportion of textisms produced or number of categories used, on the basis of
Texting in high school and university students 9
Tagliamonte & Denis’ (2008) evidence that teenagers and young adults use written language
in similar ways in IM. Robust sex differences reported in the literature (e.g., Ling, 2004;
Rosen et al., 2010) led us to hypothesise that females in the present study would produce
more textisms, and more textism categories, than males. The comparison of textism data
collected by participants handwriting messages versus typing messages into their phones was
included to clarify the interpretation of previous findings and the design of future
Previous research with children (Plester et al., 2008; 2009) has shown that textism use
and creativity predict linguistic skills, even after controlling for texting experience and other
skills, although the scant evidence on textism use and literacy in young adults suggests
neutral (Kemp, 2010) or even negative relationships (Rosen et al., 2010). We predicted that
after controlling for age and texting experience, textism use would predict a small but
Method
Participants
participate in this research project, advertised as a ‘texting study’, for course credit. There
were 15 males and 38 females, with a mean age of 20 years (SD = 1.8, range 18-24 years).
Sixty-two Grade 8 and 9 students (from three state high schools in the same region), with
parental permission, responded to an invitation to participate sent via their schools. Data from
10 school students were excluded from the final analyses because they failed to complete all
of the tasks (n = 4), or because another student attempted to send translations of the test
messages to their phone during testing (n = 6). This left 52 high school participants: 32 males
Texting in high school and university students 10
and 20 females, with a mean age of 14 (SD = 0.7, range 13-15 years). All participants owned
a mobile phone, were familiar with text messaging and spoke English as their first language.
Materials
Messages. The texting task comprised two lists of five messages in Standard English (see
appendix), concerning social and education-related events. Each list contained 193 words,
with the number of characters per message ranging from 208-284. To increase the scope for
the use of textisms, the messages contained numerous words which had been abbreviated by
undergraduates at the same university in a study by Kemp (2010). Twenty-three target words
which had been abbreviated in several ways in Kemp’s study were repeated within and
between the two lists to allow scoring for consistency (see appendix). An example message is
I forgot to call Kate tonight because I was studying. I hate exams, which might be written in
textese as i 4got 2 call k8 2nite bcs i woz studyin. i h8 xamz, or i 4got 2 call kate 2night bcoz i
Literacy skills. We measured traditional literacy and language skills with the spelling and
reading subtests of the Wide Range Achievement Test, fourth edition (WRAT-4) (Wilkinson
& Robertson, 2006), and phonological decoding with the Nonword Reading Test (Martin &
triplets of words, and within each triplet all the words had the same ending sound/s (e.g.,
baker, lover, cover: see appendix). Two of the words had morphological endings which
changed the meaning of the word (e.g., the -er ending of baker and lover). The third word
happened to share the same ending, but with no morphological basis (e.g., cover). A correct
morpheme (e.g., cover) and words which have a base + ending structure (e.g., bake + er).
The test of orthographic awareness consisted of 24 nonword pairs, each with two different
spellings (e.g., keach vs. keatch). Both spellings would result in the same pronunciation, but
introduction. To increase the scope for potential variation in scores, the word pairs we chose
were the 24 items which adult participants had found the most difficult in Hayes et al.’s
Procedure
Participants attended two testing sessions a week apart, each lasting between 45 and 120
minutes. We provided task instructions at the beginning of the session, after which
Session 1. All participants completed the text messaging questionnaire. Then, in the
texting task, participants translated the (printed) sentences in both lists of messages from
Standard English into the spelling that they would use “if sending the message to a friend”,
typing one list into their own mobile phones, and writing the other list down on paper. List
order was counterbalanced across participants. Participants all had phones with alphanumeric
(not qwerty) keypads, and were asked to use the input mode that they usually employed:
predictive or multi-press. Ethical concerns regarding costs and the privacy of participants’
phone numbers meant that participants did not send their typed messages, but once they had
finished typing, transcribed all typed messages onto paper as they appeared on their phone
screen. The experimenters checked each transcription for accuracy and made corrections
where necessary. The translations of the two lists were alternated with the completion of the
Session 2. First, we administered the spelling subtest of the WRAT-4 to groups. This
standardised test requires participants to spell dictated real words of increasing difficulty.
Participants then repeated the texting task, writing the list that they had previous texted, and
texting the list that they had previously written. During this session, each participant was
taken into a separate room to complete the two reading tasks; the WRAT-4 reading subtest
and the Nonword Reading Test. Both are standardised tests in which participants read aloud
Results
Questionnaire Data
We examined questionnaire responses to compare the experience and use of text messaging
in high school and university students. In both groups, most participants reported that when
writing text messages they used abbreviated spellings “for some words, such as u for you”
(high school 46%, university 51%) or not at all (high school 37%, university 43%). Only a
minority reported using textisms “for most words” (high school 17%, university 6%).
= 5.29, SD = 1.85) than high school students (M = 2.92, SD = 1.74), t (103) = 6.75, p < .001,
Cohen’s d = 1.32, but high school students were significantly younger (M = 11.40 years, SD
= 1.71) than university students (M = 14.34 years, SD = 1.91) when they first received a
mobile phone, t (103) = 8.29, p < .001, Cohen’s d = 1.63. High school students reported
sending more text messages per day (M = 34.82, SD = 43.29) than university students (M =
18.00, SD = 19.76), and males reported sending the same number of messages per day (M =
high school and university students (U = 737.0, p = .157, two-tailed), nor between males and
Participants at both education levels predominantly used predictive text messaging; about
79% of high school students and 86% of university students. The rest used the multi-press
entry method. As planned, we conducted further analyses using data only from the 87
predictive texters. This meant excluding the relatively small number of high school (n = 11)
Textism Use across Education Level, Sex, and Data Collection Method
We converted the number of textisms used by each participant to a proportion of the total
number of words produced. Table 1 presents means and standard deviations of textism
proportions for both education levels, sexes, and data collection methods.
**Table 1**
As shown in Table 1, textism use was generally quite low, with remarkably similar
patterns across education level and sex groups, and even across written and texted messages.
variable, with data collection method (written, texted) as the repeated measure and with
education level (high school, university) and sex (male, female) as the between-subjects
factors. There were no significant main effects, and no significant interactions, confirming
that teenagers and young adults of both sexes used textisms with near-identical frequency,
Message length
The full versions of the messages were of varying length, and Table 2 shows the extent to
which the participants reduced the length of the messages in translation, written and texted.
**Table 2**
As shown in Table 2, the mean character counts of the translated messages were reduced
to 91-96% of their full versions (with Message 5 of List A reduced to about 80%, probably
because it contained relatively more words which could be reduced to a single character, such
Texting in high school and university students 14
as see and for). The character counts for the translated texted and written versions were
virtually identical, and all were longer than the maximum single-message length of 160
characters.
We examined the consistency of participants’ spelling of textisms for our target words
(e.g., tomorrow, because, weekend; see appendix). First, we calculated an individual spelling
score for each target word for every participant, by counting the number of times each
different spelling (including conventional spelling) was used. For example, the word
together, presented eight times, was spelled by one participant five times as 2gther, twice as
2gether, and once as 2gthr. We obtained consistency scores by summing the squares of the
individual spelling scores, divided by the total number of times the word was produced (in
the example, (5/8)2 + (2/8)2 + (1/8)2 = 0.47). A consistency score of 1 indicated that the same
spelling was used every time that word was written. If a participant spelled a target word
conventionally every time, no consistency score was recorded for that word, for that person.
As shown in Table 1, the overall level of spelling consistency for textisms was about 72%.
variable, data collection method (written, texted) as the repeated measure and education level
(high school, university) and sex (male, female) as the between-subjects factors. Females
were more consistent than males, F (1, 70) = 4.27, p = .04, partial ɳ2= .06, but no significant
Textism Categories
from Plester et al. (2009) and Crystal (2008). Table 3 summarises the twelve categories used.
**Table 3**
Texting in high school and university students 15
Number of categories used. Once all textisms were coded, we counted the number of
categories (out of 12) that each participant had used, for both written and typed text
**Table 4**
Density of categories. As well as the number of categories used, we were interested in the
density of categories. Where applicable, we classified textisms as representing more than one
category, in keeping with previous coding methods (e.g., Varnhagen et al., 2009). For
example, the textism 2moz (for tomorrow) represents a combined letter/number homophone
(2 + [word]), but also contains an accent stylisation (moz for morrow). Similarly, the textism
ive (for I’ve) omits both the apostrophe and capitalisation. We divided the total number of
categories by the total number of textisms produced to obtain a score which reflected the
density of categories. Table 4 also displays the density of categories for each education level,
university and high school participants, or between males and females, in terms of number of
categories employed. However, overall, participants used significantly more categories when
they wrote messages down than when they typed them into their phones, F (1, 82) = 8.65, p <
.01, partial ɳ2= .10. This suggests that although different methods of data collection yield
remarkably similar results, some differences exist, at least for this feature of message
this density was significantly higher for written than for texted messages, F (1, 81) = 6.98, p
= .01, partial ɳ2= .08, with no significant differences for sex or education level. Thus,
although the written and texted messages were remarkably similar overall, the written
Texting in high school and university students 16
messages did include a wider range of textism categories, and a higher proportion of
different ways according to education level or data collection method, we divided each
individual category score by the total number of categories used. Differences between data
collection methods were significant for only two types of textism: contractions (written, M =
0.12, SD = 0.09; texted, M = 0.07, SD = 0.08; t (85) = -5.14, p < .001, Cohen’s d = 1.12 ) and
accent stylisations (written, M = 0.13, SD= 0.17; texted, M =0.09, SD = 0.11; t (85) = -3.30,
p = .001, Cohen’s d = 0.72), which were both produced more frequently when participants
wrote, rather than texted, the messages. In terms of education level, initialisms were used
.004, SD = .009), t (84) = 4.27, p < .001, Cohen’s d = 0.93. No other differences were
significant, meaning that overall patterns of category use were very similar for both education
levels. Because there were so few differences according to education level or data collection
method, Figure 1 displays the mean proportions of each of the twelve categories for all
**Figure 1**
In interpreting the figure, it should be borne in mind that the wording of the messages that
the participants were asked to translate at least partially constrained the opportunities for
using each textism category. For example, although participants could potentially shorten,
contract, or misspell as many words as they liked, the opportunity for omitting capitals or
apostrophes, or clipping the g from -ing endings was limited by the messages’ wording. With
these constraints in mind, it can be seen from the figure that the most common textisms were
omitted capital letters and omitted apostrophes, which made up .24 and .17 of categories
letter/number homophones, symbols and spelling errors made up .04 to .06 of categories. It
should be noted that if single and combined letter/number homophones are grouped together,
as in some previous research (Plester et al., 2008, 2009), the textism categories used would
more closely resemble the patterns reported previously. Participants used initialisms and g-
clippings the least, with these each comprising less than .03 of categories produced.
Literacy Scores
Table 5 shows the university and high school students’ mean standard scores on the tests
of spelling, reading and nonword reading, which were in the average range, and also their
scores on the experimental tests of morphological and orthographic awareness. The reliability
**Table 5**
WRAT spelling or nonword reading, but unexpectedly, the high school students significantly
0.88. For the experimental tasks of morphological and orthographic awareness, one-sample t-
tests revealed that mean scores were significantly above chance levels; t (85) = 7.43, p <
.001, Cohen’s d = 1.61 and t (85) = 16.88, p < .001, Cohen’s d = 3.66, respectively, although
it is clear that the morphological task was the more difficult. Independent samples t-tests
revealed that undergraduates scored significantly higher than high school students on
Correlational Analyses
Table 6 shows the results of bivariate correlations between measures of texting experience
and behaviour, and literacy scores. The two age groups had shown such similar behaviour on
all of the texting tasks that we decided to combine them for the final analyses, even though
Texting in high school and university students 18
they had differed significantly on two of the five literacy measures (reading and
morphological awareness). This also kept the sample size as large as possible. These bivariate
correlations were calculated only for the texted messages, not the written ones, and are shown
**Table 6**
Table 6 shows that, in contrast to previous research with children, the correlations
between textism use and literacy scores were largely negative. Sending more messages per
day, using a greater number of textism categories, and using a greater proportion of textisms
were all moderately to weakly associated with lower scores on spelling, reading, nonword
reading and morphological awareness, although this may reflect the unexpectedly worse
reading performance by the university than the high school students. A longer period of
texting experience was moderately associated with lower scores on word and nonword
reading. None of the texting experience or literacy scores correlated significantly with
textism consistency.
It is possible that some of the negative relationships between texting and literacy measures
could be attributed in part to the frequency with which participants usually sent text
messages, and thus with their usual exposure to texting and textisms. Finally, then, we
conducted partial correlations between the texting and literacy measures, controlling for
number of messages sent per day. These correlations are shown in the upper/right part of
Table 6. After this control, the negative correlations between the proportion of textisms used
and real-word reading and spelling no longer reached significance, but remained significant
for nonword reading. Number of textism categories used still correlated significantly and
negatively with real- and nonword reading and spelling, and textism spelling consistency
Discussion
Texting in high school and university students 19
The aim of this study was to investigate and compare the use of textisms in high school
students of 13-15 years and university students of 18-24 years. In both handwritten and typed
and the number and types of textism categories employed, as well as relationships between
The majority of our participants used the predictive text entry method; 79% of high school
students, and 86% of university students. This is greater than the prevalence reported in
previous research; 20% (Plester et al., 2008) to 55% (Plester et al., 2009) of 10- to 12-year-
old British children, 55% of British undergraduates (Thurlow & Brown, 2003), and 77% of
Australian undergraduates (Kemp, 2010). Rapid changes in technology mean that texting
behaviour is constantly evolving, and research will need to keep pace with such changes. The
sentence character count from about 233 to about 214, a reduction of about 8%. This 214-
character average is still much longer than the 160-character maximum single-message
length, presumably because the full versions were simply too long to reduce further.
By rapidly suggesting possible words, the predictive method should greatly reduce, or
even eliminate, the need for textisms. Often the whole word is suggested after only a few
letters are typed in, meaning that typing in abbreviated spellings is often no faster than simply
typing in the first few letters and then choosing the appropriate entire word’s spelling.
However, about half the sample reported using textisms for at least some words, and textisms
still accounted for 13-16% of words in the text messages translated by participants. This is
substantially lower than the 50-54% shown by British children in translated messages (Plester
et al., 2008, although their messages were more realistic in their shorter length), although
more naturalistic elicited messages in that population produced about 35% textisms (Plester
et al., 2009). However, the present proportion is similar to the nearly 19% textisms produced
Texting in high school and university students 20
in natural text messages by British undergraduates (Thurlow & Brown, 2003), and rather
higher than the 5% in those of American undergraduates (Ling & Baron, 2007). Regardless of
text entry method, the use of textisms clearly differs with age group and message type, but
does seem generally lower in young adults than in children. However, the present study
showed no significant differences between teenagers and young adults in terms of the
proportion of textisms used. This is consistent with findings that adolescents and young
adults use similar rates of short forms in IM (Baron, 2004), and that individuals may leave
behind the desire to produce extensive textisms in early adolescence (Tagliamonte & Denis,
2008). However, the social meaning and ‘impression management’ associated with continued
use of textisms may vary with age (e.g., Green, 2003; Lewis & Fabos, 2005).
In general, we observed virtually identical texting behaviours between high school and
university students on our other measures as well. The two groups showed very similar use of
the 12 textism categories examined, except for initialisms, which were used more often by
undergraduates than by high school students. As noted earlier, the wording of the messages
themselves at least partially restricted the opportunities for use of certain categories: for
example, contraction is potentially possible with most words, but g-clipping can only be done
when there are -ing endings to clip. With these restrictions in mind, the most frequent
categories for both groups were omitted capitals (i want) and omitted apostrophes (I cant),
homophones (u), and shortenings (xcellent). Plester and colleagues’ sample of British
children’s most frequent categories were combined letter/number homophones (l8r) and non-
standard spellings (nite) (Plester et al., 2008), and combined letter/number homophones,
contractions (txt) and accent stylisations (hafta) (Plester et al., 2009). Our sample’s use of
predictive texting may have reduced the use of more creative textisms in comparison to
textisms of omission (of capitals and of apostrophes). This is because predictive texting
Texting in high school and university students 21
predicts only common abbreviations, or those that the phone user has programmed in. Thus,
someone using predictive texting is unlikely to come up with new textisms each time, but
instead to use the existing suggestions. In contrast, many phones allow the incorrect omission
of capitals and apostrophes, and so users can save time by not trying to include these features.
It was expected that predictive texting may have also have discouraged any variability in
the spelling of textisms, but spelling consistency was only 68 to 78%. If an individual is using
predictive spelling and types a spelling that is not in the phone’s dictionary, such as cos
(because), the predictive function “remembers” that spelling and will provide it as an
alternative the next time those keys are pressed in the same sequence. Textism spelling
consistency thus depends partly on the ability (or decision) to learn and remember one’s own
previous spellings, and thus we expected that there might be significant correlations between
textism consistency and texting frequency, and/or spelling ability. However, none of these
exists, participants may not have found it important to write their textisms the same way
every time. Alternatively, texters might vary their spelling for different reasons: poorer
spellers because they cannot remember the exact version they used last time, and better
spellers because they are having fun with language, or tailoring their textisms to suit the
message recipient. Further, texters of any ability may vary the textism they use for a word
according to the length of the message or the time available to write it. Overall, the current
results suggest that the use of predictive texting does not standardise all aspects of textism
spelling, and variability continues to exist both between and within texters, regardless of
Previous researchers (Plester et al., 2008, 2009) have collected textism data by having
(child) participants handwrite messages as they imagine they would type them, but the
ecological validity of this method is not clear. In the present study we had participants type
Texting in high school and university students 22
and handwrite the same messages, and found that in both conditions, the use of textisms was
remarkably similar. No significant differences were found between texted and written data in
terms of the proportion of textisms to total words produced, nor in terms of consistency
scores. This suggests that as early as age 13 years, individuals can reflect upon their own
texting habits and reproduce textisms on paper in a representative way, without the confines
imposed by a mobile phone keypad. There were a few differences; participants used more
textism categories overall, and more categories per textism, when writing messages down
than when typing them. Specifically, participants used more accent stylisations and
contractions when they wrote messages down. Given that ethical and practical requirements
often restrict the use of mobile phones in school settings, future research may benefit from
these findings. While the use of written data could be justified in conducting a study which
focused on the proportion and/or consistency of textisms used, the use of texted data would
In terms of sex differences, we did not observe the predicted greater use by females of
textisms and of textism categories that has been reported in previous research (e.g., Ling,
2004; Plester et al., 2009; Rosen et al., 2010). The text messages dictated here were rather
long, and involved mainly social information (e.g., Kate didn’t lose weight and is still
worrying), two features which are normally more associated with messages written by
females than males (Baron, 2004). The nature of these messages might thus have restricted
the scope of the sex differences that would occur in naturalistic messages.
The final important aim of the current study was to investigate the links between various
aspects of texting behaviour, and measures of literacy and language skill, in high school
picture: the number of text messages sent per day, proportion of textisms produced, and
number of textism categories used all correlated negatively with the standardised measures of
Texting in high school and university students 23
reading, spelling, and nonword reading, and the experimental measure of morphological
awareness. (Correlations with the experimental measure of orthographic awareness were not
significant.) These results differ from those observed previously in 10- to 12-year-old
children, who have generally shown positive associations between literacy skills and use of
text messaging and of textisms (Plester et al., 2008, 2009). They also differ from research
with university students, which has revealed no significant differences on language measures
according to frequency of texting (Massengill Shaw et al., 2007) or use of textisms after
When partial correlations were calculated controlling for number of text messages sent per
day, the proportion of textisms correlated significantly and negatively with nonword reading
(but no longer real word reading and spelling), and number of categories still correlated
significantly and negatively with real- and nonword reading and spelling. It seems that some
of the negative relationships observed between textism use and conventional literacy skills
might be accounted for, at least in part, by texting frequency rather than the use of textisms
themselves. However, the negative relationships are still unexpected, given previous results.
Of course, it cannot be inferred that greater use of text messaging is the cause of poorer
attainment on literacy and language measures. Teenagers and young adults with lower
linguistic abilities may be motivated to text more frequently by the greater freedom and
flexibility provided by this non-conventional form of written language (Plester et al., 2009).
Alternatively, young people who are less competent in literacy may use texting to mask poor
spelling skills. However, neither of these explanations can be the whole story, as the use of
textisms, and of textism categories, did not individually contribute significantly to language
outcomes, although their combined contributions reached significance for reading, nonword
reading, and spelling. It must also be noted that the individuals tested here all used mainly
predictive texting, not the multi-press entry method used most often by participants in
Texting in high school and university students 24
previous studies. Thus, the types of textisms that the current participants were regularly using
might have been less creative than those experienced by previous samples, as is suggested
here by the relatively high numbers of textisms of omission (omitted apostrophes and
capitals) compared to, for example, the letter-number homophones more common in previous
studies (e.g., Plester et al., 2008, 2009). Future research should investigate potential
There are some limitations in how well the current findings can be generalised to the real
vary widely, from 7.7 words in samples from American undergraduates (Ling & Baron,
British children (Wood, Plester, & Bowyer, 2008, as cited in Plester & Wood, 2009). The text
messages designed for this experimental study were very long in comparison, averaging
about 40 words per message. As previously noted, the proportion of textisms used per
message also differs substantially according to participants’ age and familiarity with texting,
as well as with the nature of the message (translated, elicited, or naturalistic). It is not clear
how well the current participants’ use of textisms represented their everyday use, but the 13-
16% observed seems to fit with previous reports. It is also difficult to know how well
participants’ reported frequency of text-messaging, and use of textisms, reflected their actual
behaviour. Future research might aim to analyse naturalistic messages collected from
individuals’ phones, as some researchers have successfully done (e.g., Ling, 2004; Ling &
Baron, 2007; Plester, Wood, Jackson, & Wilde, 2009), rather than using translation or
dictation exercises.
In conclusion, this study has shown that the moderate use of a variety of textisms remains
an important part of writing text-messages, even with the predictive entry method, for both
adolescent and young adult texters. In contrast to the largely positive associations observed in
Texting in high school and university students 25
children, in this sample we saw significant negative correlations between texting behaviours
and more traditional literacy and language skills, although some of these seem to be
accounted for by frequency of text-messaging overall, rather than any specific use of
textisms. Further research will be required to examine the causes behind these correlations. If
less competent language users are drawn to the creativity and flexibility of textese in a way
that improves their language skills through exposure to written language (as has been
suggested with younger children), then mobile phones could prove useful in educational
settings. However, if negative relationships indicate that frequent texting somehow interferes
with the development of language and literacy skills, or provides an opportunity for less able
students to mask poor spelling, quite different conclusions may be drawn. Texting and other
increasingly younger ages, and further research will be necessary to clarify the interactions
between conventional language and literacy skills and the frequent use of written
Table 1
Means and Standard Deviations for Proportion of Textisms and Consistency Scores for
High school
Male (n = 22) .16 (.11) .16 (.11) .71 (.12) .71 (.12)
Female (n = 18) .16 (.16) .15 (.15) .78 (.14) .72 (.11)
Total (n = 40) .16 (.14) .15 (.13) .74 (.13) .72 (.11)
University
Male (n = 14) .13 (.16) .13 (.15) .67 (.07) .66 (.10)
Female (n = 32) .14 (.14) .14 (.14) .73 (.09) .70 (.11)
Total (n = 46) .14 (.14) .14 (.14) .71 (0.10) .70 (0.11)
Texting in high school and university students 27
Table 2
Character Counts for Full Messages and Mean Character Counts for Participants’ Texted
and Written Translations, and Translations’ Length as Proportion of Full Versions. Standard
Deviations in Parentheses.
List A List B
Texted translations
Written translations
Table 3
Table 4
Means and Standard Deviations for Number of Categories Used and Density of Categories
High school
Male (n = 22) 6.23 (2.43) 6.86 (2.49) 1.19 (0.09) 1.20 (0.14)
Female (n = 18) 6.67 (3.01) 7.00 (2.81) 1.18 (0.11) 1.25 (0.28)
Total (n = 40) 6.43 (2.68) 6.93 (2.61) 1.18 (0.10) 1.22 (0.21)
University
Male (n = 14) 6.57 (3.32) 7.64 (2.85) 1.13 (0.09) 1.22 (0.18)
Female (n = 32) 6.87 (2.70) 6.94 (2.68) 1.21 (0.17) 1.21 (0.17)
Total (n = 46) 6.78 (2.87) 6.93 (2.61) 1.19 (0.15) 1.23 (0.14)
Texting in high school and university students 30
Figure 1. Mean (and standard error of the mean) proportion of each category type to total
categories produced, collapsed across education level and data collection method. (Note that
Table 5
Means and Standard Deviations for Literacy Scores for High School and University Students
Task
Table 6
Bivariate Correlations between Texting Measures and Literacy Scores (Lower Left) and Partial Correlations Controlling for Texting Frequency
(Upper Right)
Age No. yrs No. Prop No. Textism Spelling Reading NWR MA OA
texting msg/day textisms categ. consist
Age - .75** - -.05 .18 -.24* -.08 -.47** -.20 .28* .07
No. years texting .67** - - .06 .21 -.09 -.12 -.45** -.21 .18 .09
Proportion textisms -.05 .05 .08 - .72** .50** -.21 -.22 -.30** -.14 -.14
Number of categories .12 .14 .13 .75** - .21 -.25 -.37* -.31* -.14 -.07
Textism consistency -.21 -.07 -.002 .50** .21 - .01 -.07 -.17 -.16 .03
WRAT Spelling .03 -.09 -.32** -.27* -.33 .02 - .56** .64** .07 .38*
WRAT Reading -.35** -.46** -.31** -.25* -.37** -.08 .58** - .74** .02 .19
Nonword Reading -.11 -.27* -.32** -.29** -.31** -.17 .64** .77** - -.10 .33**
Morph Awareness .28** .07 -.29** -.21* -.26* -.14 .23* .20 .11 - .11
Orth Awareness .04 .05 -.09 -.16 .12 .02 .39** .25* .36** .15 -
Appendix
Text Messaging Task Sentences for Transcribing
4. Let’s get together tomorrow between exams, please. I’m excited about my formal
pictures. I tried printing them but failed because the format is different on my printer
or something. Annoying, but whatever. See you soon. (223)
5. I totally forgot my parents aren’t going to be there on the weekend. I’m very excited to
celebrate the end of exams together at their place in Margate. You and Kate could
sleep over. Can’t wait. Bye for now. (212)
Target words included in consistency scoring (number of times words repeated across both
lists): never (4), whatever (4), over (4), to (11), today (4), tomorrow (4), tonight (4), together
(8), for (5), formal (4), forget (4), forgive (4), you (9), your (6), weekend (4), because (6),
pictures (4), tickets (4), going (4), wait (4), celebrate (4), Kate (4), excited (4), exams (4), text
(4).
References
Australian Bureau of Statistics (2008). Census At School (2008). Findings based on data
websitedbs/cashome.nsf/Home/Teacher+Resources
Australian Psychological Society (2004). Psychosocial impact of mobile phone use amongst
mobile_phones/
http://www.businessmonitor.com/telecommunications/australia.html
Baron, N. (2004). See you online: Gender issues in college student use of instant messaging.
X04269585
Crystal, D. (2008). Txting: The gr8 db8. Oxford: Oxford University Press.
Dietrich, J.A., & Brady S.A. (2001). Phonological representations of adult poor readers: An
Drouin, M., & Davis, C. (2009). R u txting? Is the use of text speak hurting your literacy?
Green, N. (2003). Outwardly mobile: Young people and mobile technologies, in J. E. Katz
(Ed.) Machines that become us: The social context of personal communication technology
Hayes, H., Treiman, R., & Kessler, B. (2006). Children use vowels to help them spell
10.1016/j/jecp.2005.11.001
Texting in high school and university students 36
James, D. (2007). Just one more call, QUT Links Alumni Magazine, 10, 1, retrieved 16
Feb2007.pdf.
Holmes, V. M., & Carruthers, J. (1998). The relation between reading and spelling in
Kemp, N. (2010). Texting vs. txting: Reading and writing text messages, and links with other
Lewis, C., & Fabos, B. (2005). Instant messaging, literacies, and social identities. Reading
Ling, R. (2004). Mobile connection: The cell phone’s impact on society. Morgan Kaufmann:
Amsterdam.
Ling, R., & Baron, N. (2007). Text messaging and IM: Linguistic comparison of American
college data. Journal of Language and Social Psychology, 26, 291–298. doi:
10.1177/0261927X06303480
Martin, F., & Pratt, C. (2001). Nonword Reading Test. Melbourne: Australian Council for
Massengill Shaw, D., Carlson, C., & Waxman, M. (2007). An exploratory investigation into
the relationship between text messaging and spelling. New England Reading Association
Neville, L. (2003). Cn U rEd dis? The causes and consequences of a ‘text message
Plester, B., & Wood, C. (2009). Exploring relationships between traditional and new media
Plester, B., Wood, C., & Bell, V. (2008). Txt msg n school literacy: Does texting and
42(3), 137–144.
Plester, B., Wood, C., Jackson, E., & Wilde, L. (2009, June). Children’s use of mobile phone
text messaging and its impact on literacy development in primary school: Txt as
enhancement of literacy. Poster presented to the 16th meeting of the Society for the
Plester, B., Wood, C., & Joshi, P. (2009). Exploring the relationship between children’s
Rosen, L., Chang, J., Erwin, L., Carrier, M., & Cheever, N. (2010). The relationship between
“textisms” and formal and informal writing among young adults. Communication
Tagliamonte, S.A., & Denis, D. (2008, Spring). Linguistic Ruin? LOL! Instant messaging and
Taylor, A., & Vincent. J. (2005). An SMS history, in L. Hamill & A. Lasen (Eds.), Mobile
World: Past, Present and Future (pp. 75–91). Springer: New York.
Thurlow, C., & Brown, A. (2003). Generation Txt? The sociolinguistics of young people’s
/articles/v1/n1/a3/thurlow2002003-paper.html.
construction and popular exaggeration of new media language in the print media.
http://www3.interscience.wiley.com.ezproxy.utas.edu.au/cgi-bin/fulltext /11855
4163/PDFSTART
Texting in high school and university students 38
Varnhagen, C., McFall, P., Pugh, N., Routledge, L., Sumida-MacDonald, H., & Kwong, T.
(2009). lol: New language and spelling in instant messages. Reading and Writing. doi:
10.1007/s/11145-009-9181-y
Wilkinson, G., & Robertson, G. (2006). Wide Range Achievement Test- Revised 4