Effects of the Portuguese GraphoGame on Reading, Spelling, and Phonological Awareness in Second Graders Struggling to Read

This is a self-archived version of an original article. This version may differ from the original in pagination and typographic details. Author(s): Title: Year: Version: Copyright: Rights: Rights url: Please cite the original version: CC BY-NC-ND 3.0 https://creativecommons.org/licenses/by-nc-nd/3.0/ Effects of the Portuguese GraphoGame on Reading, Spelling, and Phonological Awareness in Second Graders Struggling to Read © Authors, 2020 Published version Carvalhais, Lénia; Limpo, Teresa; Richardson, Ulla; Castro, São Luís Carvalhais, L., Limpo, T., Richardson, U., & Castro, S. L. (2020). Effects of the Portuguese GraphoGame on Reading, Spelling, and Phonological Awareness in Second Graders Struggling to Read. The Journal of Writing Research, 12(1), 9-34. https://doi.org/10.17239/jowr2020.12.01.02 2020 Carvalhais, L., Limpo, T., Richardson, U., & Castro, S.L. (2020). Effects of the Portuguese Graphogame on reading, spelling, and phonological awareness in second graders struggling to read. Journal of Writing Research, 12 (1) , 9-34. https://doi.org/10.17239/jowr-2020.12.01.02 Contact: Lénia Carvalhais, Universidade Portucalense, Rua Dr. António Bernardino Almeida, 541-619, P 4200-072, Oporto | Portugal - [email protected] Copyright: Earli | This article is published under Creative Commons AttributionNoncommercial-No Derivative Works 3.0 Unported license. Effects of the Portuguese GraphoGame on Reading, Spelling, and Phonological Awareness in Second Graders Struggling to Read Lénia Carvalhais°, Teresa Limpo*, Ulla Richardson+ & São Luís Castro* ° Portucalense University | Portugal * University of Porto | Portugal + University of Jyvaskyla | Finland Abstract: The interest in computer-assisted interventions to promote literacy has increased over the years. In this study, we developed the Portuguese version of the GraphoGame Fluent and tested its effects on reading, spelling, and phonological awareness. Second graders struggling to read were randomly assigned to two groups: GraphoGame Fluent group ( n = 15), which received a computer-assisted remedial reading intervention, or GraphoGame Math group ( n = 15), which received a computer-assisted numeracy intervention. An additional, non-playing group, composed by second graders without reading difficulties, was formed (business-as-usual group, n = 15). Results showed clear benefits of GraphoGame on spelling and phonological awareness. After the intervention and one month later, the GraphoGame Fluent group displayed spelling and phonological awareness skills similar to the business-as-usual group, and above the GraphoGame Math group. Overall, these findings indicate that Portuguese struggling readers benefit from computer-assisted interventions that combine letter-sound correspondences with more complex orthographic patterns. Keywords: computer-assisted intervention; reading; spelling; phonological awareness; Portu g uese Gra p hoGame CARVALHAIS ET AL.  EFFECTS OF THE PORTUGUESE GRAPHOGAME | 10 1. Introduction Over the last decades, there has been an increasing research focus on the development of computer-assisted interventions to promote literacy in children at risk of reading difficulties (Lyytinen, Erskine, Kujala, Ojanen, & Richardson, 2009; McCandliss, 2010; Saine, Lerkkanen, Ahonen, Tolvanen, & Lyytinen, 2011; Solheim, Frijters, Lundetrae, & Uppstad, 2018). Although most published studies were based on learning to read in English (Kyle, Kujala, Richardson, Lyytinen, & Goswami, 2013; Lyytinen et al., 2009), which is classified as an opaque orthography (Seymour, Aro, & Erskine, 2003), recent studies have focused on transparent orthographies, such as Finnish (Richardson & Lyytinen, 2014). However, little is still known about the effectiveness of computer-assisted literacy interventions in intermediate depth orthographies. This is the case of Portuguese, in which the correspondences between phonemes and graphemes are not always univocal (Seymour et al., 2003). In this study, we tested if Portuguese struggling readers benefit from a computerbased intervention in early phases of reading/writing development (Grade 2, age 78). Here, struggling readers refer to children struggling with the process of acquiring automaticity in word reading. 1.1 Development and Promotion of Reading and Spelling An important model of reading development is Ehri’s stage model (Ehri & McCormick, 1998; Saine et al., 2011). This five-stage model emphasizes the importance of letter-sound relationships in learning to read. During the prealphabetic stage children distingue visual cues as logos associated with words, shapes, and letter names. In the partial-alphabetic stage children use phonetic cues regarding the link between letters and sounds to start recognizing words (e.g., initial letters). In the full-alphabetic stage , children can fully apply the alphabetic principle to decode familiar and unfamiliar words. With reading practice, children move to the consolidated alphabetic stage , where they start joining multi-letter units into words. Finally, in the automatic stage , children are able to read words with fluency and accuracy. According to this model, children progress across theses stages through explicit and systematic reading instruction focused on letter-sound correspondences (Saine et al., 2011). This kind of teaching, providing training in phonological awareness with highly structured reading instruction (Hatcher, Hulme, & Ellis, 1994; Hatcher et al., 2006), helps children to realize that spoken words can be broken into smaller units of sounds and to learn the multitude of spellingsound correspondences (Gillon, 2004). An identical stage-like progression has been proposed for spelling (Ehri, 1992; for a review see Babayiğit, 2009). In the earliest stage, children realize that writing is different from drawings. Their first writing attempts are characterized by scribbles or letter-like shapes. With increased exposure to print and the beginning of reading 11 | JOURNAL OF WRITING RESEARCH instruction, children realize that speech is represented by alphabet letters. At this stage, children make use of partial alphabetic knowledge to spell words in writing. The progressive increase of alphabetic knowledge leads to full alphabetic spelling, in which children spell words by applying the alphabetic principle. However, given the complexity of orthographic systems, exclusive reliance on this strategy may lead to errors, mainly when spelling complex words (e.g., homophones). Progressively, children acquire more sophisticated knowledge of the spelling system, including knowledge of orthographic patterns and morphological relationships among words. Supporting the key role of the alphabetic principle in reading and spelling development, several studies showed the effectiveness of interventions targeting letter-sound correspondences to promote both reading (Fletcher, Lyon, Fuchs, & Barnes, 2007; Snowling & Hulme, 2011) and spelling (Graham, Harris, & Fink Chorzempa, 2002). However, the mainstream approach to reading and spelling instruction still fails to support at-risk children (Hatcher, Hulme, & Snowling, 2004). The use of technology through computer games seems a promising means to support children that struggle in the initial steps of learning to read and write. 1.2 Computer-Assisted Tools: GraphoGame1 There are several advantages in using computer-based teaching tools. For example, they reduce the demands of traditional instruction (Knight et al., 2020; Nicolson, Fawcett, & Nicolson, 2000), allow children to receive responsive and immediate feedback (Lyytinen, Aro, Holopainen, Leiwo, Lyytinen, & Tolvanen, 2006), promote individualized learning by adapting tasks to children’s learning progress, and increase children’s motivation to be engaged in the learning task (Richardson & Lyytinen, 2014; Vandermeulen, Leijten, & Van Waes, 2020). Computer-assisted reading interventions seem particularly effective for those at-risk for reading difficulties (Blok, Oostdam, Otter, & Overmaat, 2002; Nicolson et al., 2000). However, for the success of those tools, it is crucial to catch children’s attention, adapt the task to their learning in a playful environment, and motivate them to complete the tasks. A reading program that was developed based on these premises was the GraphoGame (Lyytinen, Ronimus, Alanko, Poikkeus, & Taanila, 2007). The GraphoGame program was created to enrich regular remedial intervention with individualized repetition of carefully selected items (Lyytinen et al., 2007). Across several motivating games, this tool trains the partial-, full-, and consolidatedalphabetic stages proposed in Ehri’s model (Ehri & McCormick, 1998). This is achieved by exposing children to phoneme-grapheme correspondences along with specific knowledge about letters, which together improve reading fluency and accuracy (Lyytinen et al., 2007). Notably, this study also found transfer effects of GraphoGame to writing, in particular, to children’s spelling skills. The game is CARVALHAIS ET AL.  EFFECTS OF THE PORTUGUESE GRAPHOGAME | 12 organized according to a hierarchy of streams, levels, and items with increasing difficulty. Children’s progress along these depend upon their learning. A typical task within GraphoGame involves two steps (Lyytinen et al., 2009). First, players listen to a spoken item, which can be presented as many times as needed. Second, players must select the written item that matches the spoken item, presented among several distractors. The written items can be letters, syllables, onset-rimes, words or pseudowords. Players are then immediately provided with visual (color) and auditory (sound) feedback, indicating a correct or incorrect answer. A correct answer allows children to move forwards in the game. After an incorrect answer, the target item appears again (with or without distractors), giving children the opportunity to learn. The rationale behind this procedure is that task difficulty increases or decreases depending on the players providing correct or incorrect answers, respectively (Richardson & Lyytinen, 2014; Ronimus, Kujala, Tolvanen, & Lyytinen, 2014). The progression throughout the game is therefore automatically computed in accordance with the responses provided by children so that a success probability of 80% is kept constant ( Kujala, Richardson, & Lyytinen, 2010; Lyytinen et al., 2009). Another important feature of the GraphoGame is the reward system used, which aims to keep children motivated to play. At the root of this system is a personal avatar, selected by the players from the very beginning, and the stamps he/she receives after each level, which are pasted in the player’s personal page. For each player, the GraphoGame web platform presents information about games’ exposure time and performance (i.e., lower and higher marginal know-how expressed in percentage of correct answers). The platform also saves several other performance indicators organized into a table. For an item, the table displays the number of times the item was presented, accuracy percentage of the first and last seven trials (representing players’ level of success at the beginning and end of each session), mean accuracy percentage of all trials, and the lowest and highest success level. Users can have access to more fine-grained results graphically presented through the Daisygraphs, which illustrate discrimination between letters (Lyytinen et al., 2009). Based on these graphs, it is possible to determine how well a player distinguishes a target item from a specific distractor. 1.3 Effectiveness of the GraphoGame GraphoGame is now a worldwide computer game adapted to several different languages (around 30 versions; more information from https://info.grapholearn.com/ partners/) and used in the four continents: Europe (e.g., Austrian, Swiss-German, Greek, Dutch, Norwegian versions), Africa (e.g., Nyanja, Kiswahili, or Afrikaans); America (e.g., Chilean Spanish, Brazilian Portuguese, Canadian); and Asia (British English as a second language, Indonesian, Pinyin, and Mandarin). It was initially developed for preventive purposes. 13 | JOURNAL OF WRITING RESEARCH Specifically, it was designed for Finnish children showing early signs of reading difficulties at the end of kindergarten. The first evidence on the effectiveness of GraphoGame came from pilot studies. Initial results were very promising, by indicating that non-reading children were able to acquire basic reading skills after 4 hours of gaming (Lyytinen et al., 2007). In the first pilot study described by Lyytinen and collaborators (2007), twelve kindergarteners were divided into two groups, who played the GraphoGame or a math-focused game (active control). Performance in a blending task – in which children were asked to say a full word form presented by the computer in segments (viz., syllables and/or phonemes) – revealed the first efficacy results of the Finish GraphoGame. Saine and colleagues (2011) conducted a longitudinal intervention study using the GraphoGame method with Finnish speakers, who were followed from Grade 1 to 3. Based on the assessment of the pre-reading skills of 166 sevenyear-old students, 50 children were identified as struggling readers and assigned to one of two intervention groups receiving remedial reading support via traditional methods or GraphoGame. Results showed that GraphoGame greatly improved children’s reading and spelling skills and that it was highly beneficial compared to traditional support. Because Finnish has a transparent orthography, the relationship between phonemes and graphemes (or vice-versa) is univocal. Thus, the design method developed to train contents in Finnish language using the GraphoGame is very straightforward. The game starts introducing the simplest correspondences (e.g., a, s, t ) and moves to correspondences that are phonetically less distinguishable (e.g., m, n, l ). Only later are larger units, such as syllables and rimes, targeted in the games, followed by whole words (Richardson & Lyytinen, 2014). Other versions of the GraphoGame were created for less transparent orthographies, such as English, and the efficacy of its different versions have been shown (Richardson & Lyytinen, 2014). For example, Kyle and collaborators (2013) developed the English GraphoGame Rime and the English GraphoGame Phoneme, which were administered to poor readers in Grade 2. Although both forms were designed to automatize grapheme-phoneme conversion skills, one game targeted phoneme-level units, whereas the other focused on orthographic rime units. Different groups of children played each game for 12 weeks, during five sessions per week (total of 11 hours of training). Results showed that, regardless of targeting phonemes or rimes, GraphoGame resulted in gains in reading, spelling, and phonological awareness compared with an untreated control group. In addition to the Phoneme and Rime versions, there is also a GraphoGame Fluent version designed to promote fast and effortless reading. The GraphoGame Fluent also aims to impact on the teaching practices of literacy in children with specific reading difficulties who need extra help for learning. One study that tested the GraphoGame Fluent with poor readers of Finish found clear gains in reading speed regarding all trained syllables, and transfer effects to the word level for certain types CARVALHAIS ET AL.  EFFECTS OF THE PORTUGUESE GRAPHOGAME | 14 of syllables (Heikkilä, Aro, Närhi, Westerholm, & Ahonen, 2013). The Finnish GraphoGame Fluent version was adapted to English, German, and Polish. The current study aimed to adapt it to Portuguese (see the Method section for more detail on the features of the adapted version). Portuguese is an orthography of intermediate depth with some orthographic complexities (Seymour, Aro, & Erskine, 2003), which are acquired throughout primary school (Mesquita, Carvalhais, Limpo, & Castro, 2020). At the end of Grade 1, children seem to have already acquired basic decoding skills (Fernandes, Ventura, Querido, & Morais, 2008). They were able to correctly read and spell an average of 6 out of 8 simple words and about half of their errors were due to partial or complete parsing of words phonemic structure. Thus, the GraphoGame Fluent version seemed particularly suitable for second graders with basic decoding skills but struggling with the process of acquiring automaticity in word reading. Indeed, the Fluent version was designed to overcome the limitations of other versions focused on individual letter-sound correspondences, and therefore more appropriate for second graders and/or children with difficulties in acquiring the alphabet principle. 1.4 Present Study As noted above, GraphoGame has been largely implemented either in transparent orthographies, such as Finnish (Saine et al., 2011), or opaque orthographies, such as English (Kyle et al., 2013). Little is still known about the advantages of GraphoGame to promote reading and spelling skills in struggling readers in orthographies of intermediate depth, such as Portuguese, as well as across different educational systems and instructional contexts. This was the main goal of the present study, in which we developed and tested the Portuguese version of GraphoGame Fluent (Richardson, 2011). After adapting the original Finish version to Portuguese, GraphoGame was implemented with second graders displaying low results in standardized reading measures. Struggling readers playing GraphoGame Fluent over a 6-week period were compared with struggling readers playing GraphoGame Math (Richardson & Lyytinen, 2014), and their peers without reading difficulties. Students’ reading, writing, and phonological awareness skills were assessed before and after training as well as one month later. Based on the previously surveyed research, we anticipated that GraphoGame Fluent would improve children’s reading, spelling, and phonological awareness skills. These gains were expected to be induced by the systematic and explicit training targeting not only phoneme-grapheme correspondences but also larger units, such as syllables, rimes, and whole words; along with training on the specific rules of the Portuguese orthography. 15 | JOURNAL OF WRITING RESEARCH 2. Method 2.1 Screening Procedure A total of 104 second graders from two schools in a middle-class urban area in Portugal were invited to participate in the study via school authorities and parents. Except for two students who failed to return parental consent and one student who transferred to a school in another country, all other children were assessed to determine their reading and cognitive abilities ( M = 7.6 years, SD = 0.3, range 7.19.1; 58 boys). Students’ reading skills were assessed with the Reading Age Test ( Teste de Idade de Leitura , hereafter TIL; Sucena & Castro, 2010; see description below under Measures). Their cognitive abilities were assessed with the Progressive Raven Coloured Matrices (PRCM; Simões, 2002). The whole sample achieved an average score of 30.98 ( SD = 16.40) in the TIL, and 26.34 ( SD = 4.18) in the PRCM. The following criteria were set to identify poor readers: (a) a score within the 1st quartile (0–25% accuracy) in the TIL test, (b) a score above the 1st quartile in the Portuguese norms of the PRCM test; and (c) teacher’s classification of the child as a poor reader. Thirty children fulfilled the three criteria. 2.2 Participants and Design The 30 children that met inclusion criteria were pseudo-randomly divided into two groups matched on sex, reading skills, and cognitive ability: GraphoGame Fluent condition and GraphoGame Math condition. An additional business-as-usual (BAU) group was formed by randomly selecting 15 students included in the screening procedure who exhibited typical reading development and without cognitive difficulties (mainstream students). Before starting the intervention, one-way Analyses of Variance (ANOVAs) revealed no group differences ( F s < 2.62, ns ) in terms of age, intellectual ability, and vocabulary (assessed with the Peabody Vocabulary Test, Portuguese version from Vicente, Sousa, & Silva, 2011). Descriptive statistics by group are displayed in Table 1. 2.3 Educational Setting Reading and spelling are important contents of the Portuguese curriculum in primary school (Buescu, Morais, Rocha, & Magalhaes, 2015). In Grade 1, teachers focus more on developing children’s decoding skills, whereas from Grade 2 onward they start targeting reading comprehension skills more explicitly. Children are expected to read texts of different genres and to improve their vocabulary. Reading comprehension is promoted by teaching children how to reformulate texts and how to identify and analyze text elements (e.g., sequence of events, main theme, characters’ intentions and emotions). Typically, teaching spelling involves explicit instruction of orthographic rules and rote memorization, trained through dictations CARVALHAIS ET AL.  EFFECTS OF THE PORTUGUESE GRAPHOGAME | 16 and error-finding activities. Of interest in the present study is the lack of use of technology through computer games to promote reading and spelling skills. Table 1. Students’ Characterization by Condition GraphoGame Fluent GraphoGame Math BAU N 15 (7 boys) 15 (6 boys) 15 (7 boys) Age (years) M 7.54 7.51 7.49 SD 0.25 0.24 0.24 min 7.20 7.20 7.10 max 7.90 7.90 7.80 Raven Coloured Matrices (raw score) M 26.27 24.00 27.20 SD 4.74 3.89 3.00 min 18.00 17.00 20.00 max 35.00 31.00 32.00 Vocabulary (Peabody) M 130.33 133.07 134.67 SD 20.02 18.89 19.88 min 91.00 102.00 106.00 max 175.00 184.00 170.00 Note. BAU = business as usual. 2.4 Interventions General procedure The intervention groups played the computer games for approximately 15 min daily for four/five sessions a week over a 6-week period. Each child completed a maximum of 28 sessions for a total of 7 hours of gaming. The school provided a quiet room with six PC computers with Windows XP and an internet connection, where five to six children at a time could play the games. Children played the games individually on separate laptops with headphones after the regular school schedule. This minimized disruption of the rest of the class and ensured that the children in the training programs were exposed to the same contents as their classmates. One of the authors (LC) was present to set up the laptops at the beginning of each 23 | JOURNAL OF WRITING RESEARCH Table 3. Results of the 3 (Condition) x 3 (Testing Time) Repeated Measures ANOVAs. ME of Condition ME of Testing Time Condition x Testing Time Interaction Measures F (2, 42) ηp2 F (2, 41) ηp2 F (4, 84) ηp2 Reading comprehension 18.08*** 0.46 92.94*** 0.82 2.48* 0.11 Reading high-frequency words 20.07*** 0.49 86.59*** 0.81 5.48*** 0.21 Reading low-frequency words 13.83*** 0.40 84.83*** 0.81 7.11*** 0.25 Reading pseudowords 14.71*** 0.41 33.24*** 0.62 5.73*** 0.21 Spelling words 4.22* 0.17 37.80*** 0.65 8.31*** 0.28 Phonological awareness: Implicit rime 7.72** 0.27 12.74*** 0.38 3.39* 0.14 Phonological awareness: Explicit syllable 9.28*** 0.31 10.78*** 0.35 5.57** 0.21 Note . ME = main effect. * p < .05. ** p < .01. *** p < .001. Table 4. Effect Sizes (Cohen’s d ) Computed for all Pairwise Comparisons between Conditions at Posttest and Follow-Up GraphoGame Fluent vs. Math GraphoGame Fluent vs. BAU GraphoGame Math vs. BAU Measures Posttes t Followup Posttest Follow -up Posttest Follo w-up Reading comprehension -0.43 0.32 -1.71 -1.16 -1.36 -1.31 Reading high-frequency words 0.27 0.12 -1.65 -1.62 -1.77 -1.72 Reading low-frequency words 0.79 0.47 -1.02 -1.04 -1.79 -1.50 Reading pseudowords 0.53 0.28 -1.25 -0.90 -1.57 -1.13 Spelling 0.87 0.61 0.23 -0.55 -0.74 -0.81 Phonological awareness: Implicit rime 1.17 1.62 0.49 0.75 -0.97 -1.13 Phonological awareness: Explicit syllable 1.37 1.43 0.34 0.60 -1.12 -1.01 Note. BAU = business as usual. CARVALHAIS ET AL.  EFFECTS OF THE PORTUGUESE GRAPHOGAME | 24 those of the students in the GraphoGame Math increased from pretest to posttest ( p < .001), and then remained stable. 3.2 High Frequency Words Reading Analyses decomposing the Condition x Testing Time interaction showed differences between conditions at pretest, F (2, 42) = 28.19, p < .001, ηp2 = 0.57, posttest, F (2, 42) = 15.99, p < .001, ηp2 = 0.43, and follow-up, F (2, 42) = 14.37, p < .001, ηp2 = 0.41. Again, results showed that students in the GraphoGame Fluent and GraphoGame Math conditions achieved lower performance than students in the BAU condition at all testing times (all p s < .001). Testing time effects within conditions were also significant for the three groups: GraphoGame Fluent, F (2, 41) = 64.55, p < .001, ηp2 = 0.76; GraphoGame Math F (2, 41) = 20.97, p < .001, ηp2 = 0.51; and BAU, F (2, 41) = 13.58, p < .001, ηp2 = 0.40. Follow-up analyses revealed that, in the GraphoGame Fluent and BAU conditions, the number of high-frequency words correctly read increased from pretest to posttest (both p s < .001) and levelled off from posttest to follow-up. For the GraphoGame Math condition, performance in the high-frequency words reading increased progressively throughout the three measurement occasions ( p s < .004). 3.3 Low Frequency Words Reading Tests of simple main effects for the interaction showed differences between conditions at pretest, F (2, 42) = 19.22, p < .001, ηp2 = 0.48, posttest, F (2, 42) = 12.55, p < .001, ηp2 = 0.37, and follow-up, F (2, 42) = 9.18, p < .001, ηp2 = 0.30. At pretest, results showed that students in the GraphoGame Fluent and GraphoGame Math conditions correctly read less words than BAU students (both p s < .001).Even though the effect was not statistically significant ( p = .06), the difference between the two GraphoGame conditions at posttest resulted in a large effect size ( d = 0.79), with students in the Fluent condition performing better than those in the Math condition. This effect decreased at follow-up ( d = 0.47). At both posttest and followup, the performance of students in both GraphoGame conditions was inferior than that of BAU students ( p s < .005). Tests of simple main effects also revealed differences between testing sessions for the GraphoGame Fluent, F (2, 41) = 74.62, p < .001, ηp2 = 0.78, GraphoGame Math, F (2, 41) = 14.14, p < .001, ηp2 = 0.41, and BAU conditions, F (2, 41) = 14.70, p < .001, ηp2 = 0.42. Likewise results on high-frequency words, the number of low-frequency words correctly read increased from pretest to posttest (both p s < .001) and then remained stable for GraphoGame Fluent and BAU conditions, but performance increased gradually from pretest to follow-up for the GraphoGame Math condition (both p s < .01). 25 | JOURNAL OF WRITING RESEARCH 3.4 Pseudowords Reading Simple effect analyses to examine the interaction between condition and testing time revealed significant differences between conditions at pretest, F (2, 42) = 24.32, p < .001, ηp2 = 0.54, posttest, F (2, 42) = 11.96, p < .001, ηp2 = 0.36, and follow-up, F (2, 42) = 5.86, p = .01, ηp2 = 0.22. Results were identical to those obtained for the other word reading measures. Specifically, students in the GraphoGame Fluent and GraphoGame Math condition performed poorer than BAU students in the three measurement occasions (all p s < .02). Concerning time effects for each condition, although there was no significant progress among BAU students ( F < 1), those in the GraphoGame Fluent, F (2, 41) = 16.37, p < .001, ηp2 = 0.41, and GraphoGame Math conditions, F (2, 41) = 30.50, p < .001, ηp2 = 0.78, were observed to increase their performance from pretest to posttest and from posttest to follow-up ( p s < .04). 3.5 Spelling Tests of simple main effects for the interaction revealed differences between conditions at pretest, F (2, 42) = 7.52, p = .002, ηp2 = 0.26, posttest, F (2, 42) = 4.47, p = .02, ηp2 = 0.18, and follow-up, F (2, 42) = 3.76, p = .03, ηp2 = 0.15. At pretest, students in the GraphoGame conditions revealed poorer spelling skills than their peers in the BAU condition (both p s < .004). However, at both posttest and follow-up students in the GraphoGame Fluent condition performed similarly to BAU students and surpassed those in the GraphoGame Math condition ( p s < .05), who continued to show poorer spelling skills than BAU students ( p s < .02). Additionally, results showed differences in spelling performance throughout time, in the GraphoGame Fluent, F (2, 41) = 41.66, p < .001, ηp2 = 0.67, GraphoGame Math, F (2, 41) = 12.32, p < .001, ηp2 = 0.38, and BAU conditions, F (2, 41) = 10.03, p < .001, ηp2 = 0.33. Specifically, results showed that spelling skills increased from pretest to posttest in the GraphoGame Fluent condition ( p < .001); increased from pretest to posttest and from posttest to follow-up in the GraphoGame Math condition ( p s < .002); and increased from posttest to follow-up in the BAU condition ( p < .001). 3.6 Phonological Awareness: Implicit Rime Analyses decomposing the Condition x Testing Time interaction showed differences between conditions at pretest, F (2, 42) = 5.32, p = .01, ηp2 = 0.20, posttest, F (2, 42) = 8.18, p = .001, ηp2 = 0.28, and follow-up, F (2, 42) = 13.10, p < .001, ηp2 = 0.38. Again, at pretest, students in the GraphoGame conditions performed worse than those in the BAU condition (all p s < .01). However, at both posttest and follow-up, students in the GraphoGame Fluent condition achieved a performance level virtually equal to that of students in the BAU condition, which was above that of students in the GraphoGame Math condition (all p s < .003). Tests of simple main effects also revealed differences between testing sessions for the GraphoGame Fluent, F (2, 41) = 14.98, p < .001, ηp2 = 0.42, and GraphoGame Math conditions F (2, CARVALHAIS ET AL.  EFFECTS OF THE PORTUGUESE GRAPHOGAME | 26 41) = 4.90, p = .01, ηp2 = 0.19, but not for the BAU condition ( F < 1). In both GraphoGame conditions, performance increased from pretest to posttest ( p s < .01) and then remained stable. 3.7 Phonological Awareness: Explicit Syllable Analyses decomposing the Condition x Testing Time interaction showed differences between conditions at pretest, F (2, 42) = 8.16, p = .001, ηp2 = 0.28, posttest, F (2, 42) = 10.43, p < .001, ηp2 = 0.33, and follow-up, F (2, 42) = 10.22, p < .001, ηp2 = 0.33. As the results obtained for spelling and implicit rime, students in the GraphoGame conditions achieved a lower performance than students in the BAU condition at pretest (both p s < .001). At posttest and follow-up, students in the GraphoGame Fluent condition performed as good as students in the BAU condition and outperformed those in the GraphoGame Math condition (all p s < .002). Regarding time effects for each condition, although there was no significant progress among BAU students ( F < 1), those in the GraphoGame Fluent, F (2, 41) = 20.28, p < .001, ηp2 = 0.50, and GraphoGame Math conditions, F (2, 41) = 3.36, p = .04, ηp2 = 0.14. Whereas the performance of students in the GraphoGame Fluent condition increased from pretest to posttest ( p < .001) and then remained stable, performance of students in the GraphoGame Math condition remained stable from pretest to posttest and increased from posttest to follow-up ( p = .03). 4. Discussion The present study investigated the effectiveness of a Portuguese version of GraphoGame, which is a technology-enhanced learning environment that aims to improve reading in children who struggle in the initial stages of reading and spelling. These children exhibit difficulties in automatizing letter-sound correspondences and slow phonological decoding mechanisms, which are associated with deficits in word and nonword reading (Ziegler, Perry, Ma-Wyatt, Ladner, & Schulte-Körne, 2003). To store letter-sound correspondences and to achieve rapid automatic retrieval, struggling readers require intensive phonological training (Écalle, Magnan, & Calmus, 2009; Lyytinen et al., 2007). However, in less transparent languages, such as Portuguese, it is important to consider larger units than just phoneme-grapheme correspondences (Richardson & Lyytinen, 2014). Following this premise, we developed the Portuguese version of GraphoGame, which focused on less transparent connections between letters and sounds and included phoneme-grapheme, syllable, rime, and word items. The effectiveness of this program was tested with children showing low reading ability, and effects were examined on reading skills as well as on spelling and phonological awareness skills. Overall, we found no effects of GraphoGame on reading tasks, but clear improvements on spelling and phonological awareness. These results were in line with other reading interventions reporting benefits on letter knowledge and 27 | JOURNAL OF WRITING RESEARCH phoneme awareness, but no effects on other literacy skills, such as reading (e.g., Duff, Hulme, Grainger, Hardwick, Miles, & Snowling, 2014; Regtvoort & van der Leij, 2007). These findings extend current research on GraphoGame effects to an orthography of intermediate depth and to the Portuguese educational system. 4.1 Reading Contrary to our expectations, there were no major improvements in reading comprehension nor in the reading of highand low-frequency words and pseudowords. Due to its focus on phoneme-grapheme conversions (Richardson & Lyytinen, 2014) and on more complex orthographic patterns, trained through syllables, rimes, and whole words (Kyle et al., 2013), we were anticipating GraphoGame to promote an automatization of children’s ability to read fluently and accurately. However, this was not observed. Though this can be partially related to the reduced sample size that could have reduced the power of our study, a similar lack of benefits on reading, including in children at risk for reading difficulties, was already reported in prior studies (Duff et al., 2014; Elbro & Petterson, 2004; Regtvoort & van der Leij, 2007). One explanation for this finding can be the reduced training time provided by GraphoGame. Despite being planned to be short, it might be the case that the 7 hours of training were not enough to improve reading comprehension in struggling readers in Grade 2. Previous research suggested that the most effective interventions tapping phonological awareness skills lasted between 5 to 18 hours of practice spread across several weeks (Ehri et al., 2001). Considering that this computer-based tool has been most tested in transparent and opaque orthographies, little is still known about the time of exposure needed to improve reading in an orthography of depth intermediate, such as Portuguese. Struggling Portuguese readers may require more training and practice to reach the reading level of the mainstreamers (Saine et al., 2011; Torgesen, 2005). Moreover, the emphasis of GraphoGame on training children to sound out the constituent graphemes of words may have affected reading performance, by slowing their reading speed. This effect was already demonstrated in a previous study (Connelly, Johnston, & Thompson, 2001). Word recognition based on phonological decoding has been reported as a slower reading strategy than orthographic processing (Regtvoort & van der Leij, 2007). Additionally, the provided training cannot be restricted to decoding abilities through the training of isolated words. It also needs to include exercises focused on comprehension. Even though the Portuguese GraphoGame included a set of games targeting reading comprehension through sentence-based exercises, results from each player’s playing report revealed that children had a reduced exposure to these games. The majority of playing time was devoted to games aimed to train phonemic awareness and orthographic patterns of consistent and inconsistent words presented in CARVALHAIS ET AL.  EFFECTS OF THE PORTUGUESE GRAPHOGAME | 28 isolation. The benefits of this type of training were clearly observed in spelling and phonological awareness. 4.2 Spelling As anticipated, the Portuguese version of the GraphoGame Fluent improved children’s spelling skills, assessed through a word dictation test. Despite the general growth across all groups, likely explained by the regular teaching at school, children playing this game exhibited a larger growth in spelling throughout the intervention. Both after training and one month later, struggling readers playing GraphoGame Fluent displayed a spelling achievement of the same level of their mainstream peers and above that of struggling readers playing the GraphoGame Math. Similar spelling benefits of GraphoGame have been reported in the literature (Kyle et al., 2013; Saine et al., 2011), indicating the effectiveness of this computerbased tool to promote writing-related skills. Struggling readers seem to need intensive and individual support to reach the level of their counterparts without reading difficulties (Saine et al., 2011). The GraphoGame was found to be an effective aid for teachers to support their at-risk students and improve their ability to correctly spell words. This is an important finding because spelling is a foundational writing skill in primary and middle grades (Limpo & Alves, 2013; Limpo, Alves, & Connelly, 2017; O'Rourke, Connelly, Barnett, & Afonso, 2020). For example, Abbott, Berninger, and Fayol (2010) found that students’ ability to spell words was a longitudinal predictor of their ability to compose text consistently from Grade 1 to 7. Future research should go beyond the examination of GraphoGame effects on spelling isolated words and examine its effects on spelling words in authentic writing tasks (e.g., text production) as well as on other writing-related measures (e.g., syntactic skills). 4.3 Phonological Awareness In line with previous studies, including research with at-risk children (Duff et al., 2014; Regtvoort & van der Leij, 2007), GraphoGame was effective in improving phonological awareness. Our findings revealed that both at posttest and follow-up, students playing GraphoGame Fluent performed as good as mainstreamers, and better than students playing GraphoGame Math. This was observed in both the implicit and explicit phonological awareness tasks. In the implicit task, children were asked to identify whether a pair of words shared or not the same rime. In the explicit task, children were asked to say what was common syllable in pairs of words. GraphoGame improved struggling readers’ ability not only to explicitly identify common syllables across words but also to implicitly recognize if two words share the same rime. Both skills were fully trained in the GraphoGame Fluent. Across several games, children had to identify the listened word among several distractors representing minimal pairs (i.e., words sharing almost sounds with the 29 | JOURNAL OF WRITING RESEARCH target item). The benefits of GraphoGame on phonological awareness is an important finding, as this skill is crucial for beginning reading and spelling (Lyon, Shaywitz, & Shaywitz, 2003; Saine et al., 2011; Snowling, Gallagher, & Frith, 2003). 5. Limitations In addition to the limitations already identified, current findings should be viewed in light of three additional limitations, which may also be useful to guide future research. First, the practical constraints imposed by the inclusion criteria (i.e., struggling readers) and the randomized design of this study resulted in a small sample size (15 children per group), which might have weakened the statistical power of the study. Future studies providing further validity on the Portuguese version of the GraphoGame Fluent should aim for larger samples to examine the replicability of current findings. Second, despite the advantages of including the GraphoGame Math condition – which allowed for a more robust test of the GraphoGame Fluent effectiveness by controlling for motivational effects associated with the computer game – we were not able to collect any measures on math-related skills. This precluded us from examining the successful implementation of this intervention. It would be important to conduct further experimental research on the effectiveness of GraphoGame Fluent that include control conditions implementing technologybased interventions as in here, along with treatment-specific measures testing their effects and confirming their benefits. Third, we did not include a BAU group of struggling readers. The BAU group in this study was composed of children with typical reading development, which allowed us to examine the extent to which GraphoGame improved the skills of struggling readers to the level of their typically developing peers. However, for a more robust examination of GraphoGame effects, future studies should include a BAU group including children with similar reading difficulties as the experimental group. Finally, we did not collect measures regarding the regular reading instruction that students were receiving in the classroom. As discussed in the literature, the regular reading instruction that intervention students receive can weaken intervention benefits (Duff et al., 2014). In the Portuguese educational system, Grade 2 is critical in what concerns the teaching of reading, with a particular focus on contextual and orthographic inconsistencies. The involvement of both struggling readers and mainstreamers in the research project could have also led teachers to give special attention to the development of their reading abilities. CARVALHAIS ET AL.  EFFECTS OF THE PORTUGUESE GRAPHOGAME | 30 6. Future Developments This study constitutes a first step in examining the effectiveness of a Portuguese version of the GraphoGame Fluent. This is a technology-enhanced tool specifically tuned to support Portuguese students’ learning to read in the first years of schooling. As revealed by our findings, the Portuguese version of GraphoGame helped struggling beginning readers to master some of the complexities of the Portuguese language. Despite being a promising tool to assist literacy acquisition in Portuguese, new studies are needed to strengthen the validity of this tool. It would be important to conduct further randomized controlled trials with larger samples and implemented at the national level, including private and public schools and students from different socioeconomic status. Larger studies would also allow an examination of the relationship between children’s performance either at baseline or during GraphoGame training and training effects on reading and writing measures. Based on the current findings, which showed limited effects of GraphoGame on reading comprehension skills, it would be particularly interesting to develop computer-based intervention programs targeting reading-related skills beyond letter-sound correspondences. For example, such a game could aim to develop vocabulary and narrative skills in oral language. This development would be important for two reasons. First, given the strong relationship between oral and written language (Muter, Hulme, Snowling, & Stevenson, 2004), it could be an effective means to promote decoding and comprehension skills as well as spelling and text production abilities. Second, it could be used with children in preschool to promote oral language and pre-literacy skills. By adopting a preventing perspective, this computerized intervention could be beneficial for preschoolers, either with typical development or with an elevated risk for emerging literacy problems. Another avenue for future research would be to create add-on training components specifically aimed to engage parents in the game. Research has already shown the key role that parents may play in children’s learning (Snow, Burns, & Griffin, 1998). For example, children using a computer-based literacy tool implemented by their parents at home for 14 weeks improved phonemic awareness and letter-sound correspondences (Regtvoort & van der Leij, 2007). Currently, there is a reduced knowledge about the effectiveness of computer-based training with parents as tutors. Overall, by targeting other skills and ages, and by considering children’s educational environment, the proposed developments for GraphoGame may constitute a secure foundation for developing literacy. 31 | JOURNAL OF WRITING RESEARCH Note 1. The name GraphoGame is used in this article because this was the name of the tool when this study was implemented. However, the rights to use the name GraphoGame were recently acquired by a private company, thus the current name of the game is GraphoLearn. Acknowledgements The research was supported by the Portuguese Foundation for Science and Technology (Grant SFRH/BPD/ 65831/2009, attributed to the first author). References Abbott, R. D., Berninger, V. W., & Fayol, M. (2010). Longitudinal relationships of levels of language in writing and between writing and reading in Grades 1 to 7. Journal of Educational Psychology , 102 , 281-298. doi:10.1037/a0019318 Babayiğit, S. (2009). Reading and spelling development in transparent alphabetic orthographies: Points of convergence and divergence. In C. Wood & V. Connelly (Eds.), Contemporary perspectives on reading and spelling (pp. 133-148). London: Routledge. Block, H., Oostdam, R., Otter, M., & Overmaat, M. (2002). Computer-assisted instruction in support of beginning reading instruction: A review. Review of Educational Research, 72 , 101-130. doi:10.3102/00346543072001101. Blomert L., & Vaessen A. (2009). 3DM Differential diagnostics for dyslexia: Cognitive analysis of reading and spelling . Amsterdam: Boom Test Publishers. Buescu, H., Morais, J., Rocha, M. R., & Magalhaes, V. M. (2015). Programa e metas curriculares de Portugue  s do ensino ba  sico [Portuguese curriculum and goals of basic education] . Lisboa: Direção-Geral da Educação Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Lawrance Earlbaum Associates. Connelly, V., Johnston, R. & Thompson, G.B. (2001). The effect of phonics instruction on the reading comprehension of beginning readers. Reading and Writing: An Interdisciplinary Journal, 14 , 423-457. doi:10.1023/A:1011114724881 Duff, F., Hulme, C., Grainger, K., Hardwick, S., Miles, J., & Snowling, M. (2014). Reading and language intervention for children at risk of dyslexia: A randomized controlled trial. Journal of Child Psychology and Psychiatry, 55 (11), 1234-1243. doi:10.1111/jcpp.12257 Écalle, J., Magnan, A., & Calmus, C. (2009). Lasting effects on literacy skills with a computerassisted learning using syllabic units in low-progress readers. Computers & Education, 52 , 554-561. doi:10.1016/j.compedu.2008.10.010 Ehri, L. C. (1992). Reconceptualising the development of sight word reading and its relationship to recoding. In P.B. Gough, L.C. Ehri & R. Treiman (Eds .) , Reading acquisition (Vol. 5, pp.107143). London: Erlbaum. Ehri, L. C., & McCormick, S. (1998). Phases of word learning: Implications for instruction with delayed and disabled readers. Reading and Writing Quarterly, 14 (2), 135-163. doi:10.1080/ 1057356980140202 Ehri, L. C., Nunes, S. R., Willows, D. M., Schuster, B.V., Yaghoub-Zadeh, Z., & Shanahan, T. (2001). Phonemic awareness instruction helps children learn to read: Evidence from National Reading panel’s meta-analysis. Reading Research Quarterly, 36 , 250-287. doi: 10.1598/RRQ.36.3.2 CARVALHAIS ET AL.  EFFECTS OF THE PORTUGUESE GRAPHOGAME | 32 Elbro, C., & Petterson, D. K. (2004). Long-term effects of phoneme awareness and letter sound training: An intervention study with children at risk for dyslexia. Journal of Educational Psychology, 96 (4), 660-670. doi:10.1037/0022-0663.96.4.660 Fernandes, S., Ventura, P., Querido, L., & Morais, J. (2008). Reading and spelling acquisition in European Portuguese: A preliminary study. Reading and Writing: An Interdisciplinary Journal, 21 , 805-821. doi: 10.1007/s11145-007-9093-7 Fletcher, J., Lyon, G., Fuchs, L., & Barnes, M. (2007). Learning disabilitiesfrom identification to intervention. New York: Guilford. Gillon, G. (2004). Phonological awareness: From research to practice. New York, NY: The Guilford Press. Gomes, I., & Castro, S. (2003). Porlex, a lexical database in European Portuguese. Psychologica , 32 , 91-108. Retrieved from https://repositorio-aberto.up.pt/bitstream/10216/6931/2/ 83626.pdf Graham, S., Harris, K. R., & Fink Chorzempa, B. (2002). Contribution of spelling instruction, writing, and reading of poor spellers. Journal of Educational Psychology, 94 , 669-686. doi: 10.1037//0022-0663.94.4.669 Hatcher, P., Hulme, C., & Ellis, A. (1994). Ameliorating early reading failure by integrating the teaching of reading and phonological skills: The phonological link-age hypothesis. Child Development, 65 , 41-57. doi:10.1111/j.1467-8624.1994.tb00733.x Hatcher, P., Hulme, C., & Snowling, M. (2004). Explicit phoneme training combined with phonic reading instruction helps young children at-risk of reading failure. Journal of Child Psychology and Psychiatry, 45 , 338-358. doi: 10.1111/j.1469-7610.2004.00225.x Hatcher, P., Hulme, C., Miles, J., Carroll, J., Hatcher, J., Gibbs, S., et al. (2006). Efficacy of small group reading intervention for beginning readers with reading-delay: A randomized controlled trial. Journal of Child Psychology and Psychiatry , 47 , 820-827. doi:10.1111/j.14697610.2005.01559.x Heikkilä, R., Aro, M., Närhi, V., Westerholm, J., & Ahonen, T. (2013). Does training in syllable recognition improve reading speed? A computer-based trial with poor readers from second and third grade. Scientific Studies of Reading, 17 (6), 398–414. doi: 10.1080/10888438. 2012.753452 Kline, R. B. (2005). Principles and practice of structural equation modeling (2nd ed.). New York: The Guilford Press. Knight, S., Shibani, A., Abel, S., Gibson, Ryan, P., Sutton, N., Wight, R., Lucas, S., Sándor, Á., Kitto, K., Liu, M., Vijay Mogarkar, R., & Buckingham-Shum, S.J. (2020). AcaWriter: A learning analytics tool for formative feedback on academic writing. Journal of Writing Research 12 (1), 141-186. https://doi.org/10.17239/jowr-2020.12.01.06 Kujala, J., Richardson, U., & Lyytinen, H. (2010). Estimation and visualization of confusability matrices from adaptive measurement data. Journal of Mathematical Psychology, 54 (1), 196207. doi:10.1016/j.jmp.2008.05.007 Kyle, F., Kujala, J., Richardson, U., Lyytinen, H., & Goswami, U. (2013). Assessing effectiveness of two theoretically motivated computer-assisted reading interventions in United Kingdom: GG Rime and GG Phoneme. Reading Research Quarterly, 48 (1), 61-76. doi:10.1002/rrq.038 Limpo, T., & Alves, R. A. (2013). Modeling writing development: Contribution of transcription and self-regulation to Portuguese students’ text generation quality. Journal of Educational Psychology, 105 , 401-413. doi:10.1037/a0031391 Limpo, T., Alves, R. A., & Connelly, V. (2017). Examining the transcription-writing link: Effects of handwriting fluency and spelling accuracy on writing performance via planning and translating in middle grades. Learning and Individual Differences, 53 , 26-36. doi:10.1016/j.lindif.2016.11.004