Individual Differences in the Effect of Orthographic/Phonological Conflict on Rhyme and Spelling Decisions

Suzanne E. Welcome; Amanda C. Alton

doi:10.1371/journal.pone.0119734

Abstract

In typical readers, orthographic knowledge has been shown to influence phonological decisions. In the present study, we used visual rhyme and spelling tasks to investigate the interaction of orthographic and phonological information in adults with varying reading skill. Word pairs that shared both orthography and phonology (e.g., throat/boat), differed in both orthography and phonology (e.g., snow/arm), shared only orthography (e.g., farm/warm), and shared only phonology (e.g., vote/boat) were visually presented to university students who varied in reading ability. For rhyme judgment, participants were slower and less accurate to accept rhyming pairs when words were spelled differently and to reject non-rhyming pairs when words were spelled similarly. Similarly, for spelling judgments, participants were slower and less accurate when indicating that word endings were spelled differently when words rhymed, and slower and less accurate when indicating that words were spelled similarly when words did not rhyme. Crucially, while these effects were clear at the group level, there were large individual differences in the extent to which participants were impacted by conflict. In two separate samples, reading skill was associated with the extent to which orthographic conflict impacted rhyme decisions such that individuals with better nonword reading performance were less impacted by orthographic conflict. Thus, university students with poorer reading skills may differ from their peers either in the reading strategies they use or in the degree to which they automatically access word form information. Understanding these relationships is important for understanding the roles that reading processes play in readers of different skill.

Citation: Welcome SE, Alton AC (2015) Individual Differences in the Effect of Orthographic/Phonological Conflict on Rhyme and Spelling Decisions. PLoS ONE 10(3): e0119734. https://doi.org/10.1371/journal.pone.0119734

Academic Editor: Benjamin Xu, The National Institutes of Health, UNITED STATES

Received: September 19, 2014; Accepted: February 1, 2015; Published: March 9, 2015

Copyright: © 2015 Welcome, Alton. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited

Data Availability: All relevant data are within the paper and its Supporting Information files.

Funding: This research was supported by a University of Missouri Research Board Grant to the first author. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

Introduction

In skilled readers, connections have been forged between a phonological (sound-based) system and an orthographic (visually-based) system. These connections may mean that literacy fundamentally changes the way in which language is processed. A number of studies have demonstrated that in skilled readers, orthographic information influences phonological decisions and recognition of spoken words e.g. [1, 2]. A classic finding is that the speed and accuracy of rhyme judgments can be impacted by word spelling, such that adult participants are faster to recognize that similarly spelled pairs (e.g. tie-pie) rhyme than rhymes that are spelled differently (e.g., rye-tie) [1]. This increase in response latency for pairs in which orthographic and phonological information conflict (incongruent word pairs) over congruent pairs has been demonstrated using both auditory and visual presentation of word pairs [3, 4].

Other paradigms have also been used to demonstrate that orthographic information influences auditory word processing. For example, Ziegler and Ferrand [2] demonstrated that auditory lexical decision times are impacted by consistency, such that inconsistent words, whose rimes can be spelled multiple ways (e.g., leap), were recognized more slowly than consistent words, whose rimes can only be spelled one way (e.g., duck). This effect of consistency extends to semantic and gender categorization tasks, indicating that orthography influences many aspects of spoken word processing [5]. In a priming paradigm, primes that shared both orthography and phonology with a target were more effective primes than those that shared phonology only [6]. During a phoneme monitoring task, adults were faster to respond to a target phoneme when it was spelled typically than when it was spelled atypically [7]. Words with more orthographic neighbors show faster reaction times in a shadowing task [8].

There is mixed evidence regarding interactions between orthography and phonology in individuals with poor phonological decoding skills. Some evidence supports the view that orthographic processing is preserved in those with difficulties in phonological decoding. In individuals with dyslexia, orthographic processing is impacted less severely and persistently than phonological processing [9]. As poor readers mature, their deficit in orthographic processing is reduced while their phonological deficit is increased (relative to normal-reading peers), suggesting that individuals with dyslexia are able to catch up in orthographic processing while falling behind in phonological processing [10]. Proficient and dyslexic adult readers showed a comparable effect of orthography on auditory rhyme judgments [11], suggesting that individuals with adequate reading experience, regardless of reading skill, show similar influence of orthography when the task is metaphonological in nature, requiring explicit identification and analysis of speech sounds.

Other findings suggest that individuals with poor phonological processing skills exhibit reduced access to orthographic information. Children with dyslexia showed less orthographic facilitation than typical reading groups in an auditory rhyme detection task [12]. Children with dyslexia also did not show a typical orthographic neighborhood effect [13], showing little benefit for the processing of words that share orthography with many other words. In an fMRI study, children with reading disabilities failed to show reliable activation of the fusiform cortex during an auditory rhyming task [14]. The fusiform gyrus typically shows activation during visual or orthographic processing, and activation of this region in typically-reading children was taken as evidence for activation of orthographic information during an auditory task.

However, there are some reports that individuals with poor phonological skills show enhanced, rather than reduced, effects of orthography. Some individuals with impaired phonological decoding skills were shown to excel at orthographic tasks [15, 16]. Among adolescents performing a visual rhyming task, individuals with dyslexia showed high error rates when phonological and orthographic information conflicted [17]. Adult literacy students, who read at grade levels 3 to 6, made more errors than reading-level matched children on rhyming trials with different spelling [18], indicating that these participants might rely on spelling information to make rhyme decisions.

While the effect of orthography on rhyme decisions and spoken language processing is clear at a group level, it is less clear whether meaningful variation in this effect exists within the adult population. However, Dich [19] demonstrated a relationship between an individual’s spelling skill and the extent to which orthography impacts the recognition of spoken words. Individuals with stronger spelling skills tended to show larger effects of orthography on an auditory lexical decision task (a greater difference between inconsistent and consistent words). A more complicated pattern of results was found when comparing effects of orthography on phonological processing between typical readers and dyslexic adults [11]. During an auditory lexical decision task, consistency effects were related to reading skill, with better readers showing a larger effect of consistency. This relationship with reading skill was not observed for an auditory rhyme judgment task. The authors attributed this pattern of results to the metaphonological nature of the rhyme task. Together, such findings suggest that individual differences in reading and spelling ability may modulate the effect of orthography on phonological tasks.

In the present studies we used visual rhyme and spelling tasks to investigate the interaction of orthographic and phonological information in adults with varying reading skill. Prior studies have demonstrated effects of conflicting orthography on phonological judgments with both auditory and visual presentation of stimuli. Those studies have demonstrated an increase in response latency for orthographically dissimilar rhyme pairs over orthographically similar rhyme pairs using both visual and auditory word presentations [20, 21]. Event-related potential (ERP) data further suggest that such effects are not fully dependent on modality. Consistent differences between non-rhyming and rhyming targets emerge about 300 ms after stimulus onset for both auditory [22] and visual [23] word pairs. Further, ERPs have revealed effects of conflicting orthography during rhyme decisions and spoken language processing with effects of orthography emerging during the same time window as effects of rhyme, using visual presentation [20, 21]. These findings make sense in light of the fact that in order to make a rhyme judgment, the phonological form of the word must be accessed, regardless of the modality of presentation. Nonetheless, differences between modalities exist in terms of the need for print-to-sound conversion and the saliency of the task-irrelevant information.

We were particularly interested in relationships with reading skill. Within the university population, there is considerable variability in reading skill, particularly the ability to read nonwords [24, 25]. Understanding the consequences of this variability in the adult population is important, as readers with varying skill might engage different mechanisms during reading, and a full understanding of the behavioral and neural underpinnings of reading mechanisms must incorporate this variability. Specifically, adults with relatively poor phonological decoding skills may differ from other readers in orthographic processing [26, 27]. On this basis, we predicted that there would be a relationship between phonological decoding skill and the extent to which conflict between orthographic and phonological information impacted an individual such that individuals with worse nonword reading skill would be more reliant on orthographic information, and thus more sensitive to phonological conflict. Most previous studies exploring these questions primarily compared effects in dyslexic and typical-reading children. In this study, we investigated effects across the reading spectrum in adults, in whom reading processes are likely fully developed. For both rhyme and spelling tasks, it is possible for conflict to exist between orthographic and phonological information, and the use of two tasks allows us to investigate the effect of this conflict on both phonological and orthographic decisions.

Experiment 1 Methods

Participants

Forty-three neurologically normal adults (38 female) took part in the study and were compensated with course credit. All were right-handed native speakers of English. The study had the approval of the University of Missouri—St Louis (UMSL) Institutional Review Board and written informed consent was obtained from all participants. Seven participants were removed because their behavioral performance indicated imperfect understanding of the task (accuracy was below 60% on two or more experimental conditions). See S1 Table: Mean (standard deviation) performance across condition for seven participants dropped from Experiment 1. Note that major results are not changed by the inclusion of these participants with low accuracy (S2 Table: Associations (Pearson’s r-values) between effects of conflict and cognitive performance including all participants from Experiment 1). For the remaining 36 participants, scores on standardized measures of reading and nonverbal intelligence ranged from below average to well above average, reflecting the variability present in the UMSL student body (Table 1).

Download:

Table 1. Demographic characteristics of the sample for Experiment 1 (N = 36).

https://doi.org/10.1371/journal.pone.0119734.t001

Psychometric Measures

All participants completed standardized measures of intelligence and reading ability. Nonverbal intelligence was assessed with the Test of Nonverbal Intelligence-3, which uses picture stimuli to evaluate skill at recognizing logical sequences (TONI-3) [28]. The Sight Word Efficiency (SWE) and Phonemic Decoding Efficiency (PDE) subtests of the Test of Word Reading Efficiency (TOWRE) were used to assess timed reading of familiar words and pronounceable pseudowords, respectively [29]. Participants completed the Adult Reading History Questionnaire (ARHQ) [30], designed to investigate participants’ childhood reading experiences and current attitudes. This questionnaire is designed to provide a cut-score (0.30) above which scores indicate a history of reading disability. Using this cut-score, 17 participants would be classified as having a positive history of reading disability. UMSL’s undergraduate student body is somewhat nontraditional, with more than 50% transfer students and a mean age of 26, and it might therefore not be surprising that a relatively large portion of the sample experienced reading difficulty at some point in the past. Despite the relatively high scores on this measure, none of the participants reported repeating grades due to academic failure. Some items in the questionnaire did not seem applicable to the college population (e.g., reading of newspapers given that many of the participants report seeking news through other media), so we did not use a cut-score, and instead treated the score as a continuous measure.

There was a significant association between performance on the SWE and PDE subtests of the TOWRE, r(34) = .48, p < .005. However, nonverbal intelligence and scores on the ARHQ were not significantly correlated with each other, r(34) = −.19, p > .10, or with either reading measure [TONI with SWE: r(34) = −.02, p > .10; TONI with PDE: r(34) = −.04, p > .10; ARHQ with SWE: r(34) = .20, p > .10; ARHQ with PDE: r(34) = −.20, p > .10]. One explanation for the lack of significant associations between self-reported reading history and current reading skill is that the ARHQ contains a mix of questions relating to current reading behavior and past reading experiences. It is possible that participants obtained high scores on the self-report measure primarily as a result of academic difficulty during childhood, and these early experiences are only weakly related to adult performance.

Materials

All participants made rhyme and spelling decisions on visually presented word pairs. In total, participants saw 80 pairs sharing both orthography and phonology (O+P+ pairs), 80 pairs sharing neither orthography nor phonology (O-P- pairs), 40 pairs sharing phonology but not orthography (O-P+ pairs), and 40 pairs sharing orthography but not phonology (O+P- pairs). Three pairs from each of 40 O-P-, O+P+, and O-P+ triplets (seek/boat, throat/boat, vote/boat) and 40 O-P-, O+P+, O+P- triplets (snow/arm, farm/arm, warm/arm) were presented to participants. In this manner, targets were preceded by words that varied in their relationship to the target. Length, concreteness [31], familiarity [31], log frequency [32], number of orthographic neighbors [33], and bigram frequency [33] were equated as much as possible for target words, orthographically related words, and orthographically unrelated words [ts (78) < 1.5, ps > .10] across conditions.

Procedure

Participants performed both a visual rhyme decision task and a visual spelling task, with task order counterbalanced across participants. All stimuli were presented in lowercase, black 18 point Arial font on a white background. Each trial began with a fixation cross that flickered for 250 ms. On each trial, a visual word was presented for 500 ms, followed by an ISI of 500 ms, and presentation of a target word for 500 ms. Trials were separated by a 1,000 ms inter-trial interval. Participants indicated their decision with a button press. For the rhyme task, participants were instructed to indicate their rhyme/no rhyme decision with a button press. For the spelling task, participants were asked to indicate whether the endings of the word pair were spelled similarly with a button press. Participants used their index finger on the “1” key and their middle finger on the “2” key to indicate their response, with response mapping counterbalanced across participants. Before each task, 12 practice trials were administered. Each task was broken into 4 blocks of trials that lasted approximately 3–4 minutes.

Experiment 1 Results

Overall Effects

A 2 (task) x 2 (phonology) x 2 (orthography) ANOVA was conducted separately for accuracy and RT. For accuracy, a main effect of task was observed, F(1, 35) = 4.17, p < .05, with better performance on the rhyme task. The main effects of phonology and orthography were not significant (Fs < 3). The interaction of task and phonology was significant, F(1, 35) = 33.7, p < .001. For the rhyming task, shared phonology led to higher accuracy (90.2% versus 85.9%), while for the spelling task, shared phonology led to less accuracy (81.4% versus 89.1%). The interaction of task and orthography was also significant, F(1, 35) = 26.2, p < .001. For the rhyming task, shared orthography led to less accuracy than non-shared (85.5% versus 90.6%), while for the spelling task, shared orthography led to more accuracy than non-shared (88.0% versus 82.5%). Thus, accuracy was higher for the rhyme task when items shared phonology and higher for the spelling task was items shared orthography. Crucially, phonology and orthography showed a significant interaction, F(1, 35) = 80.92, p < .001. When phonology was shared, shared orthography led to more accuracy (91.3% versus 80.3%). When phonology differed, shared orthography led to less accuracy (82.2% versus 92.8%). The task x phonology x orthography interaction was not significant (F < 1).

For RT, a main effect of task was observed, F(1, 35) = 25.8, p < .001, with faster responses to the spelling task. The main effect of phonology was not significant (F < 1.6), but the main effect of orthography was, F(1, 35) = 7.4, p < .01. The interaction of task and phonology was significant, F(1, 35) = 96.6, p < .001. For the rhyming task, shared phonology led to lower RTs (704 versus 760), while for the spelling task, shared phonology led to longer RTs (688 versus 648). The interaction of task and orthography was also significant, F(1, 35) = 89.3, p < .001. For the rhyming task, shared orthography led to longer RTs than non-shared (761 versus 703), while for the spelling task, shared orthography led to shorter RTs than non-shared (650 versus 685). Thus, performance was better for the rhyme task when items shared phonology and better for the spelling task was items shared orthography. Crucially, phonology and orthography showed a significant interaction, F(1, 35) = 238.3, p < .001. When phonology was shared, shared orthography led to faster RTs than differing orthography (649 versus 744). When phonology differed, shared orthography led to longer RTs (762 versus 646). The task x phonology x orthography interaction was not significant (F < 2.5).

Performance across conditions is presented in Table 2. Planned comparisons using paired-samples t-tests indicated that accuracy was higher for O+P+ trials than O-P+ trials [t(35) = 3.19, p < .005] and higher for O-P- trials than O+P- trials [t(35) = 6.25, p < .001]. RTs also significantly varied with condition [F(3, 77.8) = 102.6, p < .001]. RTs were shorter for O+P+ trials than O-P+ trials [t(35) = 7.23, p < .001], and shorter for O-P- trials than O+P- trials [t(35) = 15.88, p < .001]. Thus, orthographic overlap within a trial led to faster and more accurate rhyme decisions and slower and less accurate no-rhyme decisions.

Download:

Table 2. Mean (standard deviation) performance across condition from Experiment 1. Accuracies are percent correct and RTs are in ms.

https://doi.org/10.1371/journal.pone.0119734.t002

For the spelling task, a repeated measures ANOVA with Greenhouse-Geisser correction showed that accuracy significantly varied with condition (Table 2) [F(3, 53.1) = 31.9, p < .001]. Planned comparisons using paired-samples t-tests indicated that accuracy was higher for O+P+ trials than O+P- trials [t(35) = 5.80, p < .001] and higher for O-P- trials than O-P+ trials [t(35) = 4.25, p < .001]. RTs also significantly varied with condition [F(3, 88.2) = 78.41, p < .001]. RTs were shorter for O+P+ trials than O+P- trials [t(35) = 6.14 p < .001], and shorter for O-P- trials than O-P+ trials [t(35) = 10.79, p < .001]. Thus, phonological overlap within a trial led to better performance when words were spelled similarly worse performance when words were spelled differently.

To quantify the effects of orthographic/phonological conflict for each individual, differences between the average performance on the trials without conflict (O+P+ and O-P-) and trials with conflict (O+P- and O-P+) were computed separately for each task and dependent variable. For this particular study, classifying trials as “non-conflicting” and “conflicting” has the advantage of grouping trials identically for the rhyme and spelling tasks. As shown in Table 3, there were large individual differences in the extent to which individuals showed the effects of conflict. While all group means significantly differed from 0 (Table 3), some individuals showed reversed effects in accuracy and/or small effects in RT, while others showed robust effects of conflict.

Download:

Table 3. Differences indicating effect of orthographic conflict on rhyming decisions and phonological conflict on spelling decisions from Experiment 1.

https://doi.org/10.1371/journal.pone.0119734.t003

Relationships with Reading Skill

Given that individuals differed in the extent to which orthographic conflict impacted rhyme decisions and phonological conflict impacted spelling decisions, we explored relationships between those individual differences and performance on the standardized measures of nonverbal IQ and reading. Significant relationships emerged between effects of orthographic conflict on rhyme decisions and Phonemic Decoding Efficiency scores (PDE; Table 4). Specifically, those individuals with better phonological decoding skills were less impacted by orthographic conflict. Interestingly, this relationship was not apparent with Sight Word Efficiency or nonverbal IQ. Additionally, no significant relationships between standardized test performance and spelling effects emerged, suggesting that reading skill relates specifically to the effect of orthographic overlap on phonological judgments, not to effects of conflict more broadly.

Download:

Table 4. Associations (Pearson’s r-values) between effects of conflict and cognitive performance in Experiment 1.

https://doi.org/10.1371/journal.pone.0119734.t004

Relationships between PDE score and performance on each type of trial for the rhyme task were explored. PDE scores were not significantly correlated with accuracy for non-conflicting trials, r(34) = .25, p > .10, but were correlated with accuracy for conflict trials, r(34) = .38, p < .05 (Fig. 1) Accuracy was high in the non-conflict condition, regardless of PDE score. However, accuracy in the conflict condition was lower in those with lower PDE scores. A similar pattern was seen in RT (Fig. 2), though relationships failed to reach significance [non-conflict: r(34) = .11, p > .10; conflict: r(34) = −.28, p < .10]. Four participants had scaled scores on the standardized measure of reading less than 85. With those participants excluded, the association between PDE and non-conflicting trials approached significance, r(30) = .32, p < .10, while the association between PDE and conflicting trials did not, r(30) = .18, p > .10.

Download:

Fig 1. Accuracy (percent correct) of rhyme decisions made on nonconflicting (O+P+ and O-P-) trials and conflicting (O-P+ and O+P-) trials in Experiment 1.

https://doi.org/10.1371/journal.pone.0119734.g001

Download:

Fig 2. Reaction time for rhyme decisions for nonconflicting (O+P+ and O-P-) trials and conflicting (O-P+ and O+P-) trials in Experiment 1.

https://doi.org/10.1371/journal.pone.0119734.g002

Experiment 1 Discussion

Clear effects of orthography were observed for a rhyme decision task, in keeping with earlier findings that spelling can influence phonological judgments [1, 2, 6]. Accuracy was lower and RTs longer for rhyming pairs spelled differently and non-rhyming pairs that shared spelled. Similarly, conflicts between orthography and phonology impacted a spelling judgment task. Accuracy was higher and RTs shorter for word endings spelled similarly when words rhymed and for word endings spelled differently when words did not rhyme. Individuals showed this effect to different degrees, with some participants showing much larger effects of conflict than others. One predictor of the extent to which orthographic conflict impacted rhyme decisions was phonological decoding ability. Participants with worse nonword reading performance were less accurate in making phonological decisions in the face of orthographic conflict, indicating that these individuals might rely more heavily on orthographic information or experience more difficulty in inhibiting orthographic processing.

Performing the rhyme and spelling tasks within the same experimental session could have influenced results, directing participants’ attention to the spelling of word pairs. In particular, this may have exaggerated the effect of orthographic conflict on rhyme decisions. While task order was counterbalanced, half of the participants in the sample performed the spelling task first, and it is possible that those participants showed particularly large effects, coloring the results of the experiment (S1 Appendix: Results from Experiment 1 participants who performed the rhyme task first). To further address this possibility, a second group of participants was recruited to perform the rhyme and spelling tasks separately. This second group of participants also completed a measure of reading comprehension to examine relationships between orthographic/phonological interaction and comprehension skill.