Neuroscience Research on Literacy

Jill Kerper Mora

Neuroscience Research: The Reading Brain

In his book The Reading Brain (2008), French neurolinguist Stanislas Dehaene describes two pathways to the semantic region of the brain. One is through phonology of a word as decoded through orthography, the product of visual representations of language sounds (letters) being translated into sounds (in the brain) to render a pronounced word. The other pathway is where the visual image of the word is interpreted directly into meaning as a visual unit without “sounding out” the word. Dehaene proposes that these two pathways operate in the brain simultaneously, but one “gets there” to the semantic region before the other to trigger meaning.

We teachers must ask these questions:  What happens when there is no meaning for the word in the reader’s mental lexicon? Or what happens when the word cannot be understood because the word’s spelling alone does not provide a cue to its meaning within the sentence (context)? For example, the word “bow” has different meanings depending on how it is pronounced. Bowers and Bowers (2017) call words that can only be accurately decoded without reference to their meaning (semantics) “context-dependent” words. This is a term that describes 16% of words in English.

What should a teacher have an emergent reader do when s/he mispronounces the word incorrectly to render its correct meaning in the context of the sentence? Compare the meaning of b-o-w in these two sentences: “I took a bow at the end of my performance…” versus “He shot an arrow from his bow.” A teacher should cue the student to problem-solve this loss of comprehension by looking at the sentence and thinking of the context (language) where the word is used in order to comprehend its meaning. I have not seen any brain research that suggests that this sort of cueing disrupts brain functioning in any way, shape, or form.

Dual Route Models of Reading

Dehaene (2009) describes two parallel paths in the brain to word meaning. One is the spelling-to-sound route and the other is the direct-lexical route. Dehaene states that the route that the brain uses to determine the meaning of a word depends on the word’s relationship with its pronunciation of different types of words:

These brain imaging results dovetail nicely with conclusions from a great many psychological studies of reading. Do we have to pronounce words mentally before we understand them? Or can we move straight from a letter string to it meaning and skip the pronunciation stage? Both may happen, but it all depends on the type of word.” (p. 115-116)

The spelling-to-sound route is taken in the brain by “… Infrequently used words and neologisms move along a phonological route that converts letter strings into speech sounds.”  The direct-lexical route is taken by “… “frequently used words, and those whose spelling does not correspond to their pronunciation, are recognized via a mental lexicon that allows us to access their identity and meaning.” (p. 104).

Dehaene identifies a “letterbox region” in the brain that processes visual information from words’ orthography for both routes of processing, which operate simultaneously and in parallel:

“The brain’s networks for meaning, however, are not limited to simply processing single words…. The process that allows our neurons to snap together and “make sense” remains utterly mysterious. We do know, however, that meaning cannot be confined to only a few brain regions and probably depends on vast arrays of neurons distributed throughout the cortex. All visual stimuli are channeled to the left letterbox region… This package of visual information is then shuttled on one of two main routes: one that converts it into sound, the other into meaning. Both routes operate simultaneously and in parallel–one or the other gets the upper hand, depending on the word’s regularity.” (p. 119)

The neuroscience on reading in the brain has resulted in the reappraisal of models of reading. For example, Binder et al (2005: 678) make the following points about reading models to emphasize the possible neglect of the factor of semantics.

  • Like the parallel dual-route model, the triangle model assumes that all words are processed by the entire system, so there should be no difference in activation patterns for regular and irregular words. Also, like the parallel dual-route model, the triangle model predicts little, if any, activation favoring nonwords over words, since all stimuli activate the orthographic and phonological units.
  • Thus, the parallel dual-route and triangle models make very similar predictions about the effects of lexicality and regularity of pronunciation on brain activity. The principal difference between these models lies in whether the pronunciation of words is facilitated by word codes (the orthographic and phonological lexicons) or by semantic code.
  • In the triangle model, on the other hand, these areas process semantic codes and might therefore be sensitive to concreteness/imageability, taxonomic category, prototypicality, level of specificity, and other semantic factors.
  • The chief distinction between these theories lies in whether the indirect pathway to phonology is mediated by word codes or by semantic codes. The available evidence suggests that these regions are modulated by semantic factors … arguing for a semantic interpretation.

In a critique of current reading research and practice, David Share (2008) contends “… that the extreme ambiguity of English spelling–sound correspondence has confined reading science to an insular, Anglocentric research agenda addressing theoretical and applied issues with limited relevance for a universal science of reading.

According to Share, the dual route model of reading “…accounts for a range of English-language findings, but it is ill-equipped to serve the interests of a universal science of reading chiefly because it overlooks a fundamental unfamiliar-to-familiar and novice-to-expert dualism applicable to all words and readers in all orthographies.” 

Bowers and Bowers (2017) point out the following orthographic challenges attributable to English phonology and morphology:

  • English spelling is based on morphology, etymology and phonology. Phonological challenges stem from 20 vowel phonemes represented by 5 vowel letters.
  • Letter-sound correspondences often depend on the surrounding letters, which requires sub-lexical phonetic analysis. The sub-lexical route generates the wrong phonological transcription for 16% of 8,000 monosyllabic words.
  • Only 84% of these monosyllabic words are regular. Common words are more irregular than less frequently used words. The different spellings of <to>, <too>, and <two> is an example of English spellings code for distinctions in meaning.
  • English use consistent spellings across words with varied pronunciations. For example, the words pronunciations of the base <sign>: signal, signature, design, designate.
  • Morphemic spelling is used to signal grammar and syntax. For example, the –ed ending for past tense is not a phonetic spelling. The –ed has three different pronunciations: pushed, shoved, shouted.
  • Silent letters have a semantic function. For example, the single, silent <e> serves as an orthographic marker letter for the plural cancelling function in words like <please> or <nurse>.

Compare the complexities of English orthography to Spanish orthography based on the phonology and morphology of Spanish (Ardila & Cuetos, 2016): 

  • The average length of Spanish words is 8.76 letters. 60% of all Spanish words have between 7 to 10 letters. The majority of Spanish words are three syllable words (trisílabas).
  • 51% of the syllables in Spanish words are consonant-vowel CV syllables. 11% are vowel-consonant VC words. 88% of Spanish words are combinations of CV, CVC, V, or VC syllables.
  • 75% of Spanish words are stressed on the second to the last syllable (palabras graves). 19% of Spanish words are stressed on the last syllable (palabras agudas).
  • 3% of Spanish words are stressed on the third from the last syllable (palabras esdrújulas). Palabras esdrújulas always have a written accent mark over the stressed syllable.
  • 95% of all Spanish nouns and adjectives are palabras graves that end in a vowel and are stressed on the next to the last syllable.

Neuroscience and Meaning Making

As researchers and as practitioners, it is important to recognize the concept of triangulation of data sources in research studies (Bellido-García, et al., 2022; Hoffman, Ralph & Woolhams, 2015). The extensive research base of miscue analysis performed on oral reading performance and then overlaid with eye movement data provides insights into how readers process text during meaning making (Goodman & Burke, 1973). A third point of triangulation is provided through neuroanatomical brain research that uses functional magnetic resonance imaging (fMRI) to identify activation of regions in the brain during reading (Hoffmann, et. al, 2015).

“Dual-route models hold that all words can be read either via grapheme-to-phoneme rules or through access to orthographic and phonological lexica that are distinct from knowledge of word meaning. In contrast, the Triangle Model proposes that semantic knowledge plays an integral part in pronouncing words correctly. We were able to map specific elements of this cognitive model onto different cortical regions, thereby providing a direct link between cognitive theorizing and neural implementation. Specifically, we provide insights into the division of labor between semantic and phonological processes in supporting reading aloud, which has been a long-standing source of controversy among cognitive models. (p. E3719).

Neuroscientist Steven L. Strauss (2013) argues that fMRI is an inappropriate tool to study reading and dyslexia and that the neuroscientific study of reading and dyslexia needs to incorporate the psycholinguistic nature of meaning construction and its neuroanatomic foundation in cortical–subcortical circuitry. Strauss the reader may not fixate on a word long enough to have fully visualized all of the letters the word contains. Letter–sound mappings differ significantly from what is assumed in fMRI-based research. Eye movement studies reveal that proficient readers do not fixate on fully one third of the words in a visual text display. Therefore, having the reader fixate again on a miscued word may result in more accurate decoding. The spatial and temporal resolution capacity of fMRI is adequate for identifying letter–sound brain regions, but not for identifying meaning-construction neuroanatomy.

Dr. Strauss, together with Ken Goodman and Eric Paulson (2009), claim that reading must be described within a meaning construction psychological paradigm. It is an executive process beyond the technical resolution capacity of fMRI. Its neuroanatomic basis lies in feed-forward cortical–subcortical tracts. Meaning-construction relies on psycholinguistic strategies far more efficient and effective than letter–sound conversion. Different conclusions from neuroscientists and neurolinguistics about the implications of brain research for the pedagogy of reading and writing instruction demand caution in making pronouncements and sweeping claims about what “the brain research says…” and argues against rejection or marginalization of bodies of scientific data that inform literacy instruction (van Heuven & Dijkstra, 2010). Paulson (2008) found that 20%-40% are not fixated on during oral reading performance and that the percentage of miscues made on fixated vs. non-fixated words is comparable. Consequently, he concluded that miscues were not attributable to “missed” or “skipped ” words. Rather, the eyes merely deliver raw data to the brain, while the brain decides what needs attending to in order to derive meaning from the text. This suggests that eye movements are largely controlled by the recognition of meaning in order to generate reliable inferences. Following an explanation of the bottom-up processing theories and the top-down processing theories of reading in the brain, Strauss et al., (2009) state…

“… a meaning-centered, whole language model of reading is scientifically superior to the phonological processing model from the standpoint of neurobiology. As a scientific paradigm, whole language is based on a psycholinguistic model of the reading process, along with corresponding methods of classroom teaching and assessment. The model characterizes reading as an active process of meaning-construction brought about via the reader’s selective testing of meaning-based predictions against confirmatory or disconfirmatory textual and non-textual material. Confirmed predictions are incorporated into the reader’s expanding mental representation of meaning. Disconfirmed predictions are modified or discarded. Inconclusive predictions are tentative and await further confirmatory or disconfirmatory evidence … Neuroimaging findings are entirely consistent with the whole language model, and in no way distinguish the whole language model from the phonological processing model. (p. 022)

Neuroscientist Steven Strauss (2013; e587) concludes that…

The emerging concepts from this research clearly indicate that the higher cortical structures control the transmission of information from the deeper structures. Eye movement analysis, a widely used reading research tool for over a century, simultaneously supports the emerging neuroscientific view of cortical control and the meaning construction model of reading. We conclude that emerging neuroscience provides evidence for the meaning construction view of reading, and that the transactional socio-psycholinguistic character of reading is an instantiation of the memory-prediction model of brain function.”

Application of Neuroscience to Biliteracy Instruction

Dehaene (2009:104) identifies two pathways to word recognition: The spelling-to-sound route and the direct-lexical route. The distinction between the two pathways is based on the word that is being decoded. “Infrequently used words and neologisms move along a phonological route that converts letter strings into speech sound… Frequently used words, and those whose spelling does not correspond to their pronunciation, are recognized via a mental lexicon that allows us to access their identity and meaning.” There are three aspects of a word: 1) el léxico fonológico, which refers to the word as it is pronounced in oral language 2) el léxico ortográfico, or the word as it is spelled or encoded through letter-sound correspondence and spelling patterns, and 3) el léxico semantic, which is the meaning the word conveys in its function as a name or label for an object or concept. This is called the Triangle Model of reading (Dehaene, 2014).

The Triangle Model provides a framework for planning and implementing phonics instruction in Spanish and English to ensure that the teacher address the three aspects of a word during instruction, while avoiding working with words in isolation and out of context (Coltheart, 2006; Liversedge et al., 2012). This contextualized teaching is effective because often the semantics (meaning) of a word resides in its function within a syntactic structure of a sentence. For example, consider the difference in meaning between these two phrases: Un hombre pobre; un pobre hombre. The positioning of the noun before the adjective versus the adjective before the noun determines the meaning of the word “pobre.” Syntax occurs at the phrase and the sentence levels, which make the isolated word undefined in the mental lexicon of the reader. Consequently, a common word may be semantically context dependent. 

Wong, et al. (2016) and Fedeli, et al. (2021) explain the impact of different types of learning experiences on the networking of the bilingual brain. In a comprehensive review of the literature on experiences and the bilingual brain, Wong and colleagues (2016) make this observation: “While the general language network may be similar across languages and even between languages used within a bilingual individual, there appear to be more variations in the way these subnetworks for the component processes are engaged and assembled. For instance, simultaneous acquisition of reading in two orthographies lends itself to divergent pathways for reading in each language, whereas sequential reading acquisition gives rise to largely overlapping reading circuits in both languages. (p. 3)

Fedeli, et al. (2021) draw this conclusion from their research: “Whereas input leads to activation of a single language system in monolinguals, in individuals who speak (or sign) in two languages both systems are simultaneously activated in production and comprehension automatically. L2 Exposure, L2 Proficiency, and L2 AoA (in interaction with Exposure) differentially affected the structural organization of linguistic pathways and the communication between regions relevant for language control. These results are in line with established brain models of dual-language representation and monitoring.” (p. 104978)

The issue in teaching for transfer in biliteracy development then becomes the analysis to distinguish between the concepts and literacy skills that are universal across the languages and are therefore transferable, and the concepts and skills that are language specific A sequence of instruction for initial Spanish literacy is based on the regularities of the grapheme-phoneme (letter-sound) relationships in Spanish orthography (Dehaene, 2015). Priority is given in instruction to teaching letters and letter-sound correspondences that are more frequent and more regular before teaching the more usual, less regular, and more complex letter-sound relationships.

Vocabulary and Concepts in the Bilingual Brain 

Researchers propose two predominant hypotheses about lexical and conceptual representations: One theory is called the word association hypothesis that proposes that a direct association in established between words in two languages. Another theory, called the concept mediation hypothesis is that the connection between the two languages is via an underlying conceptual system through which bilinguals have “lexical access” to words and their meanings (Peña, Kester & Sheng, 2022; Schwartz & Kroll, 2006). The implications of these two theories about how bilingual learners access their knowledge of word meanings in two languages is that if either or both theories are true, both vocabulary learning and concept learning must be addressed through effective instruction. Another factor is the frequency of exposure to words as a determiner of the strength of the associative networks in each of the learners’ languages (Li & Clariana, 2019; Roxbury, et al, 2014). The brain research also supports the notion that there are distinct cerebral regions where Spanish semantics are stored in memory and another where English semantics are stored in a mental lexicon. The mental lexicon or collection of concept units of each language is activated to comprehend text by recognizable phonological (sounds) and orthographic (spellings) representations that the reader identifies by mapping language features onto print.

The implication of this brain activity for dual language teachers is that pathways to access the semantic storage regions must be created. In addition, pathways linking the Spanish and English semantic regions for bilingual learners to take advantage of their full linguistic and conceptual repertoires. When bilingual learners’ vocabulary in both languages is considered, their total vocabulary size is generally comparable to the total vocabulary size of their monolingual peers (Patterson & Pearson, 2022). Researchers also report findings that suggest that bilinguals have a greater brain memory capacity for words because they engage a wider cross-linguistic activation of the lexico-semantic system and use more effective encoding strategies. They have a propensity to make associations at the conceptual level because they have extensive experience directly associating word forms from their two languages. Bilingual learners also develop more skill in the automatic monitoring context as a source of information for clues to novel words (Francis, et al., 2019). 

Below please find citations of research studies to support the information about neuroscience research and the bilingual brain. Thank you for your attention. I invite your feedback and comments. JKM


Ardila, A., & Cuetos, F. (2016). Applicability of dual-route reading models to Spanish. Psicothema, 28(1), 71-75.

Binder, J. R., Medler, R. D., Conant, L. L., & Liebenthal, E. (2005). Some neurophysiological contraints on models of word naming. NeuroImage, 27(2005), 677-693.

Bowers, J. S., & Bowers, P. N. (2017). Beyond phonics: The case for teaching children the logic of the English spelling system. Educational Psychologist, 52(2), 124-141.

Castles, A., Rastle, K., & Nation, K. (2018). Ending the Reading Wars: Reading acquisition from novice to expert. Psychological Science, 19(1), 5-51.

Coltheart, M. (2006). Dual route and connectionist models of reading: An overview. London Review of Education, 4(1), 5-17.

Compton-Lilly, C. F., Mitra, A., Guay, M., & Spence, L. K. (2020). A confluence of complexity: Intersections among reading theory, neuroscience and observations of young readers. Reading Research Quarterly, 55(S1), S185-S195.

Dehaene, S. (2015). Aprender a leer: De las ciencias cognitivas al aula [Learning to read: From the cognitive sciences to the classroom]. Siglo Veintiuno Editores.

Dehaene, S. (2014). El cerebro lector [The reading brain]. Siglo Veintiuno Editores.

Dehaene, S. (2009). Reading in the brain: The new science of how we read. Viking Penguin.

Dehaene, S., & Dehaene-Lambertz, G. (2016). Is the brain prewired for letters? Nature Neuroscience, 19(9), 1192-1193.

Deniz, F., Nunez-Elizalde, A. O., Huth, A. G., & Gallant, J. L. (2019). The representation of semantic information across human cerebral cortex during listening versus reading is invariant to stimulus modality. Journal of Neuroscience, 39(39), 7722-7736.

Ebe, A. (2008). What eye movement and miscue analysis reveals about the reading process of young bilinguals. In A. D. Flurkey, E. J. Paulson, & K. S. Goodman (Eds.), Scientific realism in studies of reading (pp. 131-149). Routledge.

Fedeli, D., Del Maschio, N. S., Simone, Rothman, J., & Abutalebi, J. (2021). The bilingual structural connectome: Dual-language experiential factors modulate distinct cerebral networks. Brain and Language, 220(104978), 1-11.

Fedorenko, E., Blank, I. A., Siegelman, M., & Mineroff, Z. (2020). Lack of selectivity for syntax relative to word meanings throughout the language network. Cognition, 203(104348), 1-24.

Francis, W. S., Strobach, E. N., Penalver, R. M., Martínez, M., Gurrola, B. V., & Soltero, A. (2019). Word-context associations in episodic memory are learned at the conceptual level: Word frequency, bilingual proficiency, and bilingual status effects on source memory. Journal of Experimental Psychology Learning, Memory, and Cognition, 45(10), 1852-1871.

Frost, R. (2012). Towards a universal model of reading. Behavioral and Brain Sciences, 35, 263-329.

Hoffman, P., Ralph, M. A. L., & Woollams, A. M. (2015). Triangulation of the neurocomputational architecture underpinning reading aloud. PNAS Journal Proceedings of the National Academy of Sciences, 112(28), E3719-E3728.

Liversedge, S. P., Blythe, H., & Drieghe, D. (2012). Beyond isolated word recognition. Behavioral and Brain Sciences, 35(5), 293-294.

Li, P., & Clariana, R. B. (2019). Reading comprehension in L1 and L2: An integrative approach. Journal of Neurolinguistics, 50, 94-105.

Li, P., & Grant, A. (2015). Second language learning success revealed by brain networks. Bilingualism: Language and Cognition, 19(4), 657-664.

Patterson, J. L., & Pearson, B. Z. (2022). Bilingual lexical development, assessment, and intervention. In B. A. Goldstein (Ed.), Bilingual language development and disorders in Spanish-English speakers (Third ed., pp. 119-148). Brookes Publishing.

Paulson, E. J. (2008). Miscues and eye movements: Functions of comprehension. In A. D. Flurkey, E. J. Paulson, & K. S. Goodman (Eds.), Scientific realism in studies of reading (pp. 247-264). Routledge.

Peña, E. D., Kester, E. S., & Sheng, L. (2022). Semantic development in Spanish-English bilinguals. In B. A. Goldstein (Ed.), Bilingual language development and disorders in Spanish-English speakers (Third ed., pp. 149-170). Brookes Publishing.

Rastle, K. (2012). Rethinking phonological theories of reading. The Behavioral and Brain Sciences, 35(5), 303-304.

Roxbury, T., McMahon, K., & Copland, D. (2014). An fMRI study of concreteness effects in spoken word recognition. Behavioral and Brain Function, 10(34), 1-14.

Schwartz, A. I., & Kroll, J. F. (2006). Bilingual lexical activation in sentence context. Journal of Memory and Language, 55(2006), 197-212.

Strauss, S. L. (2013). We need a paradigm shift in research on reading and dyslexia: Fundamental problems with fMRI studies of written language processing. Journal of the Neurological Sciences, 333, e579-e628.

Strauss, S. L. (2010). Neuroscience and dyslexia. In A. A. McGill-Franzen, Richard L. (Ed.), Handbook of reading disability research (pp. 79-90). Routledge Taylor & Francis Group.

Strauss, S. L., Goodman, K. S., & Paulson, E. J. (2009). Brain research and reading: How emerging concepts in neuroscience support a meaning construction view of the reading process. Educational Research and Review, 4(2), 021-033.

Taylor, J. S. H., Rastle, K., & Davis, M. H. (2013). Can cognitive models explain brain activation during word and pseudoword reading? A meta-analysis of 36 neuroimaging studies. Psychological Bulletin, 139(4), 966-791.

van Heuven, W., & Dijkstra, T. (2010). Language comprehension in the bilingual brain: fMRI and ERP support for psycholinguistic models. Brain Research Reviews.

Wong, B., Yin, B., & O’Brien, B. (2016). Neurolinguistics: Structure, function, and connectivity in the bilingual brain. BioMed Research International, Article ID 7069274, 22.