the effect of syntax on reading in neglect dyslexia

The effect of syntax on reading in neglect dyslexia

Naama Friedmanna♥, Lital Tzailer-Grossab, Aviah Gvionacd aTel Aviv University, bLoewenstein Hospital Rehabilitation Center,

cOno Academic College, dReuth Medical Center

Individuals with text-based neglect dyslexia omit words on the neglected side of the sentence or text, usually on the left side. This study tested whether the syntactic structure of the target sentence affects reading in this type of neglect dyslexia. Because Hebrew is read from right to left, it enables testing whether the beginning of the sentence and its syntactic properties determine if the final, leftmost, constituent is omitted or not. The participants were 7 Hebrew-speaking individuals with acquired left text-based neglect dyslexia, without syntactic impairments. Each participant read 310 sentences, in which we compared 5 types of minimal pairs of sentences that differed in the obligatoriness of the final (left) constituent. Complements were compared with adjuncts, obligatory pronouns were compared with optional resumptive pronouns, and the object of a past tense verb was compared with the object of a present tense verb, which can also be taken to be an adjective, which does not require an object. Questions that require a verb were compared with questions that can appear without a verb, and clauses that serve as sentential complements of a verb were compared with coordinated clauses, which are not required by the verb. In addition, we compared the reading of noun sequences to the reading of meaningful sentences, and assessed the neglect point in reading 2 texts. The results clearly indicated that the syntactic knowledge of the readers with neglect dyslexia modulated their sentence reading. They tended to keep on reading as long as the syntactic and lexical-syntactic requirements of the sentence had not been met. In 4 of the conditions twice as many omissions occurred when the final constituent was optional than when it was obligatory. Text reading was also guided by a search for a "happy end" that does not violate syntactic or semantic requirements. Thus, the syntactic structure of the target sentence modulates reading and neglect errors in text-based neglect dyslexia, suggesting that the best stimuli to diagnose mild text-based neglect dyslexia are sentences in which the leftmost constituent is optional, and not required by syntax. Another finding of this study is a dissociation between neglect dyslexia at the text and at the word levels. Two of the participants had neglect dyslexia at the text level, manifested in omissions of words on the left side of text, without neglect dyslexia at the word level (namely, without omissions, substitutions, or additions of letters on the left side of words).

1. Introduction

Text-based neglect dyslexia is a reading disorder in which the readers neglect the words in the

space contralateral to the lesioned hemisphere, usually on the left part of the sentence or text

(Cubelli, Nichelli, Bonito, De Tanti, & Inzaghi, 1991; Ellis, Flude, & Young, 1987;

Kinsbourne & Warrington, 1962; Schwartz, Ojemann, & Dodrill, 1997; Subbiah & ♥ Acknowledgment: This research was supported by the Israel Science Foundation (grant no. 1296/06,

Friedmann). We thank Yaron Sacher and Nachum Soroker for their valuable comments.

Syntax affects reading in neglect dyslexia 2

Caramazza, 2000; Worthington, 1996; Young, Newcombe, & Ellis, 1991; see also Vallar,

Burani, & Arduino, 2010 for a review). Because Hebrew, the language tested in this study, is

read from right to left, text-based neglect dyslexia in Hebrew is usually manifested in the

omission of the ends of sentences, which allows for the assessment of the effect of the

syntactic structure of the beginning of the sentence on the rate of omissions of its end.

The aim of this study was to test whether preserved syntactic knowledge modulates reading in

neglect dyslexia by impelling the participants to allocate attention to the neglected hemi-space

until the syntactic requirements of the sentence are met. Our general hypothesis is that

individuals with acquired text-level neglect dyslexia whose syntactic ability is preserved will

tend to maintain the grammaticality of the sentences they read. Thus, they will try not to omit

obligatory constituents, and will tend to stop reading the sentence only at points that create a

grammatical sentence. We tested this hypothesis using various minimal pairs of sentences,

comparing sentences in which the final constituent is obligatory to similar sentences in which

the final constituent is optional.

Whereas we are not aware of previous research on the effect of syntax on reading in text-

based neglect dyslexia, some evidence from two sources might be suggestive: studies of a

possible effect of the meaning of the words read on reading in text-based neglect dyslexia,

and lexical and morphological effects on reading in word-based neglect dyslexia.

As for evidence for the effect of the meaning of the words read on text-based neglect dyslexia,

Schwartz et al. (1997) report on the reading of 64 English-speaking epileptic patients. Under

the effect of sodium amobarbital injection to the right hemisphere, the epileptic patients

substituted and deleted words on the left side of the sentences. Interestingly, words were

deleted more frequently when the sentence still made sense without them, semantically and

syntactically. These patients also made word substitutions on the left of the sentences,

substituting the target word with a syntactically similar word, such as reading “a” instead of

“the”. The analysis of the sentences in which no substitutions occurred revealed that

substitution of the left word would have created nonsensical sentences. Thus, there are

indications that reading can be affected by the properties of the sentence (although it should

be kept in mind that these patients did not have neglect dyslexia, and that they substituted

whole words on the left side of the line, whereas individuals with text-based neglect dyslexia

rarely substitute whole words on the left side of the text. Patients with text-based neglect

dyslexia usually omit whole words, Young et al., 1991). Another relevant study was reported

by Kartsounis and Warrington (1989). They describe a man with left text-based neglect


dyslexia, whose reading was affected by properties of the sentence he read. When the

sentences made sense, he read more sentences correctly and omitted fewer words on the left

than when the target sentence was semantically implausible. Manipulating the semantic

coherence of the sentences (by scrambling the words) affected his reading, with more word

omissions in less coherent sentences. The same pattern was evinced in his performance in

visuospatial tasks, which was significantly better when the stimuli were meaningful. Karnath

and Huber (1992) also reported an effect of plausibility of the sentence on its reading. They

tested a participant with text-based neglect dyslexia, and found that his reading accuracy and

eye movements were affected by the acceptability of the sentence he read. The participant

read written stories composed of 9 sentences each, in which sentences continued in the next

line. The stories were manipulated according to whether an omission of the first (left) words

on the second line would have created linguistically plausible continuation of the sentence or

not. The findings were that the participant omitted the first words that still maintained the

acceptability of the sentences in 80% of the sentence continuations that were acceptable with

an omission, whereas only 14% omissions were made in the sentences in which omission was

inacceptable. The patient's eye movements were also affected by the linguistic demands of the

sentences, as although his return sweeps typically ended in the middle of the second line, they

were followed by sequences of short saccades backwards toward the beginning of the second

line, which tended to stop as soon as linguistically acceptable continuation of the text was

found.

These data thus suggest that sentence meaning and structure may affect reading in text-based

neglect dyslexia. Additional suggestive evidence for a possible linguistic effect on reading in

neglect dyslexia comes from lexical effects on word-based neglect dyslexia. In word-based

neglect, which can occur independently of visuospatial neglect or text-based neglect dyslexia,

the readers neglect (omit, substitute, or add) letters in one side of the word, typically the left

side. Several studies found that the lexical properties of the target word such as lexicality,

frequency, morphological structure, orthographic neighbourhood, and regularity, as well as

the lexicality of the response affect the reading errors of some individuals with word-based

neglect dyslexia (Arduino, Burani, & Vallar, 2002, 2003; Arguin, & Bub, 1997; Behrmann,

Moscovitch, Black, & Mozer, 1990; Caramazza & Hillis, 1990; Cubelli et al., 1991; Ellis et

al., 1987; Ellis, Young, & Flude, 1993; Kinsbourne & Warrington, 1962; Patterson & Wilson,

1990; Reznick & Friedmann, 2009; Riddoch, Humphreys, Cleton, & Fery, 1990; Stenneken,

van Eimeren, Keller, Jacobs, & Kerkhoff, 2008; see Vallar et al., 2010 for a review),

https://www.researchgate.net/publication/47460475_Lexical_effects_in_left_neglect_dyslexia_A_study_in_Italian_patients?el=1_x_8&enrichId=rgreq-8f744098-ffdc-4c4c-9f71-0ce30d4d2c79&enrichSource=Y292ZXJQYWdlOzUxMjI1MzEyO0FTOjk4OTkwODI5ODY3MDE5QDE0MDA2MTI2NjE3NTA=

https://www.researchgate.net/publication/45660992_Neglect_dyslexia_A_review_of_the_neuropsychological_literature?el=1_x_8&enrichId=rgreq-8f744098-ffdc-4c4c-9f71-0ce30d4d2c79&enrichSource=Y292ZXJQYWdlOzUxMjI1MzEyO0FTOjk4OTkwODI5ODY3MDE5QDE0MDA2MTI2NjE3NTA=

https://www.researchgate.net/publication/233046009_Lexical_Constraints_on_Reading_Accuracy_in_Neglect_Dyslexia?el=1_x_8&enrichId=rgreq-8f744098-ffdc-4c4c-9f71-0ce30d4d2c79&enrichSource=Y292ZXJQYWdlOzUxMjI1MzEyO0FTOjk4OTkwODI5ODY3MDE5QDE0MDA2MTI2NjE3NTA=


indicating that preserved lexical knowledge as well as morphological considerations modulate

these patients’ reading. The findings regarding lexical and morphological effects on word

reading in word-based neglect dyslexia may suggest that, similarly, syntactic knowledge may

affect the reading of sentences and text in text-based neglect dyslexia.

In the following sections we first assess the syntactic abilities of the participants, and then test

in 8 experiments the effect of the syntactic structure of the sentences they read on the rate of

their neglect dyslexia errors.

2. Participants: background, syntactic abilities, and single word reading

Seven Hebrew-speaking individuals with acquired text-based left neglect dyslexia following

right-hemisphere damage participated in the study. They were 24-81 years-old (mean age =

59), 6 of them had neglect following right hemisphere CVA and one sustained a vast right

hemisphere damage with skull fractures and shards from a bomb. All participants had pre-

morbidly full control of spoken and written Hebrew – six of them had Hebrew as their native

language, and one participant, OM, spoke Hebrew for 48 years prior to his stroke. The BIT

(Behavioral Inattention Test, Wilson, Cockburn, & Halligan, 1987) indicated visuospatial

neglect for all participants, and all of them had a score below the cutoff in the conventional

subtests of space-based neglect. Individuals with severe impairment of reading at the single

word level were excluded from the study on the basis of a single word reading test and the

number of errors they had at the word level in reading sentences. None of the participants had

aphasia or developmental language or reading disorders. All of them had normal or corrected-

to-normal vision. According to their medical files, none of them had hemianopia.1 See Table

1 for background information about the participants, and Appendix A for CT scans of four of

the participants.

1 The assessment of the participants' visual fields has been done before we tested them, by the neurologists in the

rehabilitation centers. This assessment typically involves a clinical confrontation test or a computerized perimetry test or both, in which the patient provides a verbal report upon stimulus presentation to determine whether he/she detected the presence of the stimulus or not; these assessments usually include a confrontation test with maximal rightward shift of the head on the trunk and then maximal rightward shift of the gaze. These assessments indicated that none of the participants had hemianopia, but one should bear in mind the difficulty in determining whether or not a patient with visuospatial neglect has hemianopia.


Table 1

Background information on the participants with neglect dyslexia.

Occupation Age Sex

Education Etiology BIT

object

BIT

space

Word

neglect

Text

neglect

General

neglect

Participant

Teacher 70 F 15 R parietal infarct 7 119 mild left Left LA

School secretary 81 F 12 R temporo-parietal

infarct incl. internal

capsule

mild left Left BK a

Warehouse manager 58 M 10 R temporo-fronto-

parietal infarct

4 42 mild left Left SA

Company owner 58 M 12 R CVA in the

MCA territory

1 124 mild left Left AS

Border policeman 24 M 12 Trauma – R temporo-

parietal hemorrhage

and metal shards, R

craniotomy

2b 90 no left Left DS

Practical electrical

engineer

56 M 14 R frontal infarct 1 125 no left Left NA

Apparel trader 64 M 12 R basal hemorrhage 0 84 mild left Left OM

R – right hemisphere BIT space level: line crossing, letter cancellation, star cancellation, line bisection (MAX- 139, neglect – below 125). BIT object level: figure and shape copying, representational drawing bisection (MAX- 7, neglect below 6). a We could not administer BIT to BK. bDS did not do the drawing subtest, so his object-level BIT score is out of 4.

2.1. Assessment of syntactic abilities

2.1.1. Syntactic tests

Before asking whether syntax affects reading in neglect dyslexia, we needed to assure that the

syntax of the participants with neglect dyslexia was unimpaired. For this aim we assessed the

syntactic abilities of the participants with neglect dyslexia, using 4 syntactic tests from

BAFLA test battery (Friedmann, 1998): relative clause production, tense and agreement

completion, embedded and coordinated sentence repetition, and preposition completion. We

chose these tests because they assess the structures that are typically most difficult for

individuals with syntactic impairments, and hence most sensitive to syntactic impairment.


Specifically, because the target sentences in the experiments included embedded sentences,

coordinated sentences, verbs that differ in inflection, relative clauses, and prepositions, we

tested all these aspects in tasks that do not require reading. Previous studies showed, using

these tests, very poor performance by individuals with agrammatic aphasia (Friedmann, 2001,

2006), children with developmental syntactic language impairment (Fattal, Friedmann, &

Fattal-Valevski, in press; Friedmann & Novogrodsky, 2004, 2007, 2011; Novogrodsky &

Friedmann, 2006), and orally-trained children with hearing impairment (Friedmann &

Szterman, 2006, 2011).

The relative clause production task elicited relative clauses using a preference task

(Friedmann & Szterman, 2006; Novogrodsky & Friedmann, 2006). The participants were

presented with a short story about two people and were required to choose which of these

people they preferred to be. The task was constructed in such a way that the choice would

have to be formed as a relative clause. There were 12 items per participant, 6 eliciting subject

relatives, and 6 eliciting object relatives. The questions that elicited subject relatives described

two people (two men for a male participant, two women for a female participant) performing

two actions (for example, One woman gives a present, one woman receives a present; which

woman would you rather be?). The questions that elicited object relatives described two

people who are the themes of an action performed by two different figures or two actions

performed by the same figure (The doctor examines one woman, the nurse examines one

woman, which woman would you rather be?). The participants were requested to choose the

figure they preferred to be starting with "The wo/man that…". In that way, a subject relative

(The woman that receives a present) or an object relative (The woman that the nurse

examines) are elicited. The order of the subject- and object relative target sentences was

randomized. Hebrew-speaking individuals with agrammatism typically fail to produce both

subject relatives and object relatives (Novogrodsky & Friedmann, 2006).

The tense and agreement completion task included 24 sentences. The participants heard the

first sentence, which included a verb inflected for tense and agreement (person, gender, and

number) and an infinitive verb, and the beginning of a second sentence in which the verb was

missing. They were then requested to complete the sentence with a verb in the correct

inflection (the infinitive verb from the first sentence, inflected according to the tense and

subject of the second sentence). For example, "The girls wanted to draw, so they took a paper

and crayons and… (drew-past,3rd person, plural, feminine)". Hebrew-speaking agrammatic

aphasics typically fail in the tense inflection of the completed verbs, providing the correct

https://www.researchgate.net/publication/228352771_Speech_Production_in_Broca's_Agrammatic_Aphasia_Syntactic_Tree_Pruning?el=1_x_8&enrichId=rgreq-8f744098-ffdc-4c4c-9f71-0ce30d4d2c79&enrichSource=Y292ZXJQYWdlOzUxMjI1MzEyO0FTOjk4OTkwODI5ODY3MDE5QDE0MDA2MTI2NjE3NTA=


verb in incorrect, random, tense inflections (Friedmann, 2001, 2006; Friedmann &

Grodzinsky, 1997)

The embedded and coordinated sentence repetition task included 22 sentences, half with

clauses embedded to a verb (sentential complement) or to a noun (relative clause), and half

with coordinated clauses. The subordinated and coordinated sentences were closely matched

in words, constituents, and length. The participants heard a sentence, and were asked to count

to 5 and then repeat the sentence as accurately as possible. Hebrew-speaking individuals with

agrammatism typically fail to repeat any kind of embedded structure in this task (Friedmann,

2001, 2006). The counting to 5 was used to prevent rehearsal in the phonological loop

(Baddeley, 1997), and hence to preclude phonological echoing.

The preposition completion task included 76 sentences in which a preposition was missing

from the PP complement of the verb or the PP adjunct (The woman waited __ the rain; The

baby cried __ bed). The participants were requested to complete the sentence with the correct

preposition.

2.1.2. Results of the syntactic tests

The results of the syntactic tasks indicate that all the participants had good syntactic abilities.

They had almost no errors in inflection completion (Average 98% correct, SD = 3) and no

errors at all in the preposition completion tasks (each participant performed 100% correct). In

the relative clause elicitation task, they produced all the target subject relatives and most of

the target object relatives (Average 91% correct, SD = 16). Even in cases they provided a

response to the object relative condition that was not a relative clause, 100% of their

responses were grammatical and the participants never omitted the embedding marker, in

marked contrast to individuals with agrammatism. Similarly, in the embedded sentence

repetition task they had no syntactic errors (100% correct syntactic structures) and almost no

inflection errors (97% correct inflections, SD = 4). They produced almost only grammatical

sentences in this task (99.3% grammatical sentences – with one a single ungrammatical

repetition).

2.2. Oral reading of single words

Because we were interested in the way the beginning of the sentence affects the reading of the

final (left-hand) constituent of the sentence, we had to make sure that the participants could

read the first part relatively well, namely, that they did not have severe reading problems at


the word level. We therefore excluded from the study individuals who had severe reading

impairment at the word level.

2.2.1. Single words reading test

To assess the participants' reading at the single word level, we used the TILTAN screening

test (Friedmann & Gvion, 2003). The Tiltan test includes 128 single words of various types

that can reveal the different types of dyslexia: First of all, for the identification of left neglect

dyslexia at the word level, all the words in the list are such that when read with a neglect error

of the left side, another existing word can be created (such as rice, which can be read as nice,

ice, or price following a letter substitution, omission, or addition error respectively); for 104

of the 128 words, right neglect errors (letter substitution, omission, or addition on the right of

the word) create another existing word. To detect letter position dyslexia (Friedmann &

Gvion, 2001; Friedmann & Rahamim, 2007), the test included 63 words for which a

transposition of middle letters can create another existing word (words like trail , beard, and

pirates). The test also included 52 words with a lexical potential for transposition that

involves an exterior letter (such as own than can create won). All the words in the test had at

least 6 orthographic neighbors, for identifying visual dyslexia. The test also included function

words and morphologically complex words, for identifying deep dyslexia and some additional

types of dyslexia. Another type of stimuli in the test was words that are sensitive to surface

dyslexia: Every word in Hebrew is sensitive to surface dyslexia, because no word can be read

unambiguously via the sublexical route. The test included the types of irregular words that are

most sensitive to surface dyslexia: words that are irregular beyond the underspecification of

vowels, and potentiophones – words that can be read via grapheme-to-phoneme conversion as

other words (like desert and dessert, or now, which can be read as know). Thirty word pairs in

which a between-word migration creates other existing words were included in the test to

identify attentional dyslexia. Because migrations between words can occur vertically as well

as horizontally (Friedmann, Kerbel, & Shvimer, 2010), and the single words were presented

in a list, one above the other, we analyzed in Table 4 both vertical and horizontal migrations

between words.

2.2.2. Word reading results

The results of the single word reading task, presented in Table 2, indicate that five of the

participants had mild word-based neglect dyslexia (with neglect errors on 7%-13% of the


words2) in addition to their text-based neglect dyslexia, and none of them made more than

25% errors in reading single words.

Interestingly, two of the participants, DS and NA, had only a single error in a letter on the left

side of the word, an error rate that did not differ significantly from that of 132 healthy

Hebrew-reading controls, p > .05. Thus, although they had text-based neglect dyslexia, they

did not have neglect dyslexia at the word level, suggesting a dissociation that was rarely

reported in the literature, of text-based neglect dyslexia without word-based neglect dyslexia.

As far as we know, such a dissociation was previously reported only for the patient described

in Kartsounis and Warrington (1989). This patient had a severe text-based neglect, which led

to omissions of words on the left side of the text (he did not read more than three words from

the extreme right of each line). In contrast, in single words, he had no or almost no neglect

errors on the left of the words. In reading 22 proper names he made no letter errors at all, and

in reading 96 compound words he made two errors on the left word of the compound. In one

case this error was on the leftmost letter, in the other case it was in the left and middle of the

word. Kartsounis and Warrington summarize this patient's report by saying that "In spite of

the fact that paralexic errors affecting the beginnings of words is common in patients

presenting with gross neglect, the patient showed no evidence of neglect dyslexia." (p. 1257).3

Indications for dissociation of the other direction, with neglect dyslexia at the word level but

not at the text level, are more often reported (De Lacy Costello & Warrington, 1987 [whose

patient had a non-lateralized impairment at the text level, without word omissions on the end

or the beginning of any line, together with word-based neglect dyslexia]; Haywood &

Coltheart, 2001 [whose patient also made several errors on the letters on the right side of

words]; Kinsbourne & Warrington, 1962 [who reported that the text scanning of two

participants with word-based neglect dyslexia was "entirely normal"]; Patterson & Wilson,

1990). Such a dissociation, with neglect dyslexia at the word level but not at the text level,

was also reported for individuals with developmental neglect dyslexia (the child reported by

Friedmann & Nachman-Katz, 2004; and the 21 children and adolescents reported by

2 Their few between-word migrations occurred almost only on the left of the words (one patient had one migration between the words that did not occur on the left of the word). Thus, these migrations are more likely to be related to their word-based neglect dyslexia, rather than to attentional dyslexia. (Indeed some studies showed a tendency to fill out the position of the letter in the neglected side of the word with a letter from a neighboring word, Gvion, Friedmann, & Faran, 2003). 3 CS, a patient reported by McIntosh, Rossetti, and Milner (2002), had text-based neglect and made only few errors of misreading at the word level in poem reading, but her single word reading was not assessed, and it is possible that in text reading, word-based neglect will be less pronounced due to the contribution of the sentence context.


Nachman-Katz & Friedmann, 2007, 2010). These individuals made letter substitution,

omission, and addition errors on the left side of words, but never omitted words on the left

side of the sentence or text.

This double dissociation suggests that different mechanisms are responsible for neglect

dyslexia at the word and sentence level, in line with the claims of Ellis et al. (1987), Subbiah

and Caramazza (2000), and Young et al. (1991). Furthermore, Vallar et al. (2010) also refer to

a double dissociation between neglect at the word level and general visuospatial neglect,

which indicate separate mechanisms underlying the two. They summarize that (word-based)

neglect dyslexia may occur independent of left visuospatial neglect, and that, on the other side

of the dissociation, right-brain-damaged patients with left visuospatial neglect may not show

left neglect dyslexia (at the word level) (see also Cubelli et al., 1991), the latter is the pattern

we see for the two participants in the current study.

Please insert Table 2 here

3. General Material and Procedure

3.1. General material

To assess the effect of syntax on sentence reading in text-based neglect dyslexia, we created a

list of 310 sentences that were designed for 6 comparisons.4 Each of the tests allowed for a

comparison between two syntactic structures that differed in the obligatoriness of their final

component. Each contrast compared minimally-different sentences, one sentence included an

obligatory final constituent, the other included an optional final constituent. The sentences of

the various conditions were presented together, randomly ordered, in the same list. They were

randomized in a way that there was a distance of at least 15 sentences between each sentence

and its minimal pair sentence, and no more than 2 sentences of the same type appeared

consecutively. We will report below each condition separately, for clarity of presentation.

The sentences and word sequences were presented in Arial font size 14. All the sentences

were 3-5 words long. The sentences were presented as a list, in the middle of an A4 page,

aligned to the right (recall that Hebrew is read from right to left, so sentences were aligned to

their beginning); There were 34 or 35 sentences per page.

4 The sentence list actually included 390 sentences, but we do not report two conditions that turned out to be

problematic (one condition was supposed to compare SVO to topicalized OSV and OVS sentences, but the accusative marker that opens every object-topicalization sentence is homographic with the feminine 2nd person pronoun, which led to confusions in reading. Another condition compared declaratives and questions, but many of the questions turned out to allow for the omission of the final complement).


The conditions included the following comparisons (the examples below are adapted as much

as possible for English, the detailed description of each condition in the experiments below

will include examples for the Hebrew original sentences).

a) 40 sentences with verb complements vs. 40 sentences with adjuncts

Dave relied on the computer – Dave slept on the beach

b) 30 sentences with an obligatory pronoun vs. 30 relative clauses with an optional pronoun

The boy thought that Mary likes him – I met the boy that Mary likes him

c) 30 sentences with a verb in past tense that requires a complement vs. 30 sentences with the

same verb in the present (participle) tense, which can be interpreted as an adjective

The-teacher was frightening the kids – The teacher frightened the kids

d) 10 Wh questions that require a verb vs. 10 Wh questions that are grammatical without a verb

When (are) the-students studying? – Where (are) the-students studying?

e) 30 sentences with an obligatory sentential complement clause vs. 30 coordinated sentences

The-woman said that the baby slept – The-woman read and the baby slept

f) 30 simple declarative SVO sentences of 3-5 words (as a comparison for 30 meaningless

word-sequences in the same length, which were presented in a separate list).

We also compared 14 questions without a question mark with 14 questions with a question

mark, to test the effect of punctuation marks on reading. In addition, the participants read two

texts. These three conditions were not part of the randomized sentence list but were rather

presented separately as blocks.

To test our classification of sentences to those with obligatory and those with optional final

constituent, we collected the judgments of 16 Hebrew-speaking skilled readers without

language impairment aged 23-55 years. We presented to them the 310 written sentences

without the final constituent, and they were asked to judge, for each sentence, whether it was

complete or incomplete as it was. The instruction was: "Mark the sentences that seem

ungrammatical to you (sentences in which you think something is missing). Treat each

sentence as if it is a single sentence, outside of context". Following this procedure, we

excluded or changed classification for 5 sentences that more than half of the judges (9) judged

differently from our original classification. Two sentences that we considered with obligatory

final constituent were classified by the judges as optional and were added to the adjunct

condition, and three sentences were excluded from the lists of sentences (2 sentences with a

present tense verb, one subordinated sentence).


3.2. Procedure

Each page with sentences was placed on a table in front of the participants, so that its middle

will be in line with the center of the reader's body. The participants were requested to read

aloud each sentence as accurately as possible. Before reading the word sequences the

participants were told that the page contains word sequences rather than meaningful

sentences. No time limit was imposed during testing, and no response-contingent feedback

was given by the experimenter, only general encouragement. Participants were told that

whenever they needed a break they could ask for it. Each participant was tested individually

in a quiet room, in several sessions (according to the participants' abilities, with a minimum of

4 sessions per participant).

3.3. Omission analysis

During the testing sessions, every response that differed from the target was written in detail

by the experimenter, and words read correctly were scored with a plus sign. Pauses longer

than 5 seconds that occurred within the sentence were also marked in the transcription. In

addition, all the sessions were audio-recorded and the transcription from the session was

checked and corrected or completed if needed using the recordings.

We analyzed omissions of the relevant final component(s). We classified as omission both a

complete omission of the relevant final component and a stop followed by a long pause of at

least 5 seconds before the relevant final component, following which the reader continued

reading part or all of the remaining part of the sentence.

We excluded from the analysis (from the total number of the read sentences in the relevant

condition) sentences in which the participant stopped reading in a certain point and continued

to read the following sentence as part of the same sentence. We also excluded sentences in

which the reader read incorrectly the critical part of the sentence that determines whether the

final constituent is obligatory or optional, changing the original syntactic structure of the

sentence, and sentences in which the reader stopped reading the sentence before the critical

part of the sentence. In total, 13.3% of the sentences were excluded from the analyses for

these reasons of errors in the sentence area that was not part of our research question. No

significant difference was found between the number of sentences excluded in the obligatory

condition (13%) and in the optional condition (13.7%), T = 10, p = 1.


3.4. Statistical analyses

The comparison between the omissions in each of the two conditions in each experiment was

done both at the participant level and at the group level. The comparisons between the

performance on two conditions for each participant were done using chi squared test, the

comparisons on the group level were done using Wilcoxon Signed Ranks test (reported with a

capital T).

4. Experiment 1: Adjuncts vs. complements

The natural place to start comparing obligatory and optional sentence-final components is

with complements and adjuncts. Whereas complements obligatorily appear with the verb, and

are part of the lexical-syntactic information in the lexical entry of the verb, adjuncts are

optional constituents, and are not required by the verb. Thus, the phrase "for the winter" in

"Anne longed for the winter" functions as a complement of the verb "longed", and is therefore

obligatory, and "Anne longed" is ungrammatical, whereas "for the winter" in "Anne moved to

Sydney for the winter" is an adjunct, and hence optional and can be omitted. Similarly, "in the

blooming field" in "The farmer believed in the blooming field" is obligatory and the sentence

is ungrammatical without it, but it is optional and can be omitted in "The farmer sneezed in

the blooming field".

4.1. Material

We used this property of complements and adjuncts and created pairs of sentences with the

same structure (example 1), in which the final phrase in one sentence is obligatory (1a), and in

the other sentence- optional (1b). The test included 40 sentences with a complement and 40

with an adjunct. Each minimal pair had the same preposition after the verb. Twenty of the

sentences of each type included a preposition that was bound to the noun that followed it

(namely, was part of the same orthographic word, so the final constituent was a single word)

and 10 were free prepositions. We compared the number of omissions of the optional and

obligatory prepositional phrases. (In the examples the relevant final constituent is underlined,

these constituents were not underlined in the sentences presented to the participants).

(1) a. Ha-ganan hishva bein ha-praxim

The gardener compared between the flowers

b. Ha-ganan yashan bein ha-praxim

The gardener slept between the flowers


4.2. Results

As shown in Table 3, constituents that served as adjuncts, and hence, as optional components,

were omitted significantly more often than complements of verbs, T = 0, p = .02. Namely, the

participants omitted more final constituents when they could be omitted without violating the

grammaticality of the sentence, than when the omission of the constituent would have created

an ungrammatical sentence, because of the violation of the predicate argument structure

requirements of the verb. Each of the participants omitted more adjuncts than arguments, for

two of them this difference was significant at the participant level.

Table 3

Rates of complement and adjunct omissions.

Participant Complement omission Adjunct omission Statistical comparison

LA 13% 22% χ2

= 0.98, p = .32

BK 25% 58% χ2

= 6.58, p = .01

SA 14% 30% χ2

= 3.33, p = .06

AS 33% 46% χ2

= 1.62, p = .20

NA 7% 10% χ2

= 0.15, p = .69

AM 5% 24% χ2

= 5.81, p = .01

Average 16% 32% T = 0.0, p = .02

5. Experiment 2. Obligatory pronouns vs. optional pronouns

5.1. Material

Hebrew object pronouns can appear in a sentence-final position in different syntactic

contexts. One context is a simple object pronoun, as in (2a), which is the obligatory

complement of the transitive verb. Object pronouns can also appear as optional sentence

components, when they serve as resumptive pronouns, i.e., in the object position in object

relative clauses, as exemplified in (2b). In Hebrew, object relative clauses can appear either

with a gap after the verb (namely, with no element written after the verb), or with a

resumptive pronoun after the verb. Thus, Hebrew resumptive pronouns in object relatives are

optional (Friedmann & Costa, 2011; Shlonsky, 1997). Put differently, transitive verbs

obligatorily require an object. In most sentences, it is therefore impossible to omit a pronoun

that is the object of a transitive verb. However, in object relatives, in which the object moves


to a different position in the sentence, it is possible not to pronounce anything after the verb, it

is only possible to add a resumptive pronoun in this position, which is not obligatory.

These two functions of the Hebrew object pronoun allowed us to conduct a very close

comparison of the omission of exactly the same lexical item in similar sentences with

different syntactic structures and hence, different syntactic requirements for the obligatoriness

of the pronoun. The test included 30 sentences with an obligatory pronoun and 30 sentences

with an optional resumptive pronoun. We compared the number of omissions of the pronoun

in the two conditions.

(2) a. ha-mora yad'aa she-ha-yalda ma'arica ota

The teacher knew that the girl adores her

b. hikarti et ha-mora she-ha-yalda ma'arica ota

I knew the teacher that the girl adores her

5.2. Results

The results, summarized in Table 4, indicated that the omission of a pronoun that appears at

the end (left) of the sentence crucially depends on its syntactic role. Pronouns that served as

the optional resumptive pronoun in object relative clauses were omitted significantly more

often than obligatory pronouns that served as the complement of a transitive verb in sentences

without a relative clause, T = 0, p = .008. This tendency held also for each individual

participant, and was statistically significant for three of them.

Interestingly, this condition is entirely syntactic – the conditions on the obligatoriness or

optionality of the pronouns in the different sentence structures relate to syntactic principles,

and are not mediated by meaning. Therefore, the very clear difference between the omission

rates in the two conditions indicates the crucial role syntax plays in modulating reading to the

end of the sentence in text-based neglect dyslexia.


Table 4

Rates of omissions of obligatory and optional object pronoun.

Participant Obligatory pronoun Optional pronoun Statistical comparison

LA 50% 79% χ2

= 3.05, p = .08

BK 25% 45% χ2

= 0.83, p = .36

SA 52% 82% χ2

= 4.63, p = .03

AS 14% 73% χ2

= 21.21, p < .001

DS 3% 20% χ2

= 4.04, p = .04

NA 13% 23% χ2

= 1.00, p = .31

AM 21% 31% χ2

= 0.61, p = .43

Average 26% 50% T = 0.0, p = .008

6. Experiment 3. Past tense and participial verbs

6.1. Material

Another way to manipulate optionality of a component is by contrasting adjectives and verbs.

For example, whereas the adjective lovely does not take a complement, the verb charm does.

We used this basic difference between adjectives and verbs to create minimal pairs between

sentences in which the verb can be interpreted as an adjective, and hence, make the

complement unnecessary, and sentences in which the verb can only be a verb, and hence its

complement was obligatory.

For this we used the present tense in Hebrew, which, similarly to the participle in English, is

often used as an adjective. Thus, for example, "is charming" can be both a verb (in "The

dancer is charming the audience") and an adjective (as in "The dancer is very charming").

We compared, as shown in example (3), sentences in which the main (transitive) verb was in

the past tense, and hence could not be interpreted as an adjective, and required its complement

(3a), with the same sentences, but in which the verb appeared in the present tense and hence

could be interpreted as either an adjective, not requiring the complement, or as a transitive

verb (3b).


(3) a. ha-saxkan hiksim et ha-kahal

The actor charmed [accusative case marker] the-audience

b. ha-saxkan maksim et ha-kahal

The actor is-charming [accusative case marker] the-audience

Thirty such sentences with a past-tense verb were compared with the same 30 sentences with

a present-tense verb. We compared the omission rates of the object of the present and past

tense verbs.

6.2. Results

Significantly more omissions occurred after the participial verb than after the same verb in the

past tense, T = 0, p = .02, as shown in Table 5. Each of the participants made more omissions

in the present tense sentences than in the past tense sentences, for one of them this difference

was statistically significant.

Table 5

Rates of omissions of the final noun phrase when the main verb is in the past or in present tense.

Participant Past Present/participle Statistical comparison

LA 11% 24% χ2

= 1.21, p = .26

BK 45% 48% χ2

= 0.02, p = .87

SA 4% 8% χ2

= 0.44, p = .50

AS 18% 40% χ2

= 3.42, p = .06

NA 0% 7% χ2

= 1.79, p = .18

AM 4% 25% χ2

= 5.01, p = .02

Average 14% 25% T = 0.0, p = .016

Notice that the sentences in the sentence pairs in this case were identical except for a single

letter, the inflection letter which marks the past or present tense of the verb (for example,

charming in the present in Hebrew is מקסים, and charmed in the past is )הקסים . Nevertheless,

this single letter difference lead to almost twice as many omissions in the reading of the

sentences with a present tense verb compared with the sentences with the past tense verbs.


7. Experiment 4. Questions with optional vs. obligatory complement

7.1. Material

In Hebrew, not all sentences include a verb (Benmamoun, 2008; Hazout, 2010). There are

also nominal sentence like "ha-yam yafe (the-sea beautiful)", meaning the sea is beautiful, or

"Yoni ba-bayit (Yoni at-home)", meaning that Yoni is at home. Similarly, Wh questions do

not have to include a verb, and a question like "Eifo Yoni? (Where Yoni)" are frequent and

grammatically correct in Hebrew. We used this property of Hebrew to compare two types of

questions that include a Wh phrase, a subject and a verb. One type is grammatical without the

verb, creating a nominal question like "Where Yoni?" (see example 4b), the other type is

ungrammatical without the verb (4a). Notice that unlike the other minimal pairs in this study,

the difference between optional and obligatory final component here related to the semantic

properties of the Wh phrase, rather than to purely syntactic or lexical-syntactic properties. The

test contained 10 questions that require the verb and 10 matched questions in which the Wh

phrase allows for an omission of the verb. We compared the rate of omissions of the final

verb in the two conditions.

(4) a. Matai ha-yeladim sixaku?

When the-children played

b. Eifo ha-yeladim sixaku?

Where the-children played

7.2. Results

Similarly to the previous experiments, there were significantly more omissions when the

question did not require the final constituent, the verb, than when it did,

T = 0, p = .02. The same tendency held for each of the participants, and was significant for

two of them, as shown in Table 6.


Table 6

Rates of omissions of the final verb when the question allowed and did not allow for a

nominal sentence.

Statistical comparison Optional question Obligatory question Participant

χ2 = 1.87, p = .17 50% 0% LA

χ2 = 2.01, p = .15 100% 78% BK

χ2 = 7.21, p = .007 91% 33% SA

χ2 = 0.89, p = .34 82% 63% AS

χ2 = 3.52, p = .06 30% 0% NA

χ2 = 4.07, p = .04 64% 20% AM

T = 0.0, p = .02 69% 32% Average

LA read only 5 questions

8. Experiment 5. Embedded vs. coordinated clauses

8.1. Material

Another comparison that we made was between subordinated and coordinated sentences. We

surmised that given that complement clauses of verbs are obligatory (like the clause that Dani

is wonderful in the sentence Marko thinks that Dani is wonderful), whereas coordinated

clauses are completely optional, there will be more omissions in coordinated, compared with

subordinated complement clauses. We compared 30 subordinated and 30 coordinated

sentences. All the sentences included 4 words: a noun phrase, a verb, another noun phrase,

and an intransitive verb. Between the two clauses, each including a noun phrase and a verb,

there was either a coordination marker "and" (ve-), in the coordinated sentences, or a

subordination marker "that" (she-), in the subordinated sentences. Both ve- and she- are single

letter morphemes that are bound to the next word, in this case, to the second noun.

(5) a. Yosi nirdam ve-Michal ne'elva

Yosi fell-asleep and-Michal was-offended

b. Yosi hivxin she-Michal ne'elva

Yosi noticed that-Michal was-offended


Differently from the previous 4 minimal pairs in the previous experiments, in this experiment

we analyzed the rate of omissions of the whole second clause, which included two words.

8.2. Results

At the group level, no significant difference was found between the subordinated and the

coordinated sentences, although at the individual level, two participants made more omissions

on the coordinated sentences, as shown in Table 7. Listening to the way the participants read

the sentences clarified why this comparison did not yield significant differences. Whereas in

all other comparisons the two conditions differed in the final word, in the current condition

the relevant component was the whole second clause, a clause of two words. Thus, once the

participants did read the third word, which was the subject of the second clause, it was no

longer possible for them to omit the final word without affecting the grammaticality of the

sentence. Given that most of their omissions in 4-word sentences tended to occur before the

final word, the syntactic effect which would have occurred only with an omission after the

second word was not demonstrated for most participants.

Table 7

Rates of omissions in subordinated and the coordinated sentences.

Participant Subordinated Coordinated Statistical comparison

LA 0% 0% χ2 = 0, p = 1

BK 52% 67% χ2 = 1.07, p = .30

SA 21% 59% χ2 = 8.19, p = .004

AS 59% 47% χ2 = 0.85, p = .36

NA 0% 14% χ2 = 4.29, p = .03

AM 7% 3% χ2 = 0.38, p = .53

Average 23% 32% T = 3, p = .16


8.3. Interim summary – syntax modulates omissions in text-based neglect dyslexia

The results of the five experiments, and especially of Experiments 1-4, indicate that the

syntactic structure of the sentence affects the tendency to omit words on the left side, which in

Hebrew is the end of the sentence. Much fewer word omissions occurred in sentences in

which the final word was required by syntax, and in which the omission of the final word

would have resulted in ungrammaticality, than in sentences that were grammatical even

without the final word.

We ran a test across the ten conditions (the five experiments, each containing two conditions,

which appeared together in the randomized list of sentences). This analysis was guided by our

preplanned comparison, of the obligatory conditions against the optional conditions. We thus

compared the omissions in the 5 sentence types that end with an obligatory component (which

included the conditions: complement, obligatory pronoun, past tense verb, obligatory

question, and subordination) with the omissions in the 5 sentence types that end with an

optional component (which included the conditions: adjunct, optional pronoun, participle

verb, optional question, and coordination). This test, which was done using the Wilcoxon

Signed Ranks test, showed that across the whole list, optional components were omitted

significantly more often than obligatory components, T = 0, p = .004. Figure 1 summarizes

the rates of omissions in all five pairs of optional and obligatory final component.

Figure 1 .Omission rate of optional (dark textured bars) and obligatory (light bars)

components in the five experiments.


A separate analysis of long pauses (longer than 5 seconds) before the optional and obligatory

elements and of complete omissions of the optional and obligatory elements indicated that in

the five conditions, each type of response yielded significantly more errors of the optional

element: there were significantly more long pauses before the optional element (192 long

pauses, 24% of the analyzed sentences) than before the obligatory element (115 long pauses,

14% of the sentences), T = 0, p = .008; and there were significantly more complete omissions

of the optional element (77, 11% of the sentences) than complete omissions of the obligatory

one (32, 4% of the sentences), T = 0, p = .02.

It seems to be the case that the allocation of attention to the left side of the sentence is affected

by whether or not the chunk of the sentence that has already been read is grammatical or not.

If the part of the sentence that has been read is not yet grammatical by itself, the attentional

mechanism continues to shift attention to the next word. Namely, syntax encourages attention

shifting to the left as long as the sentence is not grammatical.

The same individuals had different rates of omissions in the various minimal pairs. To

compare between conditions, we normalized the variability in error rates between participants

by calculating per each participant, per each test, the number of errors in each structure out of

the total number of errors this participant made in this test. Figure 2 presents the group

average of these results for each test. The analysis presented in Figure 2 indicates that, once

the total error rate across conditions is controlled for, the optionality of the leftmost

constituents had similar effects across all conditions – in all conditions optional constituents

were omitted more often than obligatory ones.

Figure 2 .Rate of omissions in sentences ending with optional (dark, textured) and obligatory (light) components out the total number of omissions in each test .The error bars represent standard errors


Figures 1 and 2, together with the individual data in Tables 3-7 raise an interesting point

regarding the hierarchy of difficulty of the various sentence structures. The optional pronoun

condition and the optional-verb questions yielded the largest omission rates for the group and

for each of the individual participants (BK also showed elevated omission rates in the adjunct

and coordinated conditions). This finding has important clinical and diagnostic implications.

As can be seen in Tables 3-7, the different participants had different degrees of severity of

text-based neglect dyslexia, manifested in varying rates of omissions in reading. For the more

severely impaired patients, the text-based neglect dyslexia reveals itself in almost any type of

sentence. However, if we want to detect neglect dyslexia in the more mildly impaired

patients, the sentences that are more sensitive to omissions should be used. Put in a different

way, using unsuitable sentences in the diagnosis, sentences with insensitive structures

(namely, with a left-hand component that cannot be easily omitted), may lead to missing a

patient's neglect dyslexia.

9. Experiment 6. Sentences vs. word sequences

9.1. Material

If indeed the content of the written material affects reading in neglect, we would expect that

reading of meaningful word strings would differ from the reading of meaningless word strings

of the same length. To explore this, we compared sentences with word sequences. Thirty

declarative simple sentences, including ten 3-word sentences, ten 4-word sentences, and ten

5-word sentences were compared with 30 word sequences, ten 3-word, ten 4-word, and ten 5-

word sequences. The word sequences (see example 6) included masculine and feminine

nouns, in singular and plural forms.

(6) rexov praxim mita cipor

street flowers bed bird

Unlike in the analyses of the sentence reading in Experiments 1-5, where we analyzed the

omission of the final component and hence only looked at omissions at the relevant position

in the sentence, in the analysis of sentences versus sequences we counted all omissions

between the words in both the sentences and the sequences, because we did not have an a

priori assumption as to the position that was supposed to show the difference between the

conditions.


9.2. Results

The results indicate a very clear difference between the reading of sentences and of noun

sequences. The first analysis, presented in Table 8, compared the number of sentences in

which at least one word was omitted with the number of sequences in which at least one was

omitted. This tendency was significant for 5 of the participants, and significant for the group.

Table 8

Rates of items in which omissions occurred in meaningful sentences and noun sequences.

Statistical comparison Sequences Sentences Participant

χ2 = 3.27, p = .07 23% 7% LA

χ2 = 4.8, p = .03 80% 53% BK

χ2 = 18.47, p < .0001 97% 47% SA

χ2 = 0, p = 1 93% 93% AS

χ2 = 6.67, p = .01 33% 7% DS

χ2 = 26.79, p < .0001 80% 13% NA

χ2 = 18.37, p < .0001 90% 37% AM

T = 0, p = .02 71% 37% Average

The second analysis, presented in Table 9, compared the number of omissions out of the

number of words. This analysis was done for every word in the sentence or sequence. For

example, when there was a 5 seconds pause before the second word and then an omission of

the third word in a 3-word sentence, this counted as 2 omissions out of 3 words. This analysis

too indicated a clear difference between the way individuals with neglect dyslexia read

sentences and word sequences. They made significantly more omissions of words in

meaningless noun-sequences than in sentences. This was also significant for each of the

individual participants.


Table 9

Rates of word omissions in meaningful sentences and word sequences.

Statistical comparison Sequences Sentences Participant

χ2

= 4.60, p = .03 8% 2% LA

χ2

= 17.55, p < .001 43% 17% BK

χ2

= 45.61, p < .001 57% 14% SA

χ2

= 13.85, p < .001 56% 32% AS

χ2

= 5.42, p = .02 9% 2% DS

χ2

= 22.92, p < .001 25% 3% NA

χ2

= 38.76, p < .001 52% 13% AM

T = 0.0, p = .008 36% 12% Average

10. Experiment 7. Do question marks make a difference? Do question marks make a difference

10.1. Material

To assess whether punctuation marks also serve as an anchor on the left that encourages

reading to the end of the sentence, we compared sentences with and without question marks.

These sentences were adjunct questions, all the questions in this test were questions that

required the verb. There were 14 sentences in this comparison, appearing once with and once

without a question mark, with a total of 28 sentences.

Example (7) presents the way the sentence "When did the tired soldiers fall-asleep?"

(in Hebrew: when the-soldiers the-tired fell-asleep?), with and without a question mark.

מתי החיילים העייפים נרדמו ) 7(

?מתי החיילים העייפים נרדמו

10.2. Results

The results, presented in Table 10, indicate that question marks on the left of the sentence did

not facilitate reading and did not yield fewer omissions of words on the left. The rate of final

word omissions was similar in the questions with and without a question mark, and there was

no significant difference between the conditions either at the group level or at the individual

level (in fact, there was a tendency at the group level to omit more final words when the

sentence ended with a question mark, but this was not significant).


Table 10

Rates of omissions of the final component in questions with and without a question mark.

Statistical

comparison

No

question mark

With

a question mark

Participant

χ2

= 1.94, p = .16 10% 38% LA

χ2

= 2.24, p = .13 50% 80% BK

χ2

= 2.19, p = .13 7% 29% SA

χ2

= 0.14, p = .70 57% 50% AS

χ2

= 1.03, p = .30 7% 0% DS

χ2

= 0.00, p = .95 8% 7% NA

χ2

= 0.34, p = .55 14% 23% AM

T = 6, p = .1 22% 32% Average

Thus, whereas non-orthographic symbols like number signs (#), asterisks (*), and diagonal

lines at the left of the stimulus do lead to better reading in neglect dyslexia, at least at the

word level (Friedmann & Nachman-Katz, 2004; Nachman-Katz & Friedmann, 2010; Riddoch

et al., 1990; Worthington, 1996; but see Patterson & Wilson, 1990), question marks do not

yield a similar effect. This might suggest that once a symbol is considered part of the

orthographic system, it is neglected with the rest of the orthographic symbols, and hence, it

cannot facilitate reading.

11. Experiment 8. The effect of sentence structure and meaning on neglect point in text reading

Until now we evaluated the effect of syntax on reading at the sentence level. We now proceed

to assess whether the well-formedness of a sentence also affects reading and omissions in

text.

11.1. Material and analysis

Two short texts, one of 16 lines and one of and 11 lines, were administered for reading aloud.

One text was taken from a newspaper article, the other – from a short fiction story. In these

texts we counted how many of the stops (the position in the line where the participant stopped

reading) created a "happy end". We defined happy end as either a stop that created a good

ending point for the sentence – namely, created a grammatically and semantically well-


formed sentence, or a stop that occurred at a point for which the beginning of the next line

formed a good continuation, namely, that the sentence until the stop point, together with the

words in the next line, created a grammatically and semantically well-formed sentence. For

each stop position we first looked for a good ending, and if this position did not create a good

ending, we looked whether a good continuation was possible. Happy ends were judged by 3

independent judges.

To get an estimate for how many happy ends would have been created had each participant

stopped one word earlier than s/he actually did, we counted good endings and good

continuations for a hypothetical stop point a word before the actual stop for each participant

in each line. The number of happy ends in the actual stop positions were then compared to the

number of happy ends that would have been created had the participants stopped a word

earlier. If indeed the participant continued to shift attention to the left, to the next word,

because of a syntactic need – because the previous word did not create a well-formed

sentence, then we should see fewer happy ends in the word before than in the actual stop

position.

11.2. Results

A first analysis of the data counted the number of words omitted per line and the number of

lines in which an omission occurred for each participant in each text. These results are shown

in Table 11. LA, BK, and SA had a more severe text-based neglect dyslexia than did DS and

NA, who seemed to have a milder neglect dyslexia. Notice, that the neglect dyslexia of these

two milder patients was better identified in the previous experiments of sentence and

sequence reading. For example, DS, who made the fewest omissions in text (only 2 of 27

lines, 7%), made 25% omissions in sentences with a resumptive pronoun, and 33% omissions

in the noun sequences. This indicates that text reading may not be the best test for detecting

mild text-based neglect dyslexia.

The comparison of happy ends in the actual stop positions and in the word before the stop

position indicated, for both Text 1 and Text 2, that the positions in which the participants

stopped created significantly more happy ends than there would be if they had stopped a word

earlier. In the first text, 80 of the actual stop positions, out of the 94 lines the whole group

read (without DS who made no omissions in this text), created happy ends, compared to only

49 happy ends that would have been created had they stopped a word earlier, which created a

significant difference at the group level, and for 3 of the participants. Similarly, in the second


text, 54 of the stops in the 66 lines created a happy end, compared with only 39 happy ends

that would have been created had they stopped a word earlier, again, a significant difference

at the group level and for two of the participants (see Table 12 and Figure 3).

Table 11

The number of lines in which the left words were omitted and the number of omitted words

per participant.

Participant Number of lines that were

not read to their end

Average number of omitted

words in lines with omissions

Text 1 (16 lines) LA 11 5.3 BK 15 6.9 SA 12 4.6 AS 6 3.7 DS 0 0 NA 3 1.7 OM 4 1.5 Text 2 (11 lines)

BK 10 6.1 SA 9 4.4 AS 6 2.5 DS 2 2.5 NA 2 1.5 OM 9 2

2 קריאה קטע

לביקורות "לציון ראשון" הישראלית הסימפונית התזמורת זוכה לאחרונה

את המשווים ויש גבוהה רמה בעלת לתזמורת נחשבת היא .במיוחד אוהדות

.הישראלית הפילהרמונית של לאלו איכויותיה

.לציון בראשון התרבות היכל הוא האחרונות בשנים הסימפונית של ביתה

ומעולים ותיקים מישראלים מורכבת והיא שנים 13 לפני הוקמה הסימפונית

למבוגרים צעירים בין החלוקה כאשר ,נגנים 90 - כ מונים הם כשביחד חדשים

של הבית מנצח .21 בן הוא ביותר הצעיר הנגן .שווה יותר או פחות היא

התחיל אטינגר .אטינגר דני הוא המחליף והמנצח רודן מנדי הוא הסימפונית

אורח ומנצח הישראלית באופרה הבית למנצח היה יותר ומאוחר אופרה כזמר

בית תזמורת של הבית מנצח יהיה הוא הבאה מהשנה .ירושלים בתזמורת

.בירנבוים דניאל של והעוזר בברלין האופרה

Figure 3. An example of a participant's reading of Text 2. The words to the left of the line were omitted.


Table 12

Rates of stop positions that form a happy end compared with a baseline of the previous word.

Happy end in

actual stop position

Happy end

one word earlier

Text1 (16 lines) LA 10 7 χ

2 = 1.35, p = .25

BK 10 9 χ2 = 0.13, p = .72

SA 13 10 χ2 = 1.39, p = .24

AS 16 8 χ2 = 10.67, p = .001

NA 16 7 χ2 = 12.52, p < .001

OM 15 8 χ2 = 7.57, p = .006

TOTAL 80 49 T = 0, p = .02

Text2 (11 lines)

BK 6 4 χ2 = 0.73, p = .39

SA 9 4 χ2 = 4.7, p = .03

AS 10 8 χ2 = 1.22, p = .27

DS 10 7 χ2 = 2.33, p = .13

NA 11 7 χ2 = 4.89, p = .03

OM 8 9 χ2 = 0.26, p = .61

TOTAL 54 39 T = 1, p = .03

LA read only 14 of the 16 lines in Text 1 and did not read Text 2. DS made no omission on Text 1.

These results indicate that not only sentence reading, but also text reading, is guided by

syntactic and semantic considerations. Patients with neglect tend to continue shifting their

attention to the left until a position is reached in which they can stop shifting their attention to

the left, on the basis of the sentence structure and possibly also its meaning. This is why they

stopped where they did in reading the texts, rather than one word earlier.

11.3. Do punctuation marks modulate omissions?

Another question we tested in text reading was whether the omissions occur in punctuation

marks or whether they are triggered by syntactic considerations. For this aim, we calculated

the number of stops that occurred in punctuation marks (comma, full stop, colon), out of the

number of lines with a punctuation mark that were not read fully.

As summarized in Table 13, in fact most of the stops did not occur in a punctuation mark.


This finding joins our finding that a question mark at the end of a question did not reduce

omissions of the final words, in suggesting that it is the syntactic structure of the sentence,

rather than the visual clue of the punctuation marks, that modulates stop positions.

Table 13

Rates of stops at punctuation marks in text

Stops at punctuation

marks

Lines with punctuation which

were not read to the end

LA 4 11

BK 7 22

SA 5 18

AS 5 11

DS 0 1

NA 2 4

OM 3 10

TOTAL 26 77

12. Discussion

In this study we explored the effect of preserved syntactic knowledge on the reading of

sentences and text in neglect dyslexia. We tested whether sentences with different syntactic

structures yield different reading patterns, and whether individuals with acquired text-based

neglect dyslexia continue to shift attention to the neglected hemispace as long as the syntactic

requirements of the sentence have not been met. This tendency is expected to create fewer

omissions of obligatory constituents than of constituents that are not required by syntax. We

tested this hypothesis using various minimal pairs of sentences, comparing sentences in which

the final (leftmost) constituent is obligatory with sentences in which the final constituent is

optional. We also compared meaningful and meaningless word sequences, and analyzed the

stop position in text paragraphs.

The syntactic structure of the sentences was found to be a crucial determinant of omissions in

the reading of the participants with text-based neglect dyslexia. In most of the comparisons

between the structure pairs, participants tended to omit optional final (left) constituents more

often than obligatory constituents, a tendency which was statistically significant at the group

and at the individual participant levels. Adjuncts were omitted significantly more often than


verb complements, and optional resumptive pronouns were omitted significantly more often

than obligatory pronouns. Significantly more omissions occurred after a present participle

verb, which can be taken to be an adjective that does not require a complement, than after the

same verb in the past tense. Significantly more omissions were also found when a written

question did not require a verb than when it did. Another minimal pair, coordinated and

subordinated sentences, yielded difference in the expected direction, which was not

statistically significant.

These findings have theoretical as well as clinical implications. Theoretically, they indicate

that shifting of attention to the left in reading can be modulated not only by external visual

attention attractors (such as blinking lamp or a visual symbol to the left of the stimulus,

Friedmann & Nachman-Katz, 2004; Nachman-Katz & Friedmann, 2010; Riddoch et al., 1990;

Worthington, 1996; but see Patterson & Wilson, 1990). Shifting of attention can also be

motivated by internal, linguistic factors. Specifically, the syntactic structure of the beginning

of the sentence dictates whether or not the final constituent is required for the grammaticality

of the sentence, and readers with text-based neglect dyslexia tend to continue shifting their

attention to the left more often when omission of the final constituent would cause

ungrammaticality. The readers were found to be attentive to various types of syntactic

licensing conditions, including requirements on the well-formedness of pronouns and the

lexical-syntactic demands of verbs (see Haegeman, 1994 and Shapiro, 1997 for a description

of lexical-syntactic information, and see Shetreet, Palti, Friedmann, & Hadar, 2007, for

evidence that lexical-syntactic information is predominantly encoded in the left hemisphere,

which was preserved for the participants in the current study).

The clinical implications of this study relate to the diagnosis of text-based neglect and

possibly also to the hierarchy of sentence types in treatment. Firstly, sentence reading was

found to be a more sensitive diagnostic tool for text-based neglect dyslexia than text reading.

Secondly, the results of this study show that sentences in which the left constituent is optional

form much better stimuli in a test for the diagnosis of text-based neglect dyslexia than

sentences in which the last element is obligatory. In fact, one might miss the neglect dyslexia

of a participant if only obligatory constituents occur on the left side of the line.

Furthermore, the differences found in the magnitude of the syntactic effects in the various

sentence pairs form an internal hierarchy of obligatoriness of final elements. As a result, even

within the sentences with optional constituents, certain sentence types are more likely than

others to reveal mild text-based neglect dyslexia. Based on these findings, it will be possible


to create a screening test for text-based neglect dyslexia, and in that way, reveal even the

mildest neglect dyslexia. For example, on the basis of the current results the best sentences to

identify mild text-based neglect dyslexia in Hebrew would be sentences (object relatives) that

end with an optional resumptive pronoun, and questions that end with a verb, which can also

be grammatical without the verb. Clearly, for languages read from left to right, a study should

be run to identify the most sensitive initial optional constituents which would be sensitive to

neglect dyslexia (examples that come to mind are sentential adverbs like "yesterday" or

"evidently" in English, and subject pronouns in pro drop languages such as European

Portuguese, Italian, or Spanish).

Along the same lines, the comparison of meaningless noun-sequences to declarative sentences

yielded a clear difference in reading, with 5 of the participants making omissions in 80-97%

of the sequences. This indicates that word sequences, probably in every language, would be a

very sensitive marker for text-based neglect dyslexia.

These findings regarding the hierarchy of optionality of various final components are also

suggestive for a possible direction for treatment of individuals with text-based neglect

dyslexia. Patients can be guided explicitly to pay attention to the syntactic structure of the

sentence they read, and to the lexical-syntactic and thematic requirements of the verbs in it.

For example, they can think how many arguments the verb takes, and which complements it

requires. Such knowledge is implicit and preserved for them, and as we saw, they use it

implicitly in reading, but the use of this knowledge can be made explicit, and consequently,

more effective. One can view sentence reading in neglect dyslexia as a constant competition

between preserved syntax and impaired attention shifting. A treatment that reinforces the

syntactic considerations may help in putting more weight in favor of attention shift. Treatment

could then put emphasis on the syntactic and thematic well-formedness of the sentences the

patients read. The patients can be taught to check whether all the requirements of the sentence

have been met and whether it has indeed finished. According to the severity of neglect

dyslexia, for the more severely impaired individuals, this treatment can start with the

sentences in which the left constituent is the most obligatory one, in which it will be easiest

for the participants to identify omission, and then proceed to left constituents that are less

obligatory.

Another finding with implications for theory and treatment that emerges from the study is that

orthographic symbols do not serve as attention capturers, and do not promote attention

shifting to the left of the line. The individuals with neglect dyslexia in our study did not omit


fewer final elements in sentences that ended with a question mark than in sentences without

question marks. Moreover, the stop point in text reading did not tend to occur in punctuation

markers such as comma, full stop, colon etc. It is possible that questions marks and other

punctuation marks are considered part of the orthographic system and hence do not facilitate

reading. This is in contrast to non-orthographic symbols, which, when placed left to the target,

attract attention and enhance reading in neglect dyslexia, at least at the word level (Friedmann

& Nachman-Katz, 2004; Nachman-Katz & Friedmann, 2010; Riddoch et al., 1990;

Worthington, 1996; but see Patterson & Wilson, 1990).

The reading pattern in text reading was similar to the pattern found in sentence reading. In the

texts, most participants stopped reading at a specific position and omitted the rest of the line if

the words that were already read created a grammatical sentence. When the sentence was

ungrammatical without its end, the participants tried to find a continuation to the sentence: In

sentence condition, it made them read the sentence to its end. In text reading, it made them

either continue reading the same line or continue to the next line, when the initial words in the

next line served as syntactically and semantically good continuation to the words that were

read in the previous line. In the analysis of text reading, the effect of the syntactic and

semantic requirements could be seen in the comparison between the actual stop positions and

a hypothetical stop at a word before the actual stop position. This comparison indicated that the

actual stops formed significantly more often a happy end, namely, a good ending of the sentence or

a good continuation to the next line, than had the participants stopped a word earlier.

These results at the sentence and text level are consistent with some results from neglect

dyslexia at the word level. In reading single words, as we already mentioned in the

Introduction, many studies reported that various lexical factors affect reading of some

individuals with word-based neglect dyslexia. More similarly to the type of effect we saw in

the current study, there are several properties of the target words that not only affected

accuracy rate, but were specifically found to cause readers with word-based neglect dyslexia

to continue reading the word. One such factor was the form of the letters: in Hebrew, some of

the letters have a different form when they appear at the end (left side) of the word and when

they appear in non-final position. When one of these letters appeared in a word in a non-final

form, namely, in a non-final position, it made the readers continue shifting their attention to

the left of it, toward the end of the word. This is because stopping right after a letter in a non-

final form would result in violation of the orthographic well-formedness of the word, and

thus, Hebrew readers with acquired neglect dyslexia almost never omitted letters after (to the


left of) a non-final form letter (Friedmann & Gvion, 2005). Another factor that was found to

attract continued attention shifting to the left at the word level in Hebrew was the word's

morphology. Because in Hebrew most words are created from a 3-consonant roots,

individuals with word-based neglect dyslexia never omit a root letter, and even if they omit a

very high rate of non-root morphemes on the left, they continue reading the word until they

find the 3 letters of the root (Reznick & Friedmann, 2009). This is in line with findings from

Italian, where at least some individuals with left word-level neglexia read more words

correctly when they included an accent on the right side of the word (Cubelli & Beschin,

2005; Rusconi, Cappa, Scala, & Meneghello, 2004).

Another finding of this study is the dissociation found for two individuals between neglect

dyslexia at the text and word levels. Two individuals, DS and NA, had neglect dyslexia at the

sentence and text level, with no neglect dyslexia at the word level. Some researchers believe

that omissions of words on one side that occur in text reading of neglect patients are a

reflection of their visuospatial neglect, whereas errors to one side of individual words in text

reflect neglect dyslexia (Haywood & Coltheart, 2001). The dissociation we found goes in the

same vein with this observation: the participants had visuospatial neglect and neglect errors in

reading text, possibly as part of their visuospatial neglect, but had no neglect of letters at the

word level. The perception of neglect at the word level as a separate phenomenon is also

consistent with findings of neglect dyslexia at the word level without general neglect and

without omissions of words on the text level (De Lacy Costello & Warrington, 1987;

Friedmann & Nachman-Katz, 2004; Nachman-Katz & Friedmann, 2007; Nachman-Katz &

Friedmann, 2010; Haywood & Coltheart, 2001; Patterson & Wilson, 1990).

Neglect is one of the most multi-faceted disorders (Cubelli et al., 1991). The current study

supports yet another dissociation that can be found in neglect, between reading at the word

and sentence level. This study also provides sensitive markers to reveal text-based neglect

dyslexia – sentence with optional components on the left, and presents another factor that has

strong influence on the probability of errors on the left of the sentence: syntactic structure.

Syntactic structure, this study shows, determines what's left.

References

Arduino, L. S., Burani, C., & Vallar, G. (2002). Lexical effects in left neglect dyslexia: A study in Italian patients. Cognitive Neuropsychology, 19, 421-444.

Arduino, L. S., Burani, C., & Vallar, G. (2003). Reading aloud and lexical decision in neglect dyslexia patients: A dissociation. Neuropsychologia, 41, 877-885.

Arguin, D., & Bub, M. (1997). Lexical constraints on reading accuracy in neglect dyslexia. Cognitive Neuropsychology, 14, 765-800.


Baddeley, A. (1997). Human memory: Theory and practice. East Sussex: Psychology Press. Behrmann, M., Moscovitch, M., Black, S. E., & Mozer, M. (1990). Perceptual and conceptual

mechanisms in neglect dyslexia. Brain, 113, 1163-1183. Benmamoun, E. (2008). Clause structure and the syntax of verbless sentences. In R. Freidin, C. P.

Otero, & M. L. Zubizarreta (Eds.), Foundational issues in linguistic theory. Essays in honor of Jean-Roger Vergnaud. Cambridge, MA: MIT Press.

Caramazza, A., & Hillis, A. E. (1990). Levels of representation, co-ordinate frames, and unilateral neglect. Cognitive Neuropsychology, 7, 391-445.

Cubelli, R., & Beschin, N. (2005). The processing of the right-sided accent mark in left neglect dyslexia. Brain and Language, 95, 319-326.

Cubelli, R., Nichelli, P., Bonito, V., De Tanti, A., & Inzaghi, M. G. (1991). Different patterns of dissociation in unilateral spatial neglect. Brain and Cognition, 15, 139-159.

De Lacy Costello, A., & Warrington, E. K. (1987). The dissociation of visuospatial neglect and neglect dyslexia. Journal of Neurology, Neurosurgery, and Psychiatry, 50, 1110-1116.

Ellis, A. W., Flude, B. M., & Young, A. W. (1987). "Neglect dyslexia" and the early visual processing of letters in words and nonwords. Cognitive Neuropsychology, 4, 439-464.

Ellis, A. W., Young, A. W., & Flude, B. M. (1993). Neglect and visual language. In I. H. Robertson & J. C. Marshall (Eds.), Unilateral neglect: Clinical and experimental studies (pp. 233-255). Hillsdale, NJ: Erlbaum.

Fattal, I., Friedmann, N., & Fattal-Valevski, A. (in press). The crucial role of thiamine in the development of syntax and lexical retrieval: A study of infantile thiamine deficiency. Brain. doi:10.1093/brain/awr068

Friedmann, N. (1998). BAFLA: Friedmann Battery for Agrammatism. Tel Aviv University. (in Hebrew)

Friedmann, N. (2001). Agrammatism and the psychological reality of the syntactic tree. Journal of Psycholinguistic Research, 30, 71-90.

Friedmann, N. (2006). Speech production in Broca's agrammatic aphasia: Syntactic tree pruning. In Y. Grodzinsky & K. Amunts (Eds.), Broca’s region (pp. 63-82). New York: Oxford University Press.

Friedmann, N., & Costa, J. (2011). Last resort and no resort: Resumptive pronouns in Hebrew and Palestinian Arabic hearing impairment. In A. Rouveret (Ed.), Resumptive pronouns at the interfaces (pp. 223-239). Language Faculty and Beyond series, John Benjamins.

Friedmann, N., & Grodzinsky, Y. (1997). Tense and agreement in agrammatic production: Pruning the syntactic tree. Brain and Language, 56, 397-425.

Friedmann, N., & Gvion, A. (2001). Letter position dyslexia. Cognitive Neuropsychology, 18, 673-696. Friedmann, N., & Gvion, A. (2003). TILTAN: A test battery for dyslexias. Tel Aviv University. Friedmann, N., & Gvion, A. (2005). Letter form as a constraint for errors in neglect dyslexia and letter

position dyslexia. Behavioural Neurology, 16, 145-158. Friedmann, N., Kerbel, N., & Shvimer, L. (2010). Developmental attentional dyslexia. Cortex, 46,

1216-1237. Friedmann, N., & Nachman-Katz, I. (2004). Neglect dyslexia in a Hebrew-reading child. Cortex, 40,

301-313. Friedmann, N., & Novogrodsky, R. (2004). The acquisition of relative clause comprehension in

Hebrew: A study of SLI and normal development. Journal of Child Language, 31, 661-681. Friedmann, N., & Novogrodsky, R. (2007). Is the movement deficit in syntactic SLI related to traces

or to thematic role transfer? Brain and Language, 101, 50-63. Friedmann, N., & Novogrodsky, R. (2011). Which questions are most difficult to understand? The

comprehension of Wh questions in three subtypes of SLI. Lingua, 121, 367-382. doi: 10.1016/j.lingua.2010.10.004

Friedmann, N., & Rahamim, E. (2007). Developmental letter position dyslexia. Journal of Neuropsychology, 1, 201-236.

Friedmann, N., & Szterman, R. (2006). Syntactic movement in orally-trained children with hearing loss. Journal of Deaf Studies and Deaf Education, 11, 56-75.

Friedmann, N., & Szterman, R. (2011). The comprehension and production of wh questions in


children with hearing impairment. Journal of Deaf Studies and Deaf Education, 16(2), 212-235. doi: 10.1093/deafed/ENQ052

Gvion, A., Friedmann, N., & Faran, M. (2003, September). The interaction between attentional dyslexia and right neglect dyslexia: Two case studies. Presented at the British Aphasiology Society, Biennial International Conference, Newcastle, UK.

Haegeman, L. (1994). Introduction to Government and Binding theory (2nd ed.). Oxford: Blackwell. Haywood, M., & Coltheart, M. (2001). Neglect dyslexia with a stimulus-centered deficit and without

visuospatial neglect. Cognitive Neuropsychology, 18, 577-615. Hazout, I. (2010). Verbless sentences and clause structure. Linguistic Inquiry, 41(3), 471-485. Karnath, H. O., & Huber, W. (1992). Abnormal eye movement behaviour during text reading in

neglect reading: A case study. Neuropsychologia, 30, 593-598. Kartsounis, L. D., & Warrington, E. K. (1989). Unilateral visual neglect overcome by cues implicit in

stimulus arrays. Journal of Neurology, Neurosurgery, and Psychiatry, 52, 1253-1259. Kinsbourne, M., & Warrington, E. K. (1962). A variety of reading disability associated with right

hemisphere lesions. Journal of Neurology, Neurosurgery, and Psychiatry, 25, 339-344. McIntosh R. D., Rossetti, Y., & Milner, D. A. (2002). Prism adaptation improves chronic visual and

haptic neglect: A single case study. Cortex, 38, 309-320. Nachman-Katz, I., & Friedmann, N. (2007). Developmental neglect dyslexia: Characteristics and

directions for treatment. Language and Brain, 6, 75-90. (In Hebrew) Nachman-Katz, I., & Friedmann, N. (2010). An empirical evaluation of treatment directions for

developmental neglect dyslexia. Procedia Social and Behavioral Sciences, 6, 248-249. Novogrodsky, R., & Friedmann, N. (2006). The production of relative clauses in SLI: A window to the

nature of the impairment. Advances in Speech-Language Pathology, 8, 364-375. Patterson, K., & Wilson, B. (1990). A rose is a rose or a nose: A deficit in initial letter identification.

Cognitive Neuropsychology, 7, 447-477. Reznick, J., & Friedmann, N. (2009). Morphological decomposition in the early stages of

orthographic-visual analysis: Evidence from neglexia. Language and Brain, 8, 31-61. (in Hebrew).

Riddoch, M. J., Humphreys, G. W., Cleton, P., & Fery, P. (1990). Interaction of attentional and lexical processes in neglect dyslexia. Cognitive Neuropsychology, 7, 479-517.

Rusconi, M. L., Cappa, S. F., Scala, M., & Meneghello, F. (2004). A lexical stress effect in neglect dyslexia. Neuropsychology, 18, 135-140.

Schwartz, T. H., Ojemann, G. A., & Dodrill, C. B. (1997). Reading errors following right hemisphere injection of sodium amobarbital. Brain and Language, 58, 70-91.

Shapiro L. P. (1997). Tutorial: An introduction to syntax. Journal of Speech, Language, and Hearing Research, 40, 254-272.

Shetreet, E., Palti, D., Friedmann, N., & Hadar, U. (2007). Cortical representation of verb processing in sentence comprehension: Number of complements, subcategorization and thematic frames. Cerebral Cortex, 17, 1958-1969.

Shlonsky, U. (1997). Clause structure and word order in Hebrew and Arabic. New York: Oxford Press.

Stenneken, P., van Eimeren, L., Keller, I., Jacobs, A. M., & Kerkhoff, G. (2008). Task-dependent modulation of neglect dyslexia? Novel evidence from the viewing position effect. Brain Research, 1189, 166-178.

Subbiah, I., & Caramazza, A. (2000). Stimulus-centered neglect in reading and object recognition. Neurocase, 6, 13-31.

Vallar, G., Burani, C., & Arduino, L. S. (2010). Neglect dyslexia: A review of the neuropsychological literature. Experimental Brain Research, 206, 219-235.

Wilson, B., Cockburn, J., & Halligan P. (1987). Development of a behavioral test of visuospatial neglect. Archives of Physical Medicine and Rehabilitation, 68, 98-102.

Worthington, A. D. (1996). Cueing strategies in neglect dyslexia. Neuropsychological Rehabilitation, 6, 1-17.

Young, A. W., Newcombe, F., & Ellis, A. W. (1991). Different impairments contribute to neglect dyslexia. Cognitive Neuropsychology, 8, 177-191.


Table 2

Distribution of errors in the TILTAN word reading test.

Middle

letter

Migrationc

Exterior

letter

Migration

Surface

errors

semantic

errors

morphological

errors

Right

neglect

Left

neglect

Letter

substitutions

(middle)

Migration

between

words

Total

errors

LA a 3 0 0 0 0 0 17 0 3 23 (16%)

BK b 0 0 3 0 2 2 18 13 7 45 (24%)

SAa,b 7 0 2 0 4 0 12 5 0 30 (23%)

AS 2 0 2 0 0 0 14 0 5 23 (12%)

DS 0 0 2 0 0 0 1 0 3 6 (3%)

NA 0 0 1 0 0 0 1 1 2 5 (3%)

OM 0 0 2 0 0 0 25 0 3 30 (16%) a LA and SA read an earlier version of the TILAN screening, which included 142 single words, and did not include word pairs, so their migrations between words are

reported only from the word list. The rest of the participants read 128 single words and 60 words presented in pairs. b BK and SA had more than 20% errors at the word level in reading the TILTAN test, but they had much fewer errors at the word level in reading sentences and

hence were included in the study. c Letter migration – migrations of letters within words. Surface errors – errors that result from reading via grapheme-to-phoneme conversion. Morphological errors –

keeping the root and changing the affix. Right neglect – letter omission, substitution, addition on the right. Left neglect – letter omission, substitution, addition on the

left.

the effect of syntax on reading in neglect dyslexia

Documents