Visual stimuli evoke a negative-going wave peaking between 120 ms and 200 ms after stimulus onset, which is often referred to as N170. Although this component is triggered by any visual stimulus, the amplitude strongly depends on the perceptual familiarity of certain stimulus classes. Several studies report larger N170 amplitudes in adult readers to word-like stimuli compared to visual control stimuli such as symbol strings [1–3].
In a longitudinal study, Maurer and colleagues investigated the development of the word-related N170 by comparing words and symbol strings in pre-literate children, literate children and adults [3–5]. In pre-literate children, the N170 did not differentiate between words and symbols , but was higher for words compared to symbols after the first two years of formal reading education . Investigations of the same children in the fifth grade  and the comparison between children and adults [4, 5] showed that the N170 specialization for printed words continues to develop with further reading experience. In skilled adult readers, the N170 was reduced compared to second grade children, suggesting a u-shaped, non-linear development of the N170 specialization for printed words. Adults and fifth graders also have shorter N170 latencies and stronger left lateralization than second grade children .
The findings regarding the visual recognition of pseudowords are not consistent. While some studies found no differences between words and pseudowords [1, 3], other studies reported higher amplitudes for pseudowords relative to words  or higher amplitudes for words relative to pseudowords . These differential findings likely result from the large variability of tasks and different subject groups used across studies.
Brem et al.  demonstrated that a brief training of grapheme-phoneme correspondence in pre-literate children results in an initial specialization for print. Before training, words and symbols activated similar bilateral regions in the posterior occipito-temporal cortex. Afterwards, words elicited higher activations than symbols, mainly in the left occipito-temporal cortex . These findings can be interpreted as evidence for a premature recognition of words that already exists in early childhood. Yoncheva et al.  examined the impact of a short-term training with an artificial script on the N170 in adults. They trained two groups of normal reading adults to associate visual characters with corresponding spoken words. One group was instructed to associate embedded letter-like figures with phonemes (grapheme-phoneme group), while the other group was instructed to focus on the whole visual character (whole-word group). After the training, visual characters produced a left-lateralized N170 in the grapheme-phoneme group and a right-lateralized N170 in the whole-word group. The results were interpreted as evidence for the phonological mapping hypothesis, which states that the left-lateralization of the word-related N170 to trained visual words is associated with the extent to which students focus on grapheme-phoneme-associations .
The N170 amplitude is also increased for other familiar stimuli like faces  or for visual classes for which some individuals have developed a special expertise (e.g. birds , cars , or fingerprints ). Training with novel objects (“greebles”) also leads to an increase of the corresponding N170 . Altogether, these results suggest that visual experience with certain classes of visual stimuli lead to fast specialized processing of these visual stimuli. However, faces and other visual objects produce a bilateral or right-lateralized N170 [10, 15], whereas the word-related N170 is lateralized to the left [1–3].
The source of the word-related N170 is supposed to coincide with a particular area located in the mid-portion of the left fusiform gyrus [9, 16], the putative visual word form area (VWFA, [17, 18]). This assumption is supported by evidence showing that the N170 is generated in left ventral occipito-temporal regions, as demonstrated by intracranial recordings [19, 20] and source localization estimates of electroencephalographic [3, 5, 15, 21] and magnetencephalographic recordings . However, the proposal of a visual word form area in the left ventral occipito-temporal region is controversial [23, 24]. There is evidence that the VWFA is not only devoted to the visual processing of letter strings , but is also associated with naming, viewing or generating verbs [25–27]. Additionally, the VWFA is also activated when words are presented in the tactile (i.e. in Braille [28, 29]) or auditory modality . These findings show that the left ventral occipito-temporal region is a polymodal area showing considerable functional heterogeneity.
Regardless of the underlying brain structures, the N170 reflects the earliest consistent processing of linguistic stimuli and is used to investigate fast recognition processes in good as well as poor readers. A number of studies provide evidence that the fast visual specialization for print is reduced in dyslexic adults [31, 32] and children . Developmental dyslexia is an unexpected difficulty with accurate and fluent reading which occurs despite normal intelligence, motivation and exposure to adequate reading instruction and in the absence of sensory, psychiatric and neurological disorders. The failure to develop fluent reading skills often persists into adolescence and adulthood, while reading accuracy improves over time. Neurobiologically, dyslexic children and adults show reduced activity in two left posterior brain systems, one parieto-temporal and one occipito-temporal .
Functionally illiterate adults are characterized by even greater difficulties during reading acquisition than dyslexic individuals. The term functional illiteracy refers to adults who have attended school for several years but who failed to acquire functional reading skills. Although they received instructions in reading and writing, they left school with literacy skills that are at least three to four years below the expected level of performance . As a result, they can use written language only to a very limited extent; they are unable to read and understand even short sentences . A recent survey concludes that there are about 7.5 million functional illiterates in Germany (14.5% of the adult population ). Similar prevalence rates are reported for other industrialized countries, e.g. 9% for France  or 16% for the United Kingdom . For the United States it is estimated that about 43% of the adult population are not able to use printed materials for everyday activities in an appropriate way . Functional illiteracy is often associated with specific personal obstacles in childhood concerning school (e.g. truancy, inappropriate instructions, repetition of classes) and family environment (e.g. neglect, drug abuse of parents, abuse, numerous siblings etc.). However, these negative experiences do not apply for all individuals with low literacy skills, and are also not sufficient to let someone become functional illiterate . Accordingly, some researchers propose that functional illiteracy results from cognitive deficits coupled with environmental disadvantages [35, 40, 41].
In the present study, we investigated visual word recognition in functionally illiterate adults. Participants were considered as functionally illiterate when the literacy skills were below the expected level of an average student in fourth grade (see Methods section for further details). Previous studies have demonstrated that visual word recognition, reflected by the N170, is strongly dependent on literacy skills and develops when children learn to read [4, 5]. We used an implicit reading task to explore whether the N170 of functionally illiterate adults changes as a result of literacy training. We therefore compared N170 amplitude in a group of functional illiterates with a group of normal reading adults before and after literacy training. We expected that functional illiterates show a distinct N170 for words and symbols since they attended school for several years and therefore have some knowledge in reading and writing. However, the word-related N170 was expected to be smaller than the one of normal readers because of the difference in reading skills.
Functional illiterates have difficulties reading even simple words. Therefore, they probably decode words and pseudowords in a similar way. Consequently, we assumed that words and pseudowords produce a similar N170. Normal readers, on the other hand, read real words easily due to the high familiarity. Since pseudowords are unknown to them, the N170 should be smaller for pseudowords than for real words. For symbols, no differences between functionally illiterate adults and normal reading adults were expected. Moreover, the N170 as response to words and pseudowords should be lateralized to the left, while the N170 for symbols should be similar in both hemispheres or right-lateralized.
We further investigated whether the N170 can still be modulated in adulthood. To answer this question, a group of functional illiterates participated in an intensive literacy training. The N170 for word-like stimuli (words and pseudowords) should increase during the training, whereas the N170 for symbols is not supposed to change. For the group of normal readers, no changes in fast visual recognition processes were expected.
Finally, we compared training-related changes in literacy skills and N170 amplitude between functional illiterate adults taking part in different literacy trainings differing mainly in the intensity of the delivered training. The experimental group participated in a high-intensity training (trained group) and the control group participated in a low-intensity training (untrained group).