Morphological knowledge as revealed in children’s spelling accuracy and reports of spelling...

24
Morphological knowledge as revealed in children’s spelling accuracy and reports of spelling strategies Monique Se ´ne ´chal * , Michelle T. Basque, Tina Leclaire Department of Psychology, Carleton University, Ottawa, Ont., Canada K1S 5B6 Received 25 July 2005; revised 8 May 2006 Available online 30 June 2006 Abstract The goal of the current research was to assess whether children can make strategic use of morpho- logical relations among words to spell. French-speaking children in Grade 4 spelled three word types: (a) phonological words that had regular phoneme–grapheme correspondences, (b) morphological words that had silent consonant endings for which a derivative revealed the silent ending, and (c) lex- ical words that had silent consonant endings for which no familiar derivative revealed the ending. Children were also asked to provide immediate retrospective reports of the strategies used to spell each word. Two experiments (Ns = 46 and 39) were conducted. As expected, children in Grade 4 spelled phonological words more accurately than they did words with silent consonant endings. In addition, children spelled morphological words more accurately than they did lexical words. Reports of using retrieval were associated with accurate performance across word types. Importantly, reports of using morphological strategies to spell morphological words were associated with a similar level of accuracy, as were reports of using retrieval. Even though children reported using a phonological strat- egy frequently across all word types, this strategy was associated with accurate performance only for spelling phonological words. Experiment 2 replicated the results of Experiment 1 with another set of stimuli and also showed that children’s morphological awareness predicted their spelling accuracy for morphological words as well as the reported frequency of morphological strategy use. In sum, the findings revealed that most children showed evidence of adaptive strategy use. Ó 2006 Elsevier Inc. All rights reserved. Keywords: Spelling; Morphology; Strategies; Grade 4 0022-0965/$ - see front matter Ó 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.jecp.2006.05.003 * Corresponding author. Fax: +1 613 520 3667. E-mail address: [email protected] (M. Se ´ne ´chal). Journal of Experimental Child Psychology 95 (2006) 231–254 www.elsevier.com/locate/jecp

Transcript of Morphological knowledge as revealed in children’s spelling accuracy and reports of spelling...

Journal of Experimental Child Psychology 95 (2006) 231–254

www.elsevier.com/locate/jecp

Morphological knowledge as revealedin children’s spelling accuracy and reports

of spelling strategies

Monique Senechal *, Michelle T. Basque, Tina Leclaire

Department of Psychology, Carleton University, Ottawa, Ont., Canada K1S 5B6

Received 25 July 2005; revised 8 May 2006Available online 30 June 2006

Abstract

The goal of the current research was to assess whether children can make strategic use of morpho-logical relations among words to spell. French-speaking children in Grade 4 spelled three word types:(a) phonological words that had regular phoneme–grapheme correspondences, (b) morphologicalwords that had silent consonant endings for which a derivative revealed the silent ending, and (c) lex-ical words that had silent consonant endings for which no familiar derivative revealed the ending.Children were also asked to provide immediate retrospective reports of the strategies used to spelleach word. Two experiments (Ns = 46 and 39) were conducted. As expected, children in Grade 4spelled phonological words more accurately than they did words with silent consonant endings. Inaddition, children spelled morphological words more accurately than they did lexical words. Reportsof using retrieval were associated with accurate performance across word types. Importantly, reportsof using morphological strategies to spell morphological words were associated with a similar level ofaccuracy, as were reports of using retrieval. Even though children reported using a phonological strat-egy frequently across all word types, this strategy was associated with accurate performance only forspelling phonological words. Experiment 2 replicated the results of Experiment 1 with another set ofstimuli and also showed that children’s morphological awareness predicted their spelling accuracy formorphological words as well as the reported frequency of morphological strategy use. In sum, thefindings revealed that most children showed evidence of adaptive strategy use.� 2006 Elsevier Inc. All rights reserved.

Keywords: Spelling; Morphology; Strategies; Grade 4

0022-0965/$ - see front matter � 2006 Elsevier Inc. All rights reserved.

doi:10.1016/j.jecp.2006.05.003

* Corresponding author. Fax: +1 613 520 3667.E-mail address: [email protected] (M. Senechal).

232 M. Senechal et al. / Journal of Experimental Child Psychology 95 (2006) 231–254

Introduction

Learning to spell involves learning how to map speech sounds onto letters, learninghow to apply the orthographic and grammatical rules of a written language, and learningthe exceptions to those rules. The pioneering work of Read (1971, 1975) revealed thatyoung children, before they are formally taught to spell, use their existing knowledgeabout language and letters when they attempt to capture in writing the sound structureof words. We now acknowledge that learning to spell is more than merely memorizing let-ter sequences because it also involves developing adaptive and efficient strategies (Gentry,1982; Rittle-Johnson & Siegler, 1999; Varnhagen, 1995). The goal of the current researchwas to assess whether the frequency and efficiency of strategies reported by French-speak-ing children varied when they spelled different word types. Of special interest was whetherthe children would report using morphological spelling strategies when they representedmorphological information in their spelling.

The French orthography

The French language is of particular interest because written French represents aspectsof morphology that are not represented phonologically. For instance, inflected words androot words often end with silent consonants that indicate the links to derivatives. Considerthat the plural most often is marked with a final silent s and that the silent t in certainnouns marks the relation with derivatives such as derived verbs (e.g., chant–chanter,debut–debuter). Hence, the predominantly silent morphology (i.e., many written markershave no corresponding pronunciation) means that children must learn how to spell wordscorrectly without an oral reference (Totereau, Thevenin, & Fayol, 1997). It has been sug-gested that children who spell French correctly rely on morphological, syntactic, and lex-ical knowledge in addition to sound-to-spelling transcription rules (Alegria & Mousty,1996, 1997; Pacton & Fayol, 2003; Pacton, Fayol, & Perruchet, 2005).

Given the morphological structure of French, children might use the morphologicalrelations among known words strategically to help them spell words accurately. By think-ing of the morphological relations among words, children can choose the correct spellingfrom other possible spelling alternatives. For instance, to spell accurately the silent t inchant, writers can use their understanding of the link between chant and chanter to writethe silent consonant ending. A morphological strategy can help in two ways: (a) by alertingchildren to the presence of a silent consonant ending and (b) by allowing the selection ofthe silent letter. Senechal (2000) conducted a study to examine whether children made useof morphological information when spelling French words that contain silent final conso-nants. To do so, she compared the spelling accuracy of three word types: phonological(labeled regular), morphological, and lexical (labeled deep). Phonological words werethose that contained consistent phoneme–grapheme patterns (Waters, Bruck, & Malus-Abramowitz, 1988), whereas both morphological and lexical words were those that endedwith silent consonants. Morphological words were defined as those for which a derivativeclearly revealed the silent consonant ending. For example, rang would be classified as amorphological word because the silent letter g in rang can be determined by thinking ofits derivative rangee. Finally, lexical words were those for which the final silent consonantneeded to be memorized because a derivative, if any, would not readily and clearly revealthe final silent consonant (e.g., tabac: tabagie or tabatiere).

M. Senechal et al. / Journal of Experimental Child Psychology 95 (2006) 231–254 233

Senechal (2000) hypothesized that spelling accuracy should be linked to how deeplychildren needed to go in the orthographic structure of words to extract the correct spelling.As expected, Senechal found that phonological words were the easiest to spell and thatmorphological words were easier to spell than lexical words. The significant advantageof morphological words over lexical words suggested that children are sensitive to themorphological structure of words. In fact, Senechal interpreted these findings as evidenceof morphological strategy use, but this evidence is indirect at best. Children might not usemorphological strategies at all; instead, they may find that orthographic representations ofmorphological words are easier to store in long-term memory due to the redundancy inorthographic patterns among root words and their derivatives. The current researchexamined whether children would report using morphological strategies and whetherthe reported use of morphological strategies would be linked to spelling accuracy.

Spelling strategies

Spelling strategies can be broadly categorized into mnemonic, phonological, ortho-graphic, and morphological strategies (Varnhagen, 1995). The most advanced strategyinvolves retrieving from memory the stored orthographic representations for words. Pre-sumably, with sufficient spelling and reading practice, the orthographic representations ofwords become stored in long-term memory along with the phonological and semantic rep-resentations and can be retrieved rapidly and automatically (Ehri, 2005; Reichle & Perfetti,2003; Rittle-Johnson & Siegler, 1999; Treiman, Berch, & Weatherston, 1993). Retrieval isalso necessary for words that do not obey the rules of phonetic, orthographic, or morpho-logical regularity. For example, the written form laugh must be stored in long-term mem-ory because it cannot be derived from sound–letter correspondences or from reliance onother linguistic information such as morphological information. When complete ortho-graphic representations are not readily retrievable from long-term storage, however, chil-dren may rely on backup strategies that allow them to capture the oral language in writing.Backup strategies such as phonological, orthographic, and morphological strategies con-sequently are more deliberate and conscious than is retrieval. In addition, backup strate-gies are more effortful and can be less accurate than retrieval (Rittle-Johnson & Siegler,1999).

Phonological strategies consist of analyzing a word into phonemes and then choosingthe appropriate graphemes to represent each sound (Varnhagen, 1995; Varnhagen,Boechler, & Steffler, 1999). Although phonological strategies are used most often in earlyspelling (Gentry, 1982), there is evidence that older children also use phonological strate-gies (Steffler, Varnhagen, Friesen, & Treiman, 1998). Phonological strategies, however,might not lead to the conventional spelling of words. Additional knowledge about theorthographic structure of the written language may be necessary. Orthographic strategiescome into play as children gain experience with print and build up a collection of words inmemory (Varnhagen, 1995), enabling them to make generalizations about the kinds of let-ter sequences that do and do not occur within a language (Goulandris, 1994). In additionto phonological and orthographic information, spellers may use morphological similaritiesacross words as clues to correct spelling. For instance, Treiman, Cassar, and Zukowski(1994) found that first and second graders were more accurate at spelling the t in the worddirty, which had a stem word ending in t, than they were at spelling the t in city, which hadno stem. Moreover, Carlisle (1988) showed that fourth, sixth, and eighth graders used their

234 M. Senechal et al. / Journal of Experimental Child Psychology 95 (2006) 231–254

knowledge of morphological relations to spell derived words. Other studies have shownthat children in Grades 2 to 4 also appear to be sensitive to the morphological structureof words in their spelling (Leybaert & Alegria, 1995; Senechal, 2000; Waters et al.,1988) and that training in morphological awareness can improve spelling performance(Nunes, Bryant, & Olsson, 2003).

The research reviewed so far includes studies that have indirectly given evidence ofstrategy use by comparing children’s spelling accuracy across different word types or byexamining patterns of spelling errors. The research on morphology has been limited to thisindirect assessment of strategy use. It has been argued, however, that no inferences can bemade about strategy use from accurate spelling performance (Rittle-Johnson & Siegler,1999) and that, consequently, important converging evidence about strategy use can beobtained by asking children directly about the strategies they use while spelling (Steffleret al., 1998; Turner & Quinn, 1986; Varnhagen, 1995; Varnhagen et al., 1999). Indeed,young children seem to be quite able to describe the strategies they use to spell. Rittle-Johnson and Siegler (1999) showed that children’s reports were consistent with measuresof speed and accuracy, thereby providing support for the validity of the reports. Similarly,Steffler and colleagues (1998) found that when accuracy was high (84–100%), children inGrades 4 and 5 reported using retrieval more often than backup strategies. In addition,Varnhagen (1995) provided evidence that children’s reported strategies varied accordingto the difficulty level and familiarity of words. Children reported using retrieval (e.g., ‘‘Iremembered the word’’) for words that were well known, using phonological strategiesfor phonologically simple but somewhat unfamiliar words (e.g., cold), and using ortho-graphic strategies (e.g., ‘‘I borrowed the ght from light’’) for words that required ortho-graphic knowledge. Children, however, did not report using morphological strategiesfor the sample of words in Varnhagen’s study. Will it be the case that children report usingmorphological strategies while spelling other word types? In the current study, childrenwere asked to spell and report their strategy use on words that require morphologicalknowledge to spell the final consonant accurately because that final consonant is silent(e.g., the p in camp is silent in French, but it is revealed in the derived form camper).

The current research

In the current research, word types identical to those of Senechal (2000) were used. Sim-ilar to Senechal’s study, it was hypothesized that the degree of phonological and morpho-logical transparency should affect spelling accuracy; that is, phonological words should bespelled correctly more often than morphological and lexical words, and morphologicalwords should be spelled correctly more often than lexical words. The current researchextended Senechal’s findings because children were asked to provide immediate retrospec-tive reports of the strategies they used to spell the words. It was hypothesized that strategyuse would vary across word types. If Rittle-Johnson and Siegler (1999) were correct, thenretrieval strategies should be reported more often for phonological words because buildingan orthographic representation for them should be more straightforward than doing so forwords with silent consonant endings. Predicting the pattern of phonological strategy useacross word types is more difficult. On the one hand, the findings of Varnhagen (1995)would suggest that phonological strategies should be reported more frequently for phono-logical words; on the other hand, if tagging a silent consonant at the end of words provesto be difficult for children, then children may treat them as equivalent to phonological

M. Senechal et al. / Journal of Experimental Child Psychology 95 (2006) 231–254 235

words and, consequently, may report using phonological strategies for them as well. Theprediction for morphological strategy use is more straightforward; that is, if morpholog-ical strategies are indeed a useful backup strategy for morphological words with silent con-sonant endings, then children should report using them more frequently for morphologicalwords than for the other words.

Predictions can also be made about strategy efficiency. According to previous research,retrieval strategies should be very efficient across all word types, whereas phonologicalstrategies should be efficient only for phonological words (Rittle-Johnson & Siegler,1999; Varnhagen, 1995). If Senechal (2000) was correct, then morphological strategiesshould be efficient for morphological words with silent endings. In the current research,children in Grade 4 participated because Carlisle (1996) found that children in Grades 2and 3 only begin to use derived forms in their spelling. Grade 4 was also selected to obtaina reasonable level of accurate performance so as to assess which strategies were linked tomore accurate performance. Two experiments, each with different stimuli, were conductedto assess the reliability of the findings.

Experiment 1

Method

Participants

Parental permission was granted to 49 children, but 3 children were excluded on thebasis of school-identified learning disabilities. The final sample consisted of 46 children(25 girls and 21 boys) in Grade 4. On average, the children were 9 years 8 months ofage (SD = 4 months). Parents reported on the language spoken at home by the children.According to parents, 33% of children spoke French exclusively, 50% spoke French andEnglish, and the remainder spoke French and another language. There were no statistical-ly significant differences between the monolingual and bilingual children (ps > .69) on thedependent variables; thus, child language is not discussed further.

The children were recruited from two French-speaking schools (18 and 27 childrenincluded in each) in Ottawa, Canada. The language of instruction was French for all aca-demic subjects with the exception of the English course. The provincial curriculum forwriting states that one of the 14 goals to be achieved under syntactic knowledge is to beable to categorize words into word families at the end of Grade 4. Informal interviews withteachers revealed that children were taught to spell by dictation or by relying on syntax.No teacher reported explicitly teaching students morphological strategies to spell words.

Materials

Children in this experiment were asked to spell 24 words. These words were divided intothree categories as in Senechal (2000). Phonological words (n = 6) were those that could bespelled by simple phoneme–grapheme translation and that contained no silent consonantendings. All remaining words had silent consonant endings. Morphological words (n = 11)were those for which a derivative clearly revealed the silent letter. For these words, thefinal silent consonant could be derived by using the word’s feminine equivalent (e.g.,blond–blonde) or using a derived verb or noun (e.g., gallop: galoper, galopant, or galopade).

236 M. Senechal et al. / Journal of Experimental Child Psychology 95 (2006) 231–254

The lexical words (n = 7) were those that had a silent consonant ending that could not beclearly revealed by the use of a derivative. For instance, foulard does not have a derivative,the derivatives for noeud were rare words and did not provide a straightforward link to thebase word, and the derivatives for tabac (tabagie and tabatiere) did not reveal the silentconsonant.

The 24 words, their phonetic transcriptions, and their characteristics are presented inthe Appendix. The three word types did not differ statistically (all ps > .24) in terms ofnumber of letters and syllables, orthographic and phonological neighbors (LEXIQUE:New, Pallier, Ferrand, & Matos, 2001), or frequency of occurrence in French schoolbooksappropriate for Grades 1 to 5 (MANULEX: Lete, Sprenger-Charolles, & Cole, 2004). Theword types, however, did differ in terms of phonology-to-orthography consistency of thefinal rime because phonological words had rimes that were more consistent than those inwords with silent consonant endings (LEXOP: Peereman & Content, 1999). The entire setof 24 words and the subset of 11 morphological words had good interitem reliability(alphas of .80 and .79, respectively), but the reliability of the other two subsets was lower,presumably due to the inclusion of fewer items (alphas of .19 and .59 for phonological andlexical words, respectively).

The order of presentation of the 24 words was randomly determined with the con-straints that there never were two consecutive words from the same category and thatevery fourth word was a phonological word. Half of the children received this word list,and the other half received the list in the reverse order. During testing, researchers saideach word individually, inserted the word into a sentence that clarified its meaning, andsubsequently repeated the word. Children’s written responses were scored for accuracyof the entire word and accuracy of the final silent consonant when applicable.

Verbal reports of strategy use

After spelling each word, children answered a series of questions that asked about thestrategies they used while spelling. The first question was a general open-ended question asin Steffler and colleagues (1998). Specifically, children were asked what was going on intheir head while they were spelling the word. This question was followed by four yes/noquestions similar to those used in Rittle-Johnson and Siegler (1999). Specifically, theresearchers asked, ‘‘Did you know how to spell the word by heart?,’’ ‘‘Did you spell theword by sounding it out?,’’ ‘‘Did you use another word to help you spell the whole word?,’’and ‘‘If so, what word did you use?’’ When children responded to the open-ended questionwith a specific strategy, children were not asked the corresponding yes/no question aboutthat strategy.

Scoring the verbal report .

Based on previous research, children’s responses were coded according to six strategytypes (Rittle-Johnson & Siegler, 1999; Steffler et al., 1998; Varnhagen, 1995). Reports werecoded as (a) a retrieval strategy when children reported having a representation of theword in memory (e.g., ‘‘I learned that word,’’ ‘‘I know it by heart,’’ ‘‘I saw it in a book’’),(b) a phonological strategy when children segmented the word into phonemes (e.g., /l/, /a/,/k/) or into syllables (e.g., /ta/, /ba/), (c) a morphological strategy when children used thefeminine inflection of a word (e.g., blonde to spell blond) or used a morphologically relatedverb or noun (e.g., galoper to spell galop) to spell the test word, (d) an analogical strategy

M. Senechal et al. / Journal of Experimental Child Psychology 95 (2006) 231–254 237

to capture all other instances where children used another word with an orthography sim-ilar to that of the test word (e.g., local for bocal), (e) a rule-based strategy when childrenused a specific orthographic rule (e.g., replace n with m when placed before b or p as intambour), and (f) another strategy to capture any response that could not be scoredaccording to the previous strategies. All responses were scored by two researchers, andinterscorer agreement was 97%. Disagreements were resolved by discussion.

Procedure

Each child was tested individually in a quiet area of the school such as the library orconference room. After spelling each word, the child answered the questions about strat-egy use.

Results and discussion

Spelling accuracy

Children’s mean spelling accuracy, reported in Table 1, differed as a function of wordtype, Wilks’ lambda (2, 44) = .316, p < .001. Two planned orthogonal comparisons wereconducted to test the hypothesized difficulty level of the different word types. Planned con-trasts were also conducted on items and reported with the subscript i (i.e., Fi); these sup-plemental analyses were either identical to the analyses by participants or not significant.As expected, children spelled phonological words correctly more frequently than they didwords with silent consonant endings (i.e., morphological and lexical words),F (1, 44) = 39.85, MSe = 529.36, p < .001, Fi (1,21) = 5.92, p = .03. The second contrastshowed that children spelled morphological words correctly more frequently than theydid lexical words, supporting the hypothesis that words for which the silent consonantendings can be derived are easier to spell than words for which the silent consonant end-ings must be memorized, F (1, 44) = 57.72. MSe = 138.77, p < .001, Fi (1,21) = 2.89,p = .10.

A second analysis was conducted on children’s accurate spelling of the silent consonantendings to assess whether the same pattern of differences would hold across word types(Table 1). Children spelled the final consonant correctly more frequently for morpholog-ical words than for lexical words, F (1, 45) = 74.58, MSe = 145.37, p < .001,

Table 1Mean percentages correct performance and standard deviations for the entire words and final consonants as afunction of word type in Experiment 1

Mean SD

Whole word accuracy

Phonological words 75.0 17.5Morphological words 58.1 24.6Lexical words 39.4 25.1

Final consonant accuracy

Morphological words 65.8 24.5Lexical words 44.1 25.3

238 M. Senechal et al. / Journal of Experimental Child Psychology 95 (2006) 231–254

Fi (1,15) = 4.39, p = .055. Therefore, the pattern of findings for the final consonant is iden-tical to that obtained for the entire word, and these results replicate those obtained bySenechal (2000).

Verbal reports of strategy use

After spelling each word, children immediately reported the strategy they employedto spell that word. Three strategies—retrieval, phonological, and morphological—werereported sufficiently often to warrant further analysis. The frequencies with which chil-dren reported these strategies are presented as percentages in Table 2 (the percentagesdo not add to 100 because children could report using more than one strategy for anysingle word). All remaining strategies combined accounted for less than 8% ofresponses.

The frequency of use of each strategy was examined to assess whether it varied acrossword types. The analysis for retrieval showed a significant word type effect, Wilks’ lambda(2,44) = .543, p < .001. Consequently, the same two contrasts used to analyze spellingaccuracy were used to analyze the differences across word types. The first comparisonshowed that children reported using a retrieval strategy more frequently with phonologicalwords than with the two other word types, F (1,44) = 37.55, MSe = 235.10, p < .001.These results are consistent with the idea that phonological words, having consistent pho-neme–grapheme correspondences, are more easily encoded in the orthographic lexicon.Children reported using retrieval as frequently for morphological words as for lexicalwords, F < 1.

As indicated in Table 2, phonological strategies were reported frequently across allword types, and this may have resulted in the lack of significant findings for this variable,Wilks’ lambda (2,44) = .914, p = .14. These results suggest that children in Grade 4reported using a phonological strategy for all word types. In contrast, the use of morpho-logical strategies generally was restricted to morphological words, providing key supportfor the notion that children can make use of their knowledge of morphological derivativesto spell words, F (1, 44) = 49.71, MSe = 459.99, p < .001.

Effectiveness of strategy use

In this section, analyses were conducted to examine whether specific strategies wereassociated with accurate performance. Of special interest was whether the reported useof morphological strategies would be associated with accurate spelling. The conditionalprobabilities of spelling words correctly given reports of one strategy or another were

Table 2Mean percentages and standard deviations of reported strategies as a function of word type in Experiment 1

Word type Strategy type

Retrieval Phonological Morphological

Mean SD Mean SD Mean SD

Phonological 73.9 22.4 58.3 36.3 5.1 9.2Morphological 58.3 29.0 54.0 35.3 34.3 30.9Lexical 56.5 25.5 59.3 30.7 2.8 5.7

Table 3Conditional probabilities of spelling correctly (and numbers of children who reported using a strategy at leastonce) as a function of reported spelling strategy and word type in Experiment 1

Word type Strategy type

Retrieval Phonological Morphological

Phonological .856 (46) .638 (39) .500 (12)Morphological .711 (45) .461 (41) .728 (35)Lexical .621 (45) .369 (44) .222 (9)

M. Senechal et al. / Journal of Experimental Child Psychology 95 (2006) 231–254 239

calculated and are presented in Table 3. To test whether a strategy would prove to besuccessful, we tested the obtained probability against one set arbitrarily at 50%. Thatis, would a reported strategy be associated with a probability greater than 50% of spell-ing a word correctly? As expected from past results, children who reported usingretrieval were likely to spell the words correctly across all three word types and todo so with a probability greater than 50%, ts P 2.35, ps < .01 (one-tailed), withdfs = 45, 44, and 44 for phonological, morphological, and lexical words, respectively.In contrast, reports of using a phonological strategy were less likely to be associatedwith accurate spelling. In fact, only the spelling of phonological words had a greaterlikelihood of being associated with accurate spelling, t (38) = 2.90, p = .01 (one-tailed),whereas a phonological strategy did not lead to a greater than 50% likelihood of accu-rately spelling words with silent consonant endings. Finally, children who reportedusing morphological strategies to spell morphological words tended to spell the wordscorrectly and with a greater probability than 50%, t (34) = 4.82, p < .001. This latterfinding is consistent with the proposition that using morphological strategies canincrease children’s spelling accuracy.

Table 3 also provides information about the number of children who reported using thevarious strategies at least once. As indicated, most children reported using retrieval andphonological strategies. On average, 98% of children reported using a retrieval strategy,and 90% of children reported using a phonological strategy. In addition, most childrenreported using a morphological strategy (76%) at least some of the time. These findingssuggest that most, but not all, children in Grade 4 can make use of morphologicalstrategies.

Spelling errors

Spelling errors for the words with silent consonant endings were analyzed. Of the 406errors made, 87% were errors with the silent consonant. Clearly, silent consonant endingsrepresented the major difficulty with spelling these words accurately. Children made threeerror types: omitting the silent consonant ending altogether, omitting the silent consonantand replacing it with the letter e, and substituting the silent consonant with another one.The most frequent error made by children was the omission of the silent consonant ending(52%, with 10% of these omissions including the addition of the silent e). In 48% of cases,children substituted the silent letter with another consonant.

In sum, the findings of Experiment 1 replicated those obtained by Senechal (2000) byshowing the expected difficulty level across word types. Of specific interest was the findingthat morphological words were easier to spell than lexical words. These findings extend

240 M. Senechal et al. / Journal of Experimental Child Psychology 95 (2006) 231–254

those of previous research on strategy use that had not found that children reported mor-phological strategies by showing that most children reported using morphological strate-gies to spell morphological words some of the time and that their reported use ofmorphological strategies was efficient (Rittle-Johnson & Siegler, 1999; Steffler et al.,1998; Varnhagen, 1995). That is, children who reported using morphological strategieswere just as likely to spell words accurately as were those who reported retrieving the spell-ing from memory.

Experiment 2

Although the findings of Experiment 1 replicate those of Senechal (2000), it isimportant to demonstrate that the findings generalize to other words. Another limita-tion of the words selected in Experiment 1 is that the phonological words had finalrimes that were more consistent than those in the other two word types, and this couldaccount for the obtained advantage for the phonological words. Hence, Experiment 2was conducted on another series of words. A second goal of Experiment 2 was toexplore individual differences in strategy use. Specifically, it was of interest to assesswhether children’s awareness of the derivative morphology of the French languagewould be related to the reported use of morphological strategies. Carlisle (1995)referred to morphological awareness as the ability to think about and manipulatethe morphemic structure of words. Similarly, Green and colleagues (2003) defined mor-phological awareness as one’s sensitivity to the meaning-related structure of words.This awareness includes one’s knowledge of inflections and derivatives. Several studieshave found a solid predictive relation between morphological knowledge and readingskills (Casalis & Louis-Alexandre, 2000; Deacon & Kirby, 2004; Mahony, Singson,& Mann, 2000; Nagy, Berninger, Abbott, Vaughan, & Vermeulen, 2003) as well asbetween morphological awareness and spelling (Fowler & Liberman, 1995; Rubin, Patt-erson, & Kantor, 1991). Results based on studies of morphological awareness revealthat poor spellers are not as familiar with the morphological structure of words asare good spellers (Carlisle, 1987; Fischer, Shankweiler, & Liberman, 1985). Senechal(2000) showed that morphological awareness predicted children’s accurate spelling ofmorphological words. Could it be the case that morphological awareness also predictsthe use of morphological strategies? A second goal of Experiment 2 was to explore thisquestion.

Method

Participants

The data for 39 children (25 girls and 15 boys) in Grade 4 were included, whereas thedata for one child, identified by the school board as gifted, were excluded. On average, theparticipants were 9 years 9 months of age (SD = 5 months), and all of the children in thisexperiment attended one of the schools in Experiment 1. According to parents, 30% ofchildren spoke French exclusively, and the remainder spoke French and English. Therewere no statistically significant differences between the monolingual children and the bilin-gual children (ps > .12) on the dependent variables; thus, child language is not discussedfurther.

M. Senechal et al. / Journal of Experimental Child Psychology 95 (2006) 231–254 241

Materials

Spelling .

Experiment 2 included the same three word types as those used in Experiment 1: pho-nological (6 words), morphological (12 words), and lexical words (6 words). With theexception of the word bavard, the words in this experiment were different from those usedin Experiment 1. The 24 words, their phonological transcriptions, and their characteristicsare presented in the Appendix. The three word types did not differ statistically (all ps > .24)in terms of number of letters and syllables, orthographic and phonological neighbors, fre-quency of occurrence in French schoolbooks appropriate for Grades 1 to 5, or phonology-to-orthography consistency. Note that one word in the lexical category did have a rarederivative (escargot–escargotiere), but we judged that the infrequency of the derivative(i.e., not found in the MANULEX or LEXIQUE.3 database and with a frequency of.03 per million in the LEXIQUE.2 database) would mean that children in Grade 4 wouldtreat escargot as they would other lexical words. The entire set of 24 words and the subsetof 12 morphological words had good interitem reliability (alphas of .86 and .85, respective-ly), but the reliability for the other two subsets was lower (alphas of .58 and .58 for pho-nological and lexical words, respectively), probably due to the inclusion of fewer items.

Strategy use .

Based on the results from Experiment 1, two changes were made to the testing protocolto ensure that testing could be completed in a single session. First, the open-ended ques-tion was modified to minimize the overlap with the closed-ended questions, and it wasasked after the closed-ended questions. Second, questions about rule-based strategies wereeliminated because the reported frequency of use with this stimuli type was too low to war-rant analysis. Consequently, children answered a series of closed-ended questions thatwere followed by an open-ended question. The questions, identical to those in Experiment1, were as follows: ‘‘Did you know how to spell the word by heart?,’’ ‘‘Did you spell theword by sounding it out?,’’ ‘‘Did you use another word to help you spell the whole word?,’’and ‘‘If so, what word did you use?’’ The open-ended question differed from that in Exper-iment 1 because in Experiment 2 children were asked whether there was anything else theywould add to help a friend spell the word (Varnhagen, 1995). All responses were scored bytwo researchers, and interscorer agreement was 99%. Disagreements were resolved bydiscussion.

Morphological awareness .

This task was identical to the morphological awareness task elaborated by Senechal(2000) and inspired by the word analogy task developed by Nunes, Bindman, and Bryant(1997). During this task, the researcher said a pair of words that were linked by a morpho-logical relation (e.g., the noun danse and the derived verb danser). Afterward, the research-er said the first item of a second pair of words (e.g., saut) that had the same morphologicalrelation as the first pair of words. The participant needed to determine the missing itemthat completed the second pair of words (e.g., sauter). Within each pair of words, oneword always ended with a silent letter. This task consisted of 3 practice items and 12 testitems that are presented in the Appendix. Each child’s score was based on the number ofcorrect items. The maximum score was 12. The interitem reliability for this task was some-what low (Cronbach’s alpha = .64).

242 M. Senechal et al. / Journal of Experimental Child Psychology 95 (2006) 231–254

General spelling skills .

Children’s general spelling skills were assessed using Form B of the orthographic skillssubtest of the Test de Rendement pour Francophones (Sarazin, 1995). In this subtest, chil-dren were instructed to identify the single misspelled word in an array of four words (e.g.,bleu, feu, pneu, and vieu). Children were presented with a booklet that contained 2 practiceitems and 45 test items and were given 25 min to complete the task. Form B of the subtestis intended for children in Grade 5 or above, but it was decided to use it instead of Form Abecause a previous study showed ceiling effects for Form A in Grade 4. Each child’s scorewas based on the number of correct items. The interitem reliability of this subtest in thecurrent experiment was good (Cronbach’s alpha = .85).

Procedure

The general spelling test was group administered in the children’s regular classroom,whereas the experimental spelling task and the morphological awareness task were admin-istered individually in a quiet room. Group testing preceded individual testing, and theexperimental spelling task was administered before the morphological awareness task.The testing was conducted at the end of April.

Results and discussion

Preliminary analyses were conducted to assess whether children’s performance wouldchange over the course of the testing session as a result of the questioning procedure.The analyses, comparing children’s performance on the first and second halves of thetwo possible testing orders, failed to reveal any differences in spelling accuracy or in thefrequency of reporting retrieval, phonological, and morphological strategies (ps > .34).

Spelling accuracy

Examination of the descriptive statistics for the individual words revealed that the per-formance of one word, marquis, was different from that of others. Specifically, it had thelowest overall correct performance (.15), and it was the only word for which the percent-age correct did not increase when one considered only the spelling of the final consonant.Examination of marquis within its word type category showed that performance on it wasmore than 1 standard deviation below the word type mean. No other silent consonantword showed these characteristics. Therefore, it was decided not to include marquis in sub-sequent analyses on the basis that children might not have known this word.

Children’s spelling accuracy did vary across word types (as reported in Table 4), Wilks’lambda (2,37) = .386, p < .001. Two planned orthogonal comparisons were conducted tocompare children’s spelling of the different word types. As expected, the first comparisonrevealed that children spelled phonological words correctly more frequently than they didthe other two word types (i.e., morphological and lexical words), F (1,37) = 25.23,MSe = 69.42, p < .001, Fi (1, 20) = 11.16, p = .001. The second contrast showed that chil-dren spelled morphological words correctly more frequently than they did lexical words,F (1,37) = 20.29, MSe = 1.63, p < .001, Fi (1, 20) = 3.79, p = .08. The analysis for the accu-racy of silent endings yielded the same pattern; that is, children spelled the final silent con-sonant of morphological words accurately more often than they did that of lexical words,

Table 4Mean percentages correct performance and standard deviations for entire words and final consonants as afunction of word type in Experiment 2

Word type Mean SD

Whole word accuracy

Phonological words 70.9 24.1Morphological words 51.5 29.8Lexical words 38.5 27.6

Final consonant accuracy

Morphological words 59.0 29.9Lexical words 50.4 26.1

M. Senechal et al. / Journal of Experimental Child Psychology 95 (2006) 231–254 243

F (1, 37) = 9.16, MSe = 1.56, p < .01, Fi < 1. These findings support the view that wordsfor which the silent consonant can be derived are easier to spell than words for whichthe silent consonant must be memorized.

Verbal reports of strategy use

Children’s reports of the strategy they employed while spelling each word are presentedin Table 5. Only retrieval, phonological, and morphological strategies were analyzedbecause all of the remaining strategies combined accounted for less than 8% of responses.Within these infrequent strategy reports, we examined whether children would reportusing analogies to English to spell words accurately because in some cases the Englishequivalent revealed the silent consonant ending. There was only one child who reportedusing an analogy to an English word, and it was for the word idiot. Therefore, makinganalogies to the English language was not a strategy that the children readily reportedif they used it at all.

Two planned orthogonal comparisons were conducted to compare whether children’sreported strategies varied according to word type. The first comparison showed that chil-dren reported using a retrieval strategy more frequently with phonological words thanwith the two other word types, F (1, 37) = 28.66, MSe = 2.69, p < .001. As in Experiment1, these results suggest that phonological words, having consistent phoneme–graphemecorrespondences, are more easily memorized. In contrast to Experiment 1, however,retrieval was reported more frequently for morphological words than for lexical words,F (1, 37) = 13.54, MSe = 1.44, p < .001, suggesting that words for which the silent endingcan be tagged to a derivative may be easier to encode in long-term memory. Children in

Table 5Mean percentages and standard deviations of reported strategies as a function of word type in Experiment 2

Word type Strategy type

Retrieval Phonological Morphological

Mean SD Mean SD Mean SD

Phonological 62.8 28.5 66.7 37.3 00.0 00.0Morphological 50.6 28.0 66.9 31.4 18.0 22.4Lexical 40.6 30.8 77.8 25.7 00.0 00.0

244 M. Senechal et al. / Journal of Experimental Child Psychology 95 (2006) 231–254

Experiment 2 reported using a phonological strategy frequently across word types, butthey did so more frequently for lexical words than for morphological words,F (1,37) = 14.47, MSe = 1.59, p < .001. In comparison, they restricted their reports ofmorphological strategies to morphological words, as indicated in Table 2. On average,children reported using morphological strategies for morphological words 18% of thetime, but they did not report morphological strategies for phonological words and veryseldom (<1%) reported them for lexical words. Only the percentage for morphologicalwords was greater than zero, t (38) = 5.00, p < .001.

Effectiveness of strategy use

Analyses were conducted to examine whether specific strategies were associated withaccurate performance. The conditional probabilities of spelling words correctly givenreports of one strategy or another were calculated and are presented in Table 6. As inExperiment 1, a test of the effectiveness of the strategies was conducted by comparingthe obtained probability against one set arbitrarily at 50%. Children who reported usingretrieval were likely to spell the words correctly for the phonological and the morpholog-ical words but not for the lexical words, ts (35, 35, 31) = 10.28, 5.03, 0.63, ps < .001, .001,.26 (one-tailed) for phonological, morphological, and lexical words, respectively. In con-trast to Experiment 1, the probability of correctly spelling lexical words was only 54%when a retrieval strategy was reported (vs. 62% in Experiment 1).

The reported use of phonological strategies was as expected; that is, only the spellingof phonological words, t (35) = 10.28, p < .001 (one-tailed), had a greater likelihood ofbeing associated with accurate spelling, whereas a phonological strategy was not asso-ciated with a greater than 50% likelihood of accurately spelling words with silent con-sonant endings. Finally, children who reported using morphological strategies werelikely to spell the words correctly and to do so with a greater probability than 50%,t (24) = 4.02, p < .001 (one-tailed). As in Experiment 1, this latter finding is consistentwith the proposition that using backup morphological strategies can increase children’sspelling accuracy.

Table 6 also provides information about the number of children who reported usingeach strategy. As indicated, most children reported using retrieval and phonological strat-egies. On average, 88% of children reported using a retrieval strategy, and 94% of childrenreported using a phonological strategy. Finally, 64% of children reported using a morpho-logical strategy at least once. As in Experiment 1, these findings suggest that many, but notall, children in Grade 4 can make use of morphological strategies.

Table 6Conditional probabilities of spelling correctly (and numbers of children who reported using a strategy at leastonce) as a function of reported spelling strategy and word type in Experiment 2

Word type Strategy type

Retrieval Phonological Morphological

Phonological .853 (36) .690 (35) .500 (2)Morphological .752 (36) .432 (39) .798 (25)Lexical .544 (32) .349 (38) —a

a No children reported using the strategy.

M. Senechal et al. / Journal of Experimental Child Psychology 95 (2006) 231–254 245

Spelling errors

Spelling errors for the silent consonant endings were analyzed. Of the 361 errors made,82% were errors with the silent consonant. As in Experiment 1, silent consonant endingswere the major difficulty for spelling these words accurately. The most frequent error madeby children was the omission of the silent consonant ending (85%), with 16% of these omis-sions including the addition of the silent e. In 13% of cases, children replaced the silentletter with another consonant.

Morphological awareness

The second goal of Experiment 2 was to explore whether children’s morphologicalawareness would predict children’s reported use of morphological strategies. Childrenwere able to solve many of the analogies in the morphological awareness task(M = 8.44 [of 12], SD = 2.15) and, on average, succeeded at identifying 18.26 spellingerrors (of 45 items) on the general spelling test. The correlation coefficients among theseand other variables are reported in Table 7. An examination of the coefficients showed thatmorphological awareness was moderately related to both the spelling accuracy of morpho-logical words and the frequency of reporting morphological strategies. Interestingly, mor-phological awareness and morphological strategies were also related to spelling lexicalwords but not to spelling phonological words. Finally, spelling words with silent conso-nant endings was strongly interrelated but was not significantly related to spelling phono-logical words. These findings highlight the fact that spelling phonological words accuratelydoes not guarantee accuracy for words with silent consonant endings.

The results of fixed-order hierarchical regressions are presented in Table 8. These anal-yses tested whether the association between morphological awareness and morphologicalspelling, as well as that between morphological awareness and morphological strategies,would hold once general spelling skills were controlled. The first analysis showed that mor-phological awareness explained 7% of unique variance in spelling morphological wordsonce general spelling skills were controlled. This finding replicates that reported by Sene-chal (2000). The second analysis revealed that morphological awareness accounted for13% of unique variance in the reported frequency of using morphological strategies aftercontrolling for general spellings skills. This second finding extends that reported by Sene-chal by showing that children with greater awareness of the morphology of the French

Table 7Zero-order correlations among variables in Experiment 2

Morph. aware. Morph. strat. Sp. morph. Sp. phono Sp. lexical

Morph. strat. .453Sp. morph. .469 .512Sp. phono. .217 .200 .173Sp. lexical .449 .338 .807 .274Sp. general .385 .324 .606 .549 .629

Note. r > .31, p = .05 (two-tailed). Morph. aware., morphological awareness; Morph. strat., frequency ofreporting morphological strategies; Sp. morph., spelling accuracy for morphological words; Sp. phono., spellingaccuracy for phonological words; Sp. lexical, spelling accuracy for lexical words; Sp. general, general spellingskills as measured by the standardized test.

Table 8Fixed-order hierarchical regressions assessing the relation between morphological awareness and other variablesafter controlling for general spelling skills

Criterion and predictor R2 R2 change F change p

Accuracy for morphological words

General spelling .333 .333 18.46 .000Morphological awareness .404 .072 4.33 .045

Morphological strategies

General spelling .105 .105 4.33 .044Morphological awareness .231 .126 5.92 .020

Accuracy for phonological words

General spelling .302 .302 16.00 .000Morphological awareness .302 .000 0.00 .966

Accuracy for lexical words

General spelling .396 .396 24.24 .000Morphological awareness .446 .050 3.26 .079

246 M. Senechal et al. / Journal of Experimental Child Psychology 95 (2006) 231–254

language tend to report using morphological strategies more frequently. Finally, the lasttwo analyses demonstrate that the pattern seems to be limited to the morphologicaldomain because morphological awareness did not account for statistically significant var-iance in spelling phonological words or lexical words.1

General discussion

The goal of the current research was to describe the interplay between Grade 4 chil-dren’s understanding of oral language and their spelling. The assumption guiding thisresearch is that children can make conscious and deliberate use of morphological relationsamong words during spelling acquisition. If this assumption is correct, then spelling accu-racy and reports of strategy use should vary according to the morphological transparencyof words. To test this hypothesis, children’s spelling accuracy and reports of strategy usewere examined across three word types: phonologically transparent words, morphological-ly transparent words that are not phonologically transparent, and words that are neitherphonologically nor morphologically transparent. The results from this research are dis-cussed in terms of spelling accuracy, frequency of strategy use, efficiency of strategy use,and morphological awareness and strategy use.

1 Further examination of the morphological awareness task revealed that half of the items might be easier tosolve than the other half because the first pair of words shared the silent consonant ending with the target pair aswell as the derived morpheme; consequently, the correct answer could be obtained by matching the final syllableof the initial derivative with the target root word. For example, the correct item toiture (from toit, where the final t

is silent) could be obtained by analogy with the derivative in the initial word pair ecriture (from ecrit, where the t

is silent). This possibility of using a purely syllabic analogy with the derived word in the initial pair required twosubsequent analyses. First, the relative difficulty of the two item types was examined. This analysis revealed thatthe 6 items that could be solved with a syllabic analogy (M = 4.05) were not easier than the 6 items that clearlyrequired a morphological analogy (M = 4.38, p = .14). Second, the regression analyses reported in Table 8 wereconducted with children’s performance on the 6 items that required a morphological analogy, and these analysesresulted in an identical pattern of significant findings as reported previously.

M. Senechal et al. / Journal of Experimental Child Psychology 95 (2006) 231–254 247

Spelling accuracy

In the current research, children in Grade 4 spelled phonological words correctly morefrequently than they did words with silent consonant endings. Accordingly, spelling accu-racy seems to be related to the clearness of the relation between speech segments and gra-phemes, as was found in previous research (Varnhagen, 1995). This certainly is not a novelfinding, but the discrepancy in accuracy between the phonological words and the wordswith silent consonant endings highlights the difficulty in building complete and accurateorthographic representations when there is a mismatch between the phonology of the orallanguage and its written representation. The analysis of children’s errors supports thisclaim because children omitted the silent consonant ending in the majority of cases.

Among the words with silent consonant endings, children spelled morphological wordscorrectly more often than they did lexical words. This was true for the spelling of the wholeword as well as for the spelling of the final consonant. This finding suggests that buildingcomplete orthographic representations is easier for words for which the silent consonantending can be derived than for words for which the silent letter must be memorized. Theseresults are congruent with those of other researchers (e.g., Leybaert & Alegria, 1995;Senechal, 2000; Waters et al., 1988). The obtained pattern of findings, however, must beinterpreted cautiously because the current research included a limited number of wordsfor each category. In addition, the word characteristics were analyzed globally for eachcategory instead of matching the characteristics of specific words across categories. Theselimits might have resulted in more variability within and across categories than was opti-mal. Nonetheless, the fact that the findings were replicated across samples (Experiment 1and Senechal, 2000) and across word lists (Experiments 1 and 2) provides some evidencefor the robustness of the obtained pattern.

Frequency of strategy use

The most frequent strategies reported by far were retrieval and phonological strategies,and this is consistent with previous research (e.g., Rittle-Johnson & Siegler, 1999; Steffleret al., 1998). As expected, the frequency with which children reported using different strat-egies varied across word types. Similar to the findings of Varnhagen (1995, p. 277), chil-dren in the current study reported retrieving from memory the spelling of phonologicalwords more often than the spelling of other word types. The orthographic patterns ofwords that have regular phoneme–grapheme correspondences are easier to encode andretrieve automatically due to the overlap between their phonological and orthographicfeatures (Ehri, 2005; Reichle & Perfetti, 2003). Retrieval strategies were used less frequent-ly for words with silent consonant endings. In fact, children generally were just as likely toreport a phonological strategy as to report a retrieval strategy for words with silent con-sonant endings. The frequency of reporting phonological strategies for words with silentconsonant endings is in accord with the error pattern observed with this word type. Thatis, omission of the final letter was the major error for these words; consequently, childrenwho do not know that the words have a silent consonant might also think that a phono-logical strategy is appropriate. Encoding orthographic features that have no phonologicalvalue seems to be particularly difficult during spelling acquisition.

The most important contribution of the current research is the demonstration that mostchildren at least once reported using a backup morphological strategy when spelling

248 M. Senechal et al. / Journal of Experimental Child Psychology 95 (2006) 231–254

morphological words. Varnhagen (1995) reported that children seldom reported using thisstrategy, but the current findings clearly show that this depends on stimuli selection. Thereports of morphological strategies in the current study can be seen as an example of adap-tive strategy choice (Rittle-Johnson & Siegler, 1999). That is, children who have morpho-logical strategies in their repertoires will tend to use them when most appropriate. Thisbeing said, children in Grade 4, when spelling morphological words, reported using aretrieval strategy (58% in Experiment 1 and 54% in Experiment 2) or a phonological strat-egy (54% in Experiment 1 and 67% in Experiment 2) more often than they reported using amorphological strategy (34% in Experiment 1 and 18% in Experiment 2). This patternhighlights two things: (i) that some children eventually move from reliance on backupstrategies to retrieval while other children retain a more basic strategy and (ii) that childrendo not make optimal use of morphological strategies. Of course, these possible explana-tions should be examined in cross-sectional work, whereby one would expect a changewith spelling experience. Certainly, more research is needed to understand better how chil-dren come to use morphological relations among words and whether instruction can affectspelling performance.

Strategy effectiveness

The finding that retrieval is the most efficient strategy across word types supports theidea that children, with sufficient spelling and reading experience, become able toretrieve accurate and complete orthographic representations of words (Rittle-Johnson& Siegler, 1999; Steffler et al., 1998). Children’s ability to report on using retrievalas shown in the current research, however, does not presuppose that children have con-scious awareness of the underlying processes used during retrieval or how their ortho-graphic representations are linked to phonological, morphological, and semanticknowledge. In the current research, children did err sometimes even when they reportedusing retrieval. That is, children who reported using retrieval still made errors 14% ofthe time when spelling phonological words and up to 46% of the time when spellinglexical words, suggesting that some children might have inaccurate orthographic repre-sentations encoded in memory, as is suggested by the frequent omission of final silentconsonants.

Another important contribution of the current research is the finding that morpholog-ical strategies were efficient backup strategies to retrieval. Children who reported usingmorphological strategies tended to spell the words correctly and generally did so at thesame level as children who reported using retrieval. These findings support and extend pre-vious research that investigated indirect evidence of the value of morphological strategyuse (Carlisle, 1988; Senechal, 2000; Waters et al., 1988). The specificity of reporting mor-phological strategies when spelling morphological words and the accuracy linked withthese reports provide some converging evidence about the validity of children’s introspec-tive reports. The fact that children reported using retrieval more frequently than they didbackup morphological strategies suggests that, for some of these words, a backup strategymight not be necessary even though the backup strategy can be efficient. It would be ofinterest to examine the developmental progression of the reported use of these strategiesto assess whether a morphological strategy is more likely before morphological wordsare fully encoded and readily retrievable from children’s orthographic lexicon. We selecteda sample of children in Grade 4 based on previous research showing that younger children

M. Senechal et al. / Journal of Experimental Child Psychology 95 (2006) 231–254 249

are less likely to rely on morphology (Carlisle, 1996). It would be of interest to conduct thisresearch with younger and older children to assess the role of spelling experience on theuse of morphological strategies.

Children’s reports of using phonological strategies were efficient only for spellingphonological words and not for spelling words with silent consonant endings. Indeed,phonological strategies for silent consonant words never exceeded the 50% probabilitylevel that was set as the benchmark of strategy efficiency. This pattern of findingshighlights the fact that children need to adapt their spelling strategies to differentword types because failure to do so may lead to erroneous spelling. In addition,the consistency between the reduced accuracy and reports of inefficient strategies alsoprovides converging evidence about the validity of children’s introspective reports. Inboth experiments, however, children had more than a 30% probability of spelling lex-ical words correctly when they reported using a phonological strategy. This finding issomewhat puzzling because a phonological strategy cannot yield a correct responseand may cast doubts on the validity of children’s responses. We examined children’sresponses in Experiment 1 further because children reported specifically on the strat-egies used to spell the final letter, whereas in Experiment 2 children reported onthe entire word. For Experiment 1, we found that 83% of children who reported usinga phonological strategy to spell the final letter of lexical words correctly also reportedusing retrieval. Thus, retrieval might account for most, but not all, of the accuracyperformance. In the remaining 17% of cases, it is possible that children might haveused a strategy whereby they would guess the final consonant. Recall that 48% and13% of errors in Experiments 1 and 2, respectively, consisted of the substitution ofthe final silent consonant with another one. For these children, sometimes theyguessed correctly. These explanations, however, do not explain why children wouldreport using a phonological strategy for the final letter. Unfortunately, we did notprobe children to assess further whether their responses reflected their segmentingthe word rather then using a purely phonological strategy. This possibility could beexplored in future research.

In the current research, we imposed categorical boundaries among the phonological,morphological, and lexical words. These boundaries might appear to be arbitrary becausesome of the phonological and lexical words had derivatives; consequently, the wordsselected for each category might not have provided an adequate test that morphologicalwords form a distinct natural class of words in the children’s mind that can be used whenspelling. In our view, the acquisition of morphological knowledge is gradual and always isconstrained by the children’s lexicon. Accordingly, words that have derivatives that arerare enough to be unknown to the children should be treated as lexical words by the chil-dren. The findings that no children reported using a morphological strategy for the lexicalwords in Experiment 2 and that only 9 children reported using a morphological strategyfor a single lexical word (i.e., choix) in Experiment 1 lend some credence to this theoreticalassumption. We also view children’s use of spelling strategies as adaptive. Accordingly,children’s use of a morphological strategy should not be very helpful to spell phonologicalwords (e.g., using tambourin to spell tambour) or lexical words (e.g., using choisir to spellchoix). Again, the obtained data suggest that this is the case; that is, morphological strat-egies seldom were used and were not very helpful to spell phonological or lexical words.For instance, 7 of 9 children who reported using choisir mistakenly replaced the final x

in choix with an s. Moreover, and most important, the reverse pattern was found for

250 M. Senechal et al. / Journal of Experimental Child Psychology 95 (2006) 231–254

morphological words; that is, the backup strategy that was most adaptive to spell morpho-logical words was the use of morphological strategies. Nonetheless, future research shouldexplore whether and how children come to appreciate the limits of using morphologicalstrategies.

Morphological awareness

The current research demonstrated a reliable relation among morphological skills.That is, children in Grade 4 with greater morphological awareness tend to report usingmore morphological strategies and to spell morphological words correctly. These find-ings provide converging evidence for the key role of morphological knowledge. At thesame time, these results remain exploratory for two reasons. First, the sample size usedto conduct these analyses was quite small; consequently, these exploratory findingsshould be replicated with larger samples. Second, the morphological awareness task usedin the current study and elsewhere (Senechal, 2000) could be improved. The interitemreliability of the task was low (alpha of .64). The reliability of a given task sets the upperlimit of the correlation that task can have with other variables (Cohen, 1988). Hence, itis possible that with increased reliability, the morphological awareness task might havestronger relations with the key outcomes. The morphological awareness task also couldbe improved by including exclusively derivatives in the initial word pair and the targetword pair that do not share a common final syllable to ensure that children are usingtheir morphological knowledge, rather than using some other form of analogy, to solveeach item.

The strong association between spelling morphological words and spelling lexicalwords is also of interest. One plausible explanation for this relation might be that, dur-ing spelling acquisition, children who can distinguish between these two word typesknow that they can rely on morphological strategies to spell the endings of morpholog-ical words. Hence, morphological strategies should facilitate the process of encodingmorphological silent endings in long-term memory. These children might find the taskof learning the silent endings of lexical words easier for two reasons. First, they mighthave more resources available to encode silent consonant endings in lexical words oncethey have encoded and can readily retrieve the silent endings of morphological words.Second, knowing the distinction between the two word types, children might pay par-ticular attention to the silent endings of lexical words because they know that theymust commit their endings to memory. A similar explanation was proposed by Ehri(2005) to explain how good decoding skills might facilitate the reading of irregularwords.

Conclusion

Even though children principally learn to spell in school (Allal, 1997), teachers are stilluncertain about the amount of attention that should be placed on spelling and about thespecific spelling patterns that occur in a language (Templeton & Morris, 1999). Carlisle(1988) argued that suitable instruction might help students to develop a more explicitawareness of the connection between word forms and their spellings and, accordingly,might enhance students’ understanding of the structure of the language as well as theirability to spell. The current results certainly suggest that teaching children to make use

M. Senechal et al. / Journal of Experimental Child Psychology 95 (2006) 231–254 251

of the morphological relations among root words and their derivatives could lead to moreaccurate spelling. In addition, they suggest that teachers need to highlight for children theimportance of adaptive strategy use when words cannot be retrieved from memory (Rittle-Johnson & Siegler, 1999; Steffler et al., 1998). A complete definition of adaptive strategyuse would also include knowledge of the limits of certain strategies. For example, theuse of phonological strategies tends to be efficient with words that have regular pho-neme–grapheme correspondences but not so much with other word types. In the case ofFrench, reliance on phonology to spell is particularly problematic because morphologicalinformation often is unmarked orally.

Appendix

Spelling task for Experiment 1 and table of word characteristicsPhonological words: lac [lak], bocal [b ckal], journal [Zu nal], eclair [ekle ], canif [kanif],

tambour [tabu ].Morphological words: gras [g a], gratuit [g at�i], epais [epe], blond [bl~O], tannant [tana],

bavard [bava ], rang [ a], retard [ (E)ta ], habit [abi], debut [deby], repos [r(E)po], galop[galo].

Lexical words: tabac [taba], etang [eta], choix [Swa], esprit [esp i], noeud [nø], brebis[b Ebi].

Phonologicalwords

Morphologicalwords

Lexical words

p

Mean

SD Mean SD Mean SD

Number of letters

5.50 1.52 5.36 1.03 5.29 0.49 .94 Number of syllables 1.83 0.41 1.73 0.47 1.75 0.44 .88 Number of phonemes 4.83 0.98 3.91 1.14 3.57 1.13 .13 Word frequency 52.56 11.02 55.16 35.18 49.53 12.56 .90 Word frequency of

derivatives

40.65 12.85

Orthographicneighbors

2.83

4.17 3.18 2.99 1.29 2.56 .47

Phonologicalneighbors

8.00

12.08 13.09 9.03 16.86 14.60 .41

PO consistency

0.33 0.20 0.14 0.09 0.12 0.07 .02

Note. PO, phonology-to-orthography.

Spelling task for Experiment 2 and table of word characteristicsPhonological words: castor [kast c], divan [diva], moulin [mul~e], violon [vj cl~O], detour

[detu ], tresor [t ez c].Morphological words: bavard [bava ], marquis [ma ki], lourd [lu ], idiot [idjo], delicat

[delika], fort [f c], combat [k~Oba], sanglot [saglo], regard [ (E)ga ], transport [t asp c],poignard [pwa›a ], tapis [tapi].

Lexical words: foulard [fula ], crachat [k aSa], petard [peta ], escargot [eska go], effort[ef c], ananas [ananas].

252 M. Senechal et al. / Journal of Experimental Child Psychology 95 (2006) 231–254

Phonologicalwords

Morphologicalwords

Lexical words

p

Mean

SD Mean SD Mean SD

Number of letters

5.83 0.41 6.25 1.42 6.67 0.82 .45 Number of syllables 2.00 0.00 1.92 0.51 2.33 0.52 .20 Number of

phonemes

5.00 0.89 4.75 1.22 5.17 0.98 .73

Word frequency

51.90 6.32 48.17 7.09 46.67 14.05 .59 Word frequency

of derivatives

43.51 8.21

Orthographicneighbors

0.83

0.98 2.92 4.42 0.67 0.82 .29

Phonologicalneighbors

4.17

3.92 8.58 7.74 4.17 3.66 .24

PO consistency

0.23 0.14 0.19 0.05 0.20 0.07 .66

Note. PO, phonology-to-orthography.

Morphological awareness task for Experiment 2Practice items: 1. chien:chienne [Sj~e:Sj~en] 2. chat:(chatte) [Sa:Sat]; 1. content:contente

[k~Ota:k~Otat] 2. joyeux:(joyeuse) [Zwajø:Zwajøz]; 1. deux:deuxieme [dø:døzjem] 2. trios:(troi-sieme) [t wa:t wazjem].

Test items:

1. mechant:mechante [meSa:meSat]

2. croquant:(croquante) [k cka:k ckat] 1. rapide:rapidement [ apid: apidma] 2. heureux:(heureusement) [œ ø:œ øzma] 1. repos:reposer [ (E)po: (E)poze] 2. refus:(refuser) [ (E)fy: (E)fyze] 1. gris:grise [g i:g iz] 2. blond:(blonde) [bl~O:bl~Od] 1. cane:caneton [kan:kant~O] 2. rat:(raton) [ a: at~O] 1. danse:danser [das:dase] 2. saut:(sauter) [so:sote] 1. ecrit:ecriture [ek i:ek ity ] 2. toit:(toiture) [twa:twatyr] 1. canard:canardeau [kana :kana do] 2. renard:(renardeau) [ (E)na : (E)na do] 1. dos:dossier [do:dosje] 2. dent:dentier [da:datje] 1. grand:grandeur [g a:g adœ ] 2. lent:lenteur [la:latœ ] 1. sec:seche [sek:seS] 2. frais:fraıche [f e:f eS] 1. roux:rousse [ u: us] 2. doux:(douce) [du:dus]

References

Alegria, J., & Mousty, P. (1996). The development of spelling procedures in French-speaking, normal, andreading-disabled children: effects of frequency and lexicality. Journal of Experimental Child Psychology, 63,312–338.

Alegria, J., & Mousty, P. (1997). Lexical spelling processes in reading disabled French-speaking children. In C. A.Perfetti, L. Rieben, & M. Fayol (Eds.), Learning to spell: Research, theory, and practice across languages

(pp. 115–128). Mahwah, NJ: Lawrence Erlbaum.

M. Senechal et al. / Journal of Experimental Child Psychology 95 (2006) 231–254 253

Allal, L. (1997). Learning to spell in the classroom. In C. A. Perfetti, L. Rieben, & M. Fayol (Eds.), Learning to

spell: Research, theory, and practice across languages (pp. 129–150). Mahwah, NJ: Lawrence Erlbaum.Carlisle, J. F. (1987). The use of morphological knowledge in spelling derived forms by learning-disabled and

normal students. Annals of Dyslexia, 37, 90–108.Carlisle, J. F. (1988). Knowledge of derivational morphology and spelling ability in fourth, sixth, and eighth

graders. Applied Psycholinguistics, 9, 247–266.Carlisle, J. F. (1995). Morphological awareness and early reading achievement. In L. Feldman (Ed.),

Morphological aspects of language processing (pp. 189–209). Hillsdale, NJ: Lawrence Erlbaum.Carlisle, J. F. (1996). An exploratory study of morphological errors in children’s written stories. Reading and

Writing: An Interdisciplinary Journal, 8, 61–72.Casalis, S., & Louis-Alexandre, M.-F. (2000). Morphological analysis, phonological analysis, and learning to

read French: A longitudinal study. Reading and Writing: An Interdisciplinary Journal, 12, 303–335.Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum.Deacon, S. H., & Kirby, J. R. (2004). Morphological awareness: Just ‘‘more phonological’’? The roles of

morphological and phonological awareness in reading development. Applied Psycholinguistics, 25, 223–238.Ehri, L. C. (2005). Learning to read words: Theory, findings, and issues. Scientific Studies of Reading, 9, 167–188.Fischer, F. W., Shankweiler, D., & Liberman, I. Y. (1985). Spelling proficiency and sensitivity to word structure.

Journal of Memory and Language, 24, 423–441.Fowler, A. E., & Liberman, I. Y. (1995). The role of phonology and orthography in morphological awareness. In

L. Feldman (Ed.), Morphological aspects of language processing (pp. 57–188). Hillsdale, NJ: LawrenceErlbaum.

Gentry, J. R. (1982). An analysis of developmental spelling in GNYS AT WRK. The Reading Teacher, 36,192–200.

Goulandris, N. K. (1994). Teaching spelling: Bridging theory and practice. In G. D. A. Brown & N. C. Ellis(Eds.), Handbook of spelling: Theory, process, and intervention (pp. 408–423). New York: John Wiley.

Green, L., McCutchen, D., Schwiebert, C., Quinlan, T., Eva-Wood, A., & Juelis, J. (2003). Morphologicaldevelopment in children’s writing. Journal of Educational Psychology, 95, 752–761.

Lete, B., Sprenger-Charolles, L., & Cole, P. (2004). MANULEX: A grade-level lexical database from Frenchelementary school readers. Behavior Research Methods, Instruments, and Computers, 36, 156–166,<www.lexique.org/>.

Leybaert, J., & Alegria, J. (1995). Spelling development in deaf and hearing children: Evidence for use of morpho-phonological regularities in French. Reading and Writing: An Interdisciplinary Journal, 7, 89–109.

Mahony, D., Singson, M., & Mann, V. (2000). Reading ability and sensitivity to morphological relations. Reading

and Writing: An Interdisciplinary Journal, 12, 191–218.Nagy, W., Berninger, V., Abbott, R., Vaughan, K., & Vermeulen, K. (2003). Relationship of morphology and

other language skills to literacy skills in at-risk second-grade readers and at-risk fourth-grade writers. Journal

of Educational Psychology, 95, 730–742.New, B., Pallier, C., Ferrand, M., & Matos, R. (2001). Une base de donnees lexicales du francais contemporain

sur internet: LEXIQUE. L’Annee Psychologique, 101, 447–462, <www.lexique.org/>.Nunes, T., Bindman, M., & Bryant, P. (1997). Morphological spelling strategies: Developmental stages and

processes. Developmental Psychology, 33, 637–649.Nunes, T., Bryant, P., & Olsson, J. M. (2003). Learning morphological and phonological spelling rules: An

intervention study. Scientific Studies of Reading, 7, 289–307.Pacton, S., & Fayol, M. (2003). How do French children use morphosyntactic information when they spell

adverbs and present participles? Scientific Studies of Reading, 7, 273–287.Pacton, S., Fayol, M., & Perruchet, P. (2005). Children’s implicit learning of graphotactic and morphological

regularities. Child Development, 76, 324–339.Peereman, R., & Content, A. (1999). LEXOP: a lexical database providing orthography–phonology statistics for

French monosyllabic words. Behavior Research Methods, Instruments, and Computers, 31, 376–379, <http://leadserv.u-bourgogne.fr/bases/lexop/>.

Read, C. (1971). Preschool children’s knowledge of English phonology. Harvard Educational Review, 41,1–34.

Read, C. (1975). Children’s categorization of speech sounds in English (NCTE Research Report No. 17). Urbana,IL: National Council of Teachers of English.

Reichle, E. D., & Perfetti, C. A. (2003). Morphology in word identification: A word-experience model thataccounts for morpheme frequency effects. Scientific Studies of Reading, 7, 219–237.

254 M. Senechal et al. / Journal of Experimental Child Psychology 95 (2006) 231–254

Rittle-Johnson, B., & Siegler, R. S. (1999). Learning to spell: Variability, choice, and change in children’s strategyuse. Child Development, 70, 332–348.

Rubin, H., Patterson, P., & Kantor, M. (1991). Morphological development and writing ability in children andadults. Language, Speech, and Hearing Services in Schools, 22, 228–235.

Sarazin, G. (1995). Test de Rendement pour Francophones. Toronto: Psychological Corporation.Senechal, M. (2000). Morphological effects in children’s spelling of French words. Canadian Journal of

Experimental Psychology, 54, 76–85.Steffler, D. J., Varnhagen, C. K., Friesen, C. K., & Treiman, R. (1998). There’s more to children’s spelling than

the errors they make: Strategic and automatic processes for one-syllable words. Journal of Educational

Psychology, 90, 492–505.Templeton, S., & Morris, D. (1999). Questions teachers ask about spelling. Reading Research Quarterly, 34,

102–112.Totereau, C., Thevenin, M. G., & Fayol, M. (1997). The development of understanding of number morphology in

written French. In C. A. Perfetti, L. Rieben, & M. Fayol (Eds.), Learning to spell: Research, theory, and

practice across languages (pp. 96–114). Hillsdale, NJ: Lawrence Erlbaum Associates.Treiman, R., Berch, D., & Weatherston, S. (1993). Children’s use of phoneme–grapheme correspondences in

spelling: roles of position and stress. Journal of Educational Psychology, 85, 466–477.Treiman, R., Cassar, M., & Zukowski, A. (1994). What types of linguistic information do children use in spelling?

The case of flaps. Child Development, 65, 1318–1337.Turner, I. F., & Quinn, E. (1986). Learning English spellings: strategies employed by primary school boys.

Educational Psychology, 6, 231–241.Varnhagen, C. K. (1995). Children’s spelling strategies. In V. W. Berninger (Ed.), The varieties of orthographic

knowledge II: Relationships to phonology, reading, and writing (pp. 251–290). Dordrecht, Netherlands: KluwerAcademic.

Varnhagen, C. K., Boechler, P. M., & Steffler, D. J. (1999). Phonological and orthographic influences onchildren’s vowel spelling. Scientific Studies of Reading, 3, 363–379.

Waters, G. S., Bruck, M., & Malus-Abramowitz, M. (1988). The role of linguistic and visual information inspelling: A developmental study. Journal of Experimental Child Psychology, 45, 400–421.