Miscue Analysis

Miscue Story: Gary’s Reading

The transcript of this 46 sentence, 4-page story maintains the visible display of the text to consider the influence of text features on reading. Most of Gary’s most complex miscues in this reading happen in the first 12 lines (Figure 1). He read the title and first sentence slowly with careful enunciation and appropriate intonation.

OPEN IN VIEWER

Figure 1.

The beginning of Gary’s reading (marked typescript of Gary’s reading The Stonecutter lines 1-12) LINK: For Gary’s transcript of whole story with miscue markings (www.thosegoodmans.net click miscue samples then Gary’s marked typescript and retelling).

He split syllables in words like stonecut/ter (lines 1 and 2). That is not a miscue but we mark them to consider their influences on reading. In the second sentence, lines 3 and 4, Gary predicated building horses of … horses … self-correcting to houses and roads. In the next sentence, line 4, he read con- (a partial attempt) then $con’ tented emphasizing the first syllable which he corrected to contended. He substituted $tessels (a nonword) for tassels, read bright and $breathed (as in breath + /t/) a nonword, for breathed and made four attempts on stonecutter … $stonecarved … carve … stonecarver, and finally cut/ter. His partial attempts on words in this excerpt were all self-corrected, shown with a circled c.

Miscue analysis is both qualitative and quantitative. We analyze the quantity of miscue data and code quality; how reader’s miscues influence comprehension and show control of the reading process. We code the responses to several questions taking into consideration each miscue in the sentence. For different purposes, we have variations of questions and forms for marking and coding miscues (Y. Goodman, Flurkey, & Martens, 2014) including http://www.retrospectivemiscue.com.

In the Reading Miscue Inventory, classroom procedure (Y. Goodman, Watson, & Burke, 2005, p. 104): Question 1 is concerned with the degree to which the sentence as finally read by the reader is syntactically acceptable in the sentence and the story. Question 2 asks whether the sentence as finally read is semantically acceptable. Question 3 evaluates the degree to which meaning is changed and is only asked if the sentence, with all its miscues, is syntactically and semantically acceptable. Questions 4 and 5 evaluate the degree to which substitution miscues are graphically or phonologically similar. The transcript ends with a statistical summary (Figure 2).

OPEN IN VIEWER

Figure 2.

Statistical Profile of Gary’s reading of The Stonecutter p. 6.

Although Gary produced 8 miscues per 100 words in his reading, which might seem high, for evaluation purposes, it is necessary to consider that he sustained the grammatical structure with 91% of the sentences coded syntactically acceptable within the whole story. Eighty percent of his sentences were semantically acceptable. A hundred percent of the sentences coded syntactically and semantically acceptable retained the story’s meaning. A hundred percent of his substitution miscues showed high graphic similarity (horses/houses, were/where, and him/his) and 83%, high sound similarity ($shevery/shivery, washed/rushed) reflecting Gary’s knowledge of phonics. Gary was an effective reader with high percentages of acceptability and meaning change although he was overly cautious as shown by his high percentages of graphic and sound similarity.

Examining Gary’s miscues on the same word across the text also made his problem-solving capabilities visible. Contented occurs 3 times and Gary responded to each differently.

He self-corrected contented in its first occurrence (see Figure 1: Sentence #3, line 5) but made different miscues on subsequent occurrences indicating that he didn’t know contented and that he knew he didn’t know (Sentences #38, lines 58 and 59, and Sentence 46, line 74). Examining miscues on the same items across a text shows readers’ comprehending, the term that refers to how readers make sense during reading. Retelling represents the reader’s comprehension—the process of understanding after reading. Gary’s retelling, without an open book, was very complete including reference to almost every sentence in the story and he often rephrases as he retells trying to reconstruct the language of the text.

Miscue Analysis History and Foundational Principles

Miscue analysis research was developed at Wayne State University in Detroit, by Ken Goodman, in the late 1960’s. I was part of the team of graduate students working with him and designed the first of about 20 dissertations completed during that time. We called ourselves miscuteers, which really dates us. After I became a faculty member at the University of Michigan Dearborn, I continued my participation with the Miscue Center.

The Detroit education community provided an authentic laboratory in which to study reading. Teachers and administrators were concerned about African American and Appalachian children learning to read, and discussions often centered on their language variations. This was often a heated and legal issue at professional meetings of teacher educators and applied linguists.

During this time, linguists were calling for descriptive approaches to grammar, and teacher educators were recommending involving functional uses of language in English/ language arts curricula rather than out of context explicit grammar teaching. A few reading researchers were discussing the promise of recognizing reading as language and some teacher education programs began to include knowledge about linguistics, psycholinguistics, and sociolinguistics.

So Ken Goodman set out to study reading as a linguist would. He asked readers to read a whole story orally without interruptions and found when readers deviated from what would be expected, their deviations had linguistic characteristics related to the written text in sophisticated ways. He called these unexpected responses miscues rather than errors assuming that they resulted from the same linguistic cues and underlying processes that produced expected responses. What seems to be accurate reading are appropriate predictions by readers constructing a text in transaction with the written text.

Ken Goodman didn’t invent miscues, he discovered that they were always there. Clearly miscues happened in silent reading too but having readers read authentic whole texts orally is close to the reality of literacy events. Later, Ken Goodman realized miscue analysis was an example of scientific realism that examined real literacy events with the goal of developing a theory that explains their underlying structures and processes (K. Goodman, 2008).

Miscue analysis always involved whole authentic texts not written or rewritten for research purposes. Michael Halliday (1978), founder of systemic linguistics, provides theoretical support for this practice:

language comes to life only when functioning in some environment. We do not experience language in isolation—if we did we would not recognize it as language—(emphasis mine) but always in relation to a scenario, some background of persons and actions and events from which the things which are said derive their meaning. (p. 28)

Miscue researchers see language as taking place in several contexts as linguists do. There is a literacy event. In the event, there is a written text with readers and authors transacting with the text. The written text has structure or grammar, which makes it possible to represent meaning. Miscues are analyzed within the context of the sentence and the sentence within the context of the whole text.

As miscue analysis developed, it became obvious that all language aspects needed to be considered simultaneously to determine influences on each miscue. It’s not possible to separate single, simple cause effect relationships. Miscue analysis considers three language levels simultaneously: graphophonic, syntactic wording, and semantic based on Halliday’s systemic linguistics (1985). In order to understand the influence of grammar on reading, we analyzed every word in the texts we used with readers based on Charles Fries’ descriptive grammar (1952). We developed and continue to refine a taxonomy of questions to examine readers’ miscues (See Goodman Taxonomy in Y. Goodman et al., 2005, pp. 279–284).

In 1971, Carolyn Burke and I published the Reading Miscue Inventory (Y. Goodman & Burke, 1971) to make miscue analysis accessible for teachers. Dorothy Watson joined our author team as we continued to update our work (Y. Goodman et al., 2005; Y. Goodman, Watson, & Burke, 1987). Teachers who learn and engage in miscue analysis expand their knowledge of linguistics and build insights into how readers make sense of written language. Miscue analysis has been criticized because it requires considerable knowledge about language. But shouldn’t every researcher who studies reading and every teacher or clinician who is responsible to evaluate the language of their students have strong backgrounds in language knowledge?

Miscue Lesson: A Single Reading Process

Miscue research led to the conclusion that there is a single reading process that involves the brain selecting information from the three language levels (graphophonic, syntactic wording, and semantic) to construct meaning. Variations within this process are due to social, cultural, economic, linguistic, biological, historical, emotional, and instructional influences. However, there is only one way of making sense of print. All readers use the same universal reading strategies: sampling, predicting, confirming, and inferring as they transact with a written text to make sense of their reading and produce their own internal text. These conclusions are supported by miscue analysis studies of a range of different languages including American Sign Language (Ewoldt, 1981; Gennaoui & Chaleff, 2000) and those written in nonalphabetic orthographies (K. Goodman, Wang, Ivantosch, & Goodman, 2012). Proficient readers don’t read differently than less proficient readers, they are more flexible and efficient.

A number of colleagues are using eye-tracking research in combination with miscue analysis (eye movement miscue analysis [EMMA]) expanding on Eric Paulson’s (2000) groundbreaking research exploring the concept of a single reading process <emmaforum.org>. For over a hundred years, literature has shown that readers fixate on about two-thirds of the words of the text (Bartlett, 1932; Huey, 1908).

Our comprehensive reading model makes clear why this happens. Readers only fixate enough to confirm their predictions and then make new ones. Readers’ eyes fixate on content words about twice the rate of their fixations on function words, which provides evidence for readers’ intuitive knowledge of grammar. The eye only provides the brain with information at the points of fixation. EMMA research shows that readers fixate on words they don’t say and say words they don’t fixate. The brain is at work telling the eye where to look, making sense of its perceptions and predicting what should be there.

An example (Figure 3) from Eric Paulson’s research (K. Goodman et al., in press) shows eye movements on one sentence from a story being read orally by a young adult: He reads: “Take polenta, the cornmeal mush often served with beans and sausage floating in it.” No miscues! To hear that read orally you would think the reader has fixated on each word in order.

OPEN IN VIEWER

Figure 3.

A reader’s fixations and saccades on one sentence.

But the eye tracker shows a very different process. His eye fixations are shown by the dots and the sequence of fixations by the numbers written below each fixation; and although fixation durations vary considerably, a fixation is seldom longer than a third of a second and can be much shorter. The reader fixates first on the determiner the as he reads the sentence orally. The saccades, the lines connecting the dots, show where the eye moves but it does not take in any information as it moves. The words the reader fixated and the order in which they were fixated are: the polenta the cornmeal often beans beans sausage beans sausage floating sausage floating floating in. His eye never fixates on take, mush, served, with, and, it.

The brain controls where the eye looks and for how long and the mouth reports what the reader is already predicting while comprehending the text. Readers do not read text as a series of words but construct their own text often similar to the published text. The mouth is saying what the brain has constructed—which is not the same as what the eye has seen. EMMA research supports what we’ve learned about reading through miscue analysis and extends our understandings about the role of the brain.

This has established another theoretical conclusion: The structures and processes of silent and oral reading are essentially the same. Eye movement patterns of silent and oral reading provide this evidence and are also supported by EMMA studies that compare oral and silent reading (O’Brien de Ramirez, 2008). As a result, I encourage teachers and tutors to spend instructional reading time on silent reading and provide time and opportunities for silent reading in their curriculum. Oral reading can be used for dramatic presentations. Those interested in comparing silent and oral reading or online reading, find EMMA research useful.

Miscue Lesson: Confirming Strategies Through Self-Correction

Another conclusion relates to self-correction strategies. Although all readers use the same process to make sense of print, they vary considerably in how tentative they are especially in relation to predicting, confirming, and self-correcting. All texts vary in complexity and density as individual readers travel through them, which supports Alan Flurkey’s concept of reading as flow (Flurkey, 2008). The text is like a river that ebbs and flows rather than like water flowing through a smooth pipe. Reading slows and speeds up as water does in a river.

The concept of accuracy and fluency is inadequate as a metaphor and counterproductive. It makes getting the words right smoothly more important than making sense. It separates the role of correction from the reader’s use of complex relationships between predicting and confirming that provides readers with information to decide when self-correction is necessary. Readers concerned with accuracy and fluency are often ineffective and inefficient because they are distracted from making sense. What I’ve said may seem logical but you still may not be comfortable with the strong belief I’ve concluded from more than 50 years of miscue research that accuracy in reading is not a useful goal.

It is inefficient for all miscues to be corrected. Actually in our large database studies, no group of readers reading a whole text overtly self-corrected more than 40% of their miscues. Decisions about when to correct is related to how miscues influence the acceptability of the sentence. When readers make miscues that are fully acceptable in a sentence, they are usually unaware of those miscues. They confirm their miscues silently (no need for correction) and continue reading. This is especially true for proficient readers.

On the other hand, the highest percentages of self-corrected miscues are acceptable syntactically and semantically with only the preceding portion of the sentence but not with what follows. Such miscues cause readers disequilibrium and are usually self-corrected. Most of Gary’s miscues involved such predictions.

Readers disconfirm and self-correct such miscues to a greater extent than others in fractions of seconds. I repeat, self-correction strategies are rarely for the purpose of accuracy nor do they need to be. They are for the purpose of the reader constructing a meaningful text.

Miscue Lesson: Grammar—Reader’s Intuitive Knowledge About Language

Another foundational principle relates to grammar. The role of grammar in language processes including reading is probably the least understood and a most neglected aspect of reading. Since grammar is the rule-governed system through which meaning is conveyed in language, readers and listeners must assign a grammatical structure to each linguistic unit to get to meaning. The intonation of young readers reveal the grammatical structures they are assigning and their tentativeness in reading during oral reading when they produce a rising intonation that seems to ask: “did I get that right?” Intonation patterns confirm readers’ intuitive knowledge of grammar. Young children have been assigning appropriate grammatical structures with spoken language long before they come to school.

Readers revealed their grammatical knowledge through their miscues. For example, the youngest to the oldest (2nd to 10th graders) in the large data-based studies substituted nouns for nouns and verbs for verbs from 70% to 95% of the time. They even produced nonwords with grammatical markers, the ings and eds and other inflections such as Gary’s substitutions of $tessels for tassels and $breath+t for breathed (K. Goodman & Burke, 1973).

An analysis of fourth grade Arabic readers learning English as a second language in Dearborn (K. Goodman & Goodman, 1978) also showed readers’ developing grammatical knowledge. In a folk tale they read in English, variant forms of the word plow occur 31 times as nouns, verbs, and adjectives. Plow was read appropriately only 12 times. However, even when they produced nonwords such as the $palowing field for the plowing field, $paloe handles for plow handles, and $blowed the field for plowed, their substitutions revealed their grammatical knowledge of English.

A study of determiner miscues documented sophisticated use of grammar by second language and rural dialect readers and showed that: “All readers reflected functional control of the determiner system of English in their miscues” (K. Goodman, 1987, p. 54). Readers’ miscues on determiners were most often semantically and syntactically acceptable at every age level. Miscues on a/an, the, that, this, these, those, his, her were almost always substituted for other determiners. These substitutions revealed readers’ appropriate uses of definite and indefinite articles and their understandings of the English noun system. Although 7- to 10-year-old readers are rarely able to explain differences between definite and indefinite noun phrases, their miscues show their intuitive knowledge which supports another foundational principle: reading is language and grammar is learned in the context of its use.

Becoming aware of grammatical knowledge from readers’ miscues provides opportunities for teachers to revalue their students as competent. I developed a great deal of knowledge about linguistics through miscue analysis and recommend that literacy teacher education should include knowledge and understandings from linguistics about the grammatical system in reading.

Miscue Story—Teacher Education and Professional Development

A major focus for miscue research has been to gain insight into the reading process, and another goal has been to help teachers expand their knowledge about the teaching and learning of reading and language to support students’ reading development. We received a 3-year federal grant to provide professional development for a school district in Southern California (Cajon Valley Union School District, 1974). Carolyn Burke and I provided weekend workshops and summer programs for teachers in upper elementary grades.

We prepared engagements for teachers to explore the reading process and they used our newly published Reading Miscue Inventory (Y. Goodman & Burke, 1971). The teachers conducted miscue analysis on their own and their students’ reading and documented their growth. They developed insights into the language and background knowledge their students brought to their reading. Their instructional focus was on their ongoing curriculum, which they planned to involve students in collaboratively selected theme cycles. Both reading and writing were used and learned across the curriculum.

This project was rated an exemplary dissemination project in the state of California. As a result of the project, teachers transformed their beliefs about reading and the focus of their instruction. Students showed increase in miscue quality, predictions, and self-correction patterns. During the 3 years, targeted students identified in the fourth grade by standardized tests as being at least a half-year below grade level had a mean growth each year in excess of 1 year in reading comprehension. Some students made 3 or 4 years gain in the first 2 years.

When Carolyn and I conducted our own final interviews with the teachers, they agreed on aspects of their teaching that had the most impact on their students’ growth.

Encourage students to focus on making sense in everything they read.

We continue being involved in professional development and design research to evaluate the programs. Our purpose is to provide miscue analysis opportunities for teachers to discover their students’ strengths and consider the support they need. In such contexts, teachers take control of their own teaching as they develop knowledge about how teaching reading best supports learning to read.

Recently, Ken Goodman and I consulted with secondary English teachers in the Aurora Public Schools in Colorado using miscue and RMA (see below) with their students. Their reading/writing workshop classes involved students using tablets at home and school and regular reading and writing conferences with their teachers learning about their own literacy processes. Research conducted on this project resulted in transforming both teachers’ and students’ views about literacy. District wide evaluation highlighted the growth of teachers and students. One teacher reported on her student’s changes (Barnhouse & Butler, 2014):

This has changed his life, because he never thought he could read, he never liked reading, he wouldn’t read, he wouldn’t write, and this whole year process has changed his attitude about reading. He now thinks, “I can do this. I can read a grade level piece. I can have something to say in a conversation.”

RMA—Thinking and Talking About Reading

I developed RMA to formalize the success teachers had when they discussed miscue analysis and the reading process with their students (Y. Goodman et al., 2014). Research on RMA has shown considerable success with vulnerable readers (Bomer, 1999; Jaeger, 2015). In initial interviews, these readers articulate that they are not good readers: They are dyslexic, have an attention deficit disorder, can’t sound out words, don’t know vocabulary, don’t remember what they read, read too slowly, read too fast, don’t look up words in the dictionary, and can’t spell either. For them, RMA is liberating as well as transforming. As they discuss their miscues with others, they learn that reading behaviors that prove they are not good readers are the same behaviors that proficient readers use. As they discover their capabilities, they realize that as they read, they are in control of their own meaning making (K. Goodman, 2014a).

RMA does not teach new ways of reading, rather it engages learners in discovering what they already do as readers as literate members of a society. They respond positively to teachers they believe are truly interested in them and are liberated from instruction dependency (Board, 1982). I received an National Council of Teachers of English Research Foundation grant to study the metalinguistic knowledge that readers demonstrate during their discussions of their miscues (Y. Goodman & Flurkey, 1996). The participants were seventh graders with stanine scores of 2 to 7. Doctoral students and I met individually and regularly with them throughout one school year. Sessions were tape recorded, transcribed and analyzed, and we learned a lot.

All readers talked seriously about reading and were interested in what they were discovering. At the beginning, their views about themselves reflected instructional views of reading and negative views about themselves as readers, and this was even true for many readers in the higher stanine groups. Throughout the year, their language use and knowledge about reading, authors, and texts became more informed as they discussed why they made specific miscues. Rolando, with the lowest stanine score, was able to defend his miscues.

We documented how these students came to appreciate their own capabilities through RMA discussions. They built confidence in their control over reading and in their reading development. This was true for both proficient and less proficient readers. Bernice, in the seventh stanine in reading, initially believed that a good reader reads every word on a page correctly and does not stumble on words (Costello, 1996, p. 132). During her final interview when asked what she would do to help someone who was having trouble reading, she said: “I would kind of make them try to figure it out.… take the time to let them discover for themselves” (p. 140). She also agreed that she had improved as a reader and then stated confidently, “Oh I’ll read forever” (p. 141).

It became clear that many higher scoring readers like Bernice also lacked confidence in their own abilities and in our early conversations, often commented on what really good readers could do that they couldn’t. Even with audiences of reading researchers, when I ask how many believe they are excellent readers, not all hands go up. RMA provided opportunities for proficient readers to revalue their own literacy identities, which has important influences on their emotional responses as readers. Perhaps this is related to students taking their test results too seriously. As we learned in previous miscue studies, this study showed that test scores underestimated our participants reading competence.

A number of colleagues have used RMA successfully with various adult reader populations (Bacon, 2014; Paulson & Mason-Egan, 2007; Wurr, Theurer, & Kim, 2008). Studies of highly proficient Korean readers, learning English as a second language (Chin, 1996; Kim, 2010), demonstrated how RMA helped them discover that their competence in reading their first language could be used as a foundation to support their development of English literacy. Literacy Volunteers of Tucson tutors have been using RMA successfully with adult learners not well served by schooling. Through RMA, they become liberated and discover how literate they already are.

Dialect Features in Reading

My dissertation research was a 2-year longitudinal study documenting the reading development of six African American readers in Highland Park (Y. Goodman, 1967). I continued collecting and analyzing their readings and retellings for 7 years (Y. Goodman, 1971). Three of the children were labeled “at risk” in a prefirst grade. The others were average first grade readers. Having a rich longitudinal database made it possible for me to attend to the readers’ African American Vernacular English (AAVE). Debra Goodman analyzed 7 years of dialect features in the students’ readings and retellings. We selected the sites where AAVE dialect occurs and calculated the percentage of the readers’ dialect features (Y. Goodman, & Goodman, 2014). The miscue lessons we learned expose misconceptions about dialect in reading.

In second grade, Antonio read The Three Billy Goat’s Gruff (1966) in which he produced more dialect features than in any other of his readings. The second text sentence, which appeared 3 times in the story, is especially noteworthy:

During his first reading of the sentence, Antonio read slowly enunciating each word, stressing /z/ at the end of Who’s, // at the end of tripping, and so on. He moved from virtually no dialect readings to more as he read each subsequent sentence. The third time he read it eagerly with almost every possible dialect feature marked: “Who dat trip trappin’ ova’ mah bridge, roar’ the tro’” and after, he related the story with excitement: “He (the troll) gonna get smacked right outta the way, when the billy goat just poked his eyeballs out with his horns.”

A few years later, when the average readers including Antonio were in sixth grade, they read a story called Sheep Dog (Stovall, 1966), which revealed additional insights about dialect. Sheep Dog describes twenty four hours in the dog’s life left in charge of a herd of sheep. There are two scuffles when the dog saves the sheep from a pair of coyotes. All three readers said the fight scenes were their favorites although Antonio and Foster were much more enthusiastic than Beth. The males had considerably more dialect features in reading the fight scenes than in any other section of the story while Beth’s percentage of dialect features was actually greater for the text portions excluding the fight scene. We concluded that use of dialect features during oral reading is evidence of comprehension (Y. Goodman & Goodman, 2014).

Rejection of a readers’ dialect by teachers who insist readers enunciate endings and shift vowel sounds to match the teacher’s dialect undermines reading development. It confuses readers about what is successful reading (K. Goodman & Buck, 1973). Difference is still too often treated as deficiency (Adger, Wolfram, & Christian, 2007).

Dialect features are not consistent, and variations are evident throughout the analysis. Franklin, the reader in my study (1971) with the highest percentage of dialect features in only one of his readings reached 39%, which is high compared to his other readings and to the dialect features of the other readers. Language is influenced by the context in which it occurs and readers’ language variations reflect their sensitivity to social contexts (Rigg, 1974).

Dialect variations are not considered miscues. Rather, we consider them the reader’s expected responses to the text. We mark these language events on reader’s transcripts to discuss their implications with parents and reading professionals. Teachers need to be well versed in understanding dialect and other variations of language use in the communities in which they work. With language knowledge, teachers share such knowledge with students in order for them to appreciate that all language users shift language appropriately based on social contexts.

Final Considerations

My miscue analysis adventures and the lessons I’ve learned have greatly impacted my career in literacy research and teacher education. I get excited whenever I have the opportunity to engage in yet another miscue analysis. The more miscue phenomenon is understood, the more likely it is that teachers will involve students with authentic literacy in the context of powerful learning experiences and not focus on sequencing and direct teaching of out of context abstract units of language.

I believe that every reading professional should engage personally in miscue analysis and organize opportunities with teachers, students, parents, and readers to do miscue analysis. When I interview students and teachers in classes or workshops, it is rare for them to report having listened intently to an oral reading of a whole story without interrupting the reader with corrections and prompts. Their evaluations of readers are based on what they believe they hear readers do rather than a thorough analysis of an actual recorded reading event.

Reading professionals and teacher educators are as concerned as I am about the numbers of readers that continue to suffer as a result of archaic, mandated instructional practices. The more knowledgeable teachers, parents, and researchers are about the reading process, the more likely it is that literacy instruction in classrooms will engage learners in the richness and joy of reading and writing across the curriculum.

My motivation, the motivation of those who engage in miscue analysis is as Brian Cambourne suggests, to make learning to read barrier free (K. Goodman, 2014b).