Phonological processing deficits as a universal model for dyslexia : evidence from different orthographies

Purpose: To verify the universal nature of the phonological processing deficit hypothesis for dyslexia, since the most influential studies on the topic were conducted in children or adults speakers of English. Research strategy: A systematic review was designed, conducted and analyzed using PubMed, Science Direct, and SciELO databases. Selection criteria: The literature search was conducted using the terms “phonological processing” AND “dyslexia” in publications of the last ten years (2004–2014). Data analysis: Following screening of (a) titles and abstracts and (b) full papers, 187 articles were identified as meeting the preestablished inclusion criteria. Results: The phonological processing deficit hypothesis was explored in studies involving several languages. More importantly, we identify studies in all types of writing systems such as ideographic, syllabic and logographic, as well as alphabetic orthography, with different levels of orthographyphonology consistency. Conclusion: The phonological processing hypothesis was considered as a valid explanation to dyslexia, in a wide variety of spoken languages and writing systems.


INTRODUCTION
Dyslexia is a neurodevelopmental disorder with a strong genetic predisposition, characterized by specific difficulties in reading and spelling that could not be attributable to cognitive disabilities, lack of educational opportunities, socio-cultural environment, or obvious neurological deficits.
Within the past decades, several theories have been proposed in order to explain the diversity of linguistic and cognitive symptoms observed in developmental dyslexia.These theories conceptualize dyslexia as related to deficits that are either phonological, attentional, visual-magnocellular, auditory, or related to automatic learning (1) .
Even though there is a variety of approaches for explaining dyslexia symptoms, phonological processing difficulty has been the major theory explaining such cognitive deficits in dyslexia (2) .The phonological deficit hypothesis suggests that reading deficits can be attributed to a core deficit in manipulating linguistic information, at the phonological level, such as phonological awareness, or the ability to determine the constituent sounds which comprise spoken words.This deficit in phonological awareness leads to difficulty in learning grapheme-phoneme correspondences early on and to later difficulty in learning, decoding skills and spelling (3) .This evidence has largely come from children with difficulties in learning to read in English, a process that has similarities and differences with other languages (4) .
Whether this is a universal model to explain different manifestations, in all ages, spoken or written language contexts remain to be investigated.There have been several different accounts either contradictory or complementary to the phonological processing deficit hypothesis.The majority of studies examining the effects of auditory and visual processing deficits in dyslexia have been conducted in English, an opaque orthography.Therefore, it is essential to determine whether cognitive characteristics, such as the phonological deficits in dyslexia, may differ across languages varying in orthographic consistency.

PURPOSE
The purpose of the study was to determine whether there is enough evidence in the literature for a phonological processing deficit model for explaining dyslexia, in a wide range of writing systems and orthographies.The approach adopted was a systematic review of studies that relate phonological processing and dyslexia.

RESEARCH STRATEGY
The central question was to analyse the universal validity of the phonological processing deficit hypothesis for explaining dyslexia, regardless of the spoken language, writing system or age of population.Furthermore, the review describes whether the evidence for phonological processing difficulties comes from experimental, theoretical or intervention studies.
We conducted a search for articles in PubMed, Science Direct and SciELO, published over the past ten years, in Portuguese or English, with the combination of the terms "phonological processing" AND "dyslexia" and their equivalents in Portuguese ("processamento fonológico" AND "dislexia").The search was performed using the advanced form, in all indices, with items sorted by relevance.On PubMed and Science Direct, two separate searches were conducted: one considering only the articles with open access and also considering other items with restricted access.For PubMed and Science Direct with restricted access, we selected only the 100 most relevant articles of each database.Duplicated articles were excluded from the final sample.

SELECTION CRITERIA
We adopted as selection criteria for the analysis the inclusion of complete original articles and review articles, published in the last ten years (from January 2004 to April 2014), in Portuguese or English.The following combination of terms was used for the search: "phonological processing" AND "dyslexia" and their corresponding terms in Portuguese "processamento fonológico" AND "dislexia".We included general studies on dyslexia and other known comorbidities, such as attention deficit hyperactive disorder (ADHD), specific language impairment (SLI), dyscalculia and dysgraphia.We excluded repeated articles, articles that were not related to the topic, publications on acquired dyslexia, aphasia, psychiatric or neurological diseases, as well as other irrelevant topics for the current discussion (reading difficulties in Down syndrome, Williams syndrome etc.).

DATA ANALYSIS
Initially, the first 100 articles were selected according to the relevance in each database.The first inspection for the criteria was based on reading the titles and abstracts of the open access articles.All repeated articles were excluded as well as those not relevant to the discussion.The same procedure was taken for the restricted access articles.When there was any doubt on the exclusion criteria, a second judge analysed the article.The three authors for this study served as judges for the inclusion or exclusion of the articles.If two of the three judges agreed, the article was excluded or included.When the final database for analysis was completed, all articles were read completely in order to register all relevant details for further analysis.The studies were organized by journal, year of publication, population age, spoken and written language, main goal, experimental approach, type of study (theoretical, assessment, intervention etc.), measures used, and conclusions.In the present study, we concentrated on the discussion of the spoken language and writing system.

RESULTS
The dada was retrieved by means of a systematic review of the literature on PubMed ( From the open access of PubMed database, we excluded 27 articles, 10 for not following the inclusion criteria, 9 for having different goals from the present study, 3 were repeated in other database, 2 were publish in a language other than English or Portuguese and 3 for the type of study (case study).
From the open access of Science Direct database, we excluded 92 articles, 67 for not following the inclusion criteria, 11 for having different goals from the present study, 3 were repeated in other database, 5 were publish in a language other than English or Portuguese and 6 for the type of study (case study).
From the SciELO database, when using the search of terms in English, we excluded 2 articles (one for the language of publication and one for the type of study).When using the search terms in Portuguese, we excluded 3 articles for repetition.
This way, at final analysis, we included 47 articles from PubMed, 14 articles from Science Direct and 7 articles from SciELO, in a total of 68 articles with open access.
Considering the articles found in restricted access, we selected the 100 most relevant ones, from PubMed and Science Direct.The repeated articles found in other database or at the open access were excluded.From the database PubMed, we excluded 41 articles, 7 for not following the inclusion criteria, 12 for having different goals from the present study, 21 were repeated in other database and 1 for the type of study (case study).
From de database Science Direct, we excluded 40 articles, 13 for not following the inclusion criteria, 2 for having different goals from the present study, 18 were repeated in other database and 7 for the type of study (case study).In summary, after inspection we included 59 articles from PubMed and 60 from Science Direct, a total of 119 articles from restricted access.The articles found on the SciELO database are all in open access, therefore, there were no articles found on the restricted access group.
A total of 187 articles were registered and classified according to the categories chosen for further analysis (Table 1).The complete list is in Appendix 1.
In terms of the distribution of years of publication, there is a clear tendency of increasing the number of articles after 2009 (Figure 1).It is important to note that this review was performed on April 2014, which explains the low number of articles found in that specific year.
We found that the selected articles were published in 70 different periodicals in total, but only 14 have published five or more articles from 2004 to 2014. Figure 2 shows that Neuropsychologia is the journal with the greater number of articles (n=22).Several countries contributed for the discussion on whether phonological processing deficit is a universal theory for explaining dyslexia.As shown in Figure 3, The United States (n=40) and United Kingdom (n=24) were the two more productive countries in terms of publications.Brazil was the only country in South America with scientific publications on the topic (n=6).
Regarding the spoken languages, after inspection of all articles selected in this review, we identified 18 different languages represented.Out of the 187 articles, three studies investigated phonological processing deficits in bilingual populations (Finish-Swedish; Spanish-Swedish; English-Chinese) and one study compared phonological skills in dyslexic Spanish-, English-and Chinese-speaking children.Six articles were characterized as theoretical reviews, so there was not a specific language being investigated.Table 2 depicts the 11 most frequent languages studied.English speaking was responsible for 42% of the studies found in this review, as shown in Figure 4.
Finally, after inspection of all articles, we registered five writing system classifications: alphabetic opaque, alphabetic transparent, ideographic, logographic-syllabic, and semisyllabic (Figure 5).The most frequent writing system was alphabetic opaque (53%) but alphabetic transparent orthographies were present in 34% of the publications.The evidence of the relation between phonological processing disorders and dyslexia comes from clinical studies, both in evaluation and intervention studies.

DISCUSSION
There is a wide spread knowledge on the phonological difficulties found in individuals with dyslexia.Some authors claim that underlying phonological processing deficits would exist for all languages, but that there would be differences in the severity of written language impairments, due to differences in orthographic consistency (5) .In other words, we investigated whether this hypothesis also holds for dyslexia in more consistent orthographies, since the phonological code is more accessible in these languages, unlike in English.The present study had the purpose to explore, with a systematic review of the literature, the relation of phonological processing deficits hypothesis with dyslexia, in terms of their universal validity.
The results show a steady increase in publications from 2004 to 2014, which indicates the relevance of the debate and the need for understanding the origin of dyslexia in recent years.From the 187 articles, some are theoretical in nature (6) , others describe intervention programs based on phonological abilities (7) , and most of them are experimental or clinical studies (8,9) .In terms of the numbers of articles published, we found a wide variety of periodicals with relevant publications showing the interdisciplinary status of the discussion.The papers were found in journals representing the following areas of interest: Education, Psychology, Speech-Language Pathology, Audiology, Linguistics, and Computer Sciences.
To investigate the nature of word reading in various languages, a meta-analysis provided support for the existence of a universal reading network consisting of the left superior temporal gyrus (LSTG), the left inferior frontal gyrus (LIFG), the left occipitotemporal region, and the midfusiform Gyrus (10) .The results of this systematic review show that the universal nature of reading holds for reading disabilities, such as the ones found in dyslexia.The phonological nature of such deficits is evident for different spoken languages and different types of writing systems and orthographies (11) .
From the studies selected in the present review, it is clear that reading problems associated with dyslexia differ in regular orthographies such as Finnish (12) as compared to less regular orthographies such as French (13) .However, the underlying cause found in phonological processing skills holds for all levels of orthographic consistencies.
The relation between phonological and orthographic processing is also explored.In recent studies, the associations between auditory temporal processing and phonological processing, and between visual processing and orthographic processing, have received some support, but a lot of criticism as well (14) .Although it is not conclusive from the analysis presented here, only a small number of children with dyslexia have reported auditory or visual processing deficits.This heterogeneity of the dyslexic population may have led to such contrasting results.
The origin of phonological processing abilities in dyslexia remains to be established.Some studies are already investigating whether the failure is found on phonological representations or in the process to access these representations during reading (15) .No matter which research approach is chosen, cross-linguistic and multidimensional aspects of dyslexia have to be considered.

CONCLUSION
The phonological processing hypothesis was a valid explanation for dyslexia symptoms in a wide variety of spoken languages and writing systems.
The findings of this review add to a growing number of studies to suggest that the relationship between phonological abilities and reading is influenced by the characteristics of the orthography.The exact nature of such phonological deficits should be subjected to cross-linguistic comparisons, taking into account systematic differences of the orthographic and phonological characteristics of the languages.
*ALGPN was the principal investigator and responsible for conception and study design, data analysis, interpretation of data, correction of written manuscript and final approval of the version to be published.ECF was responsible for data acquisition, data analysis, manuscript editing, revising the study critically, final approval of the version to be published.JPAB was responsible for data acquisition, data analysis, revising the study critically, final approval of the version to be published.

Figure 5 .
Figure 5. Distribution of writing systems and their classifications represented in the review Considering the search for open access articles, we located 74 articles on PubMed, 106 on Science Direct and 12 articles on SciELO, 8 articles using the terms in English and 4 articles considering the search in Portuguese.

Table 1 .
Total number of articles found on the general search and the final number of articles after exclusions

Table 2 .
Distribution of most frequent languages investigated -related skills in Chinese and English among Hong Kong Chinese children with and without dyslexia J Exp Child Psychol 182.Ramus 2014 Neuroimaging sheds new light on the phonological deficit in dyslexia Trends Cogn Sci 183.Litt and Nation 2014 The nature and specificity of paired associate learning deficits in children with dyslexia J Mem Lang 184.Hämäläinen et al. 2014 Event-related potentials to tones show differences between children with multiple risk factors for dyslexia and control children before the onset of formal reading instruction visual and phonological representations in the processing of written words by readers with diagnosed dyslexia: evidence from a working memory task Ann Dyslexia 32.Liu et al. 2009 Speech perception deficits by Chinese children with phonological dyslexia J Exp Child Psychol 33.Abrams et al. 2009 Abnormal cortical processing of the syllable rate of speech in poor readers J Neurosci Appendix 1. List of the 187 articles included for the analysis Phonological processing and dyslexia CoDAS 2014;26(6):509-19 2013 The time course of reading processes in children with and without dyslexia: an ERP study Front Hum Neurosci 164.White-Schwoch and Kraus 2013 Physiologic discrimination of stop consonants relates to phonological skills in pre-readers: a biomarker for subsequent reading ability?Front Hum Neurosci 165.Power et al. 2013 Neural entrainment to rhythmic speech in children with developmental dyslexia Front Hum Neurosci 166.Soltész et al. 2013 Differential entrainment of neuroelectric delta oscillations in developmental dyslexia PLoS One 167.Maïonchi-Pino et al. 2013 Is the phonological deficit in developmental dyslexia related to impaired phonological representations and to universal phonological grammar?J Exp Child Psychol Appendix 1. Continuation Phonological processing and dyslexia CoDAS 2014;26(6):509-19