Rhetoric and Revolution
Kenneth Goodman's "Psycholinguistic Guessing Game"

Martin A. Kozloff
Watson School of Education
University of North Carolina at Wilmington
February, 2002



Erect your schemes with as much method and skill as you please;
yet if the materials be nothing but dirt, spun out of your own entrails
(the guts of modern brains), the edifice will conclude at last in a cobweb,
the duration of which, like that of other spiders' webs, may be imputed to
their being forgotten, or neglected, or hid in a corner. [Jonathan Swift.
The battle of the books. 1704]

Promoters and teachers of whole language argue that:

1. Whole language is more effective than other forms of reading instruction.

2. This alleged superiority reflects specific features of whole language; e.g.,

a. "Implicit" instruction that is less focused on precise learning objectives, involves less teacher direction, and requires students to construct knowledge of phonic and spelling rules (Goodman, 1986).

b. Much instruction on specific skills (e.g., phonics)  is given as-needed, during mini-lessons.

c. There is an emphasis on learning in what are called "authentic contexts"; e.g., learning phonics (which sounds go with which letters) and vocabulary during independent reading and when watching and listening to the teacher read books (Smith, 1985).

3. These design features are said to flow from a more adequate understanding of language and reading (Daniels, Zemelman, & Bizar, 2000; Powell & Hornsby, 1993).

However, recent research on reading, and experimental assessments of whole language challenge the claim of greater effectiveness.  Specifically,

1. Controlled longitudinal experimental research shows that instruction on phonemic awareness, decoding, reading fluency, spelling, and comprehension that focuses on specific skills, involves explicit communication of rules and strategies by the teacher,  precisely and logically sequenced, and provides systematic distributed practice is reliably superior for a wider range of students than implicit (less focused) instruction that requires students to construct their own knowledge (Foorman et al., 1998; Fletcher & Lyon, 1998; Gough, 1993; Liberman, 1999; National Reading Panel, 2000).

2. Evaluation research at state and county levels shows that achievement of students taught with whole language and Reading Recovery--the remedial branch of whole language—is not as high as claimed by whole language proponents and is less reliably effective than instruction provided by field tested curricula involving focused, teacher-directed instruction (Chapman, Tunmer, & Prochnow, 1999; Heibert, 1995; San Diego Unified School District, 1999; Stahl, McKenna, & Miller, 1994).

It is important as well to examine the conceptual apparatus of whole language.  What assumptions are made?  How is reading understood?  How are methods of assessment and instruction derived from the conceptualization of reading?  If the assumptions and/or conceptualization of reading are flawed, then whole language assessment and instruction derived from a flawed foundation are likely themselves to be flawed.  If so, this may help to explain the (at best) uneven effectiveness of whole language.

Goodman's Guessing Game

...when a man's fancy gets astride on his reason, when imagination is at cuffs with his senses,
and common understanding as well as common sense, is kicked out of doors, the first proselyte
he makes is himself... [Jonathan Swift. A tale of a tub. 1704]

Whole language proponents cite Kenneth Goodman's 1967 paper ("Reading: A psycholinguistic guessing game") as one of the first in their canon--the paper that fostered the whole language movement, or revolution (Goodman, 1976) and continues to guide and legitimize whole language activities (Pappas & Pettegrew, 1998).  Goodman clearly saw the paper the same way--as offering "a more viable scientific alternative" to what he dubbed "pre-existing, naive, common sense notions" about reading that "interfere with the application of modern scientific concepts of language and thought to research on reading" (Goodman, 1967, p. 126).  Let us take Goodman at his word.  Let us examine his "more viable scientific alternative" to see how he crafted a new foundation for reading research and instruction; to determine whether it satisfies the criteria for a viable or even scientific alternative; and to understand better how his ideas were so easily accepted and spawned the whole language movement.

In simplest terms, Goodman presents a conception of reading as a guessing game.  He provides no logical, empirical, or commonsensical support for this conception.  He then presents a highly selective set of passage misreadings by a child.  ONE CHILD.  These misreadings are not called errors; they are "miscues."  These misreadings are interpreted in a way that fits Goodman's guessing-game formulation (although other interpretations--from the phonic and word centered approaches he disparages--are more obvious and more reasonable).  Goodman then uses the misreading examples as verification of his conception of reading--although the only credible use of the examples would be a demonstration that it  is possible to misinterpret misreadings that way.  The paper ends by suggesting that the implication for instruction is teaching children to play the guessing game more skillfully.  Following is a closer examination of the logical structure of Goodman's paper.

The Opening Gambit

Goodman's paper begins with a common rhetorical device--caricature of a self-created adversary.  Specifically, he creates a false binary opposition of then current conceptions of reading and their associated methods of teaching:  "phonic centered" and "word centered."  He reduces these approaches to a few statements that would lead readers to agree with Goodman that these conceptions are simplistic and must be wrong.   For example,

 ...the common sense notion I seek here to refute is this: 'Reading is a precise process.  It involves exact, detailed, sequential perception and identification of letters, words, spelling patterns and larger language  units.'  In phonic centered approaches to reading, the preoccupation is with precise letter identification.  In word centered approaches, the focus is on word identification... (p. 126).

Goodman then writes, "In place of this misconception, I offer this..."--his allegedly "more viable scientific alternative" foreshadowed in the paper's abstract.

Note the artful way that Goodman sets up the reader.

1. He labels in a disparaging way the phonic and word centered approaches "common sense notions," despite the great deal of scientific research done in support of each one--especially the approach that advocated teaching phonics in a systematic way during beginning reading.  Yet, he does not cite this research or even hint that there was any.  These approaches are not presented as bodies of knowledge that may have some flaws.  Rather, in contrast to his self-valorized "scientific alternative," readers are simply to take Goodman's unsupported word and consider them mere common sense notions.

2. In contrast to standard practice in science, Goodman presents no data that the phonic and word centered approaches do not work.  He conducts no experiments--indeed, he cites no research at all--showing that whole language instruction (derived from his guessing game formulation of reading) is more effective than the phonic centered and word centered approaches he wishes to replace.  And, although he calls them "misconceptions," he does not analyze the intellectual apparatus behind the phonic centered and word centered approaches (e.g., their theories of reading) to show they are logically flawed.

In other words, Goodman does nothing to (in his own words) "refute" these common sense notions.  His only claim to readers' attention--and the only warrant for his "scientific" alternative—is an unsubstantiated opening pitch that there are two pre-existing alternatives; that these alternatives are merely common sense notions; and that they are misconceptions.

The New Model

Goodman then presents his "scientific" alternative.

...I offer this: "Reading is a selective process.  It involves partial use of available minimal language cues [Notice that he does not call them "reading" cues.  This makes it possible to falsely conflate reading and language, so that he can argue for teaching reading the same--incidental, "natural"--way that language is ordinarily learned. MK] selected from perceptual input on the basis of the reader's expectation.  As this partial information is processed, tentative decisions are made to be confirmed, rejected, or refined as reading progresses."  More simply stated, reading is a psycholinguistic guessing game.  It involves an interaction between thought and language.  Efficient reading does not result from precise perception and identification of all the elements, but from skill in selecting the fewest, most productive cues necessary to produce guesses which are right the first time.  The ability to anticipate that which will be seen, of course, is vital in reading, just as the ability to anticipate what has not yet been heard is vital in listening (pp. 127-8).

Notice that there are at least three logical errors in Goodman's opening presentation of his new and "scientific" approach to reading.  First,  Goodman's new view of reading rests on the fallacy of reification.  He transforms what is merely a metaphor into a concrete reality.  Goodman does not say that reading can (metaphorically) be seen (for purposes of analysis) as if it were a psycholinguistic guessing game.  Rather, reading is a psycholinguistic guessing game.  It is not as if readers are guessing at what words say.  Readers actually are guessing.  [The author--MK--asks readers of this document to consider whether they are in fact guessing at every word on this page.]  However, treating a metaphor as a concrete reality is a useful trick.  It means that whole language rests on a fantasy--a dreamy way of thinking--in which there is no boundary between how we think about things and how things actually are.  Once new teachers are seduced into this dream world, almost any bizarre and baseless statements can be taken as sage wisdom.  [See http://people.uncw.edu/kozloffm/wlquotes.html  for examples of whole language statements that have no basis in reality.]

Second, Goodman commits the fallacy of hyperbole, or over-generalization.  He does not say that reading is a psycholinguistic guessing game for certain aspects of reading, for some readers, at some point in their learning to read.  Rather, all of reading is a guessing game for all readers all the time.  This rhetorical device enables Goodman to lay claim to all of reading and reading instruction (word recognition, spelling, writing)--and to call it whole language.

Third, Goodman's definition of reading commits the fallacy of tautology.  After stating (above) that reading is a psycholinguistic guessing game, Goodman states, "It involves an interaction between thought and language."  [It is interesting that Goodman's definition of reading does not even have the word print in it.]  Apparently, the statement ("It involves an interaction between thought and language.") is supposed to support the idea that reading is a psycholinguistic guessing game.  However, interaction between thought and language means exactly the same thing as psycholinguistic.  Therefore, all Goodman is saying  that reading is a guessing game that involves interaction between thought and language. [Again, nothing about print.] However, all thinking uses language.  Thinking means talking to yourself--in a language.  So, what could an interaction between thinking (which means talking to yourself) and language mean?  In summary, Goodman's new definition of reading is  empty.  It means nothing at all, and it certainly has nothing to do with interacting with the printed word--the ordinary conception of reading.

Goodman's full conception consists of the following propositions--taken from his initial statement (above) and from the summary of his "model" at the end of the paper.

1. "Efficient reading does not result from precise perception and identification of all the elements."

2. Reading "involves an interaction between thought and language."

3. "Reading is a selective process."

4. This selecting process "involves partial use of available minimal language cues..."

5. Efficient reading results "from skill in selecting the fewest, most productive cues..."

6. These cues are at first graphic cues (p. 135).

7. These cues are "selected from perceptual input on the basis of the reader's expectation."  They are "guided by constraints set up through prior choices, his language knowledge, his cognitive styles and strategies he has learned" (p. 135).

8. These cues provide "partial information."

9. The reader "forms a perceptual image using these cues and his anticipated cues"  (p. 135).

10. The reader "searches his memory for related syntactic, semantic, and phonological cues."

11. This memory search "may lead to selection of more graphic cues and to reforming the perceptual image" (p. 135).

12. These cues are "necessary to produce guesses which are right the first time."

13. The reader then "makes a guess or tentative choice consistent with graphic cues.  Semantic analysis leads to partial decoding as far as possible" (p. 135).

14. This partial information " is processed, tentative decisions are made to be confirmed, rejected, or refined as reading progresses."

15. "If no guess is  possible, he checks the recalled perceptual input and tries again" (p. 135).

16. "If a guess  is still not possible, he takes another look at the text to gather more graphic cues" (p. 135).

17. "If the tentative choice is not acceptable semantically or syntactically, then he regresses, scanning from right to left along the line and up the page to locate a point of semantic or syntactic inconsistency" (p. 135).

18.  "When such a point (semantic or syntactic inconsistency.  MK) is found, he starts over at that point" (p. 135).

19. "If no inconsistency can be identified, he reads on seeking some cue which will make it possible to reconcile the anamolous (sic) situation" (p. 135).

20. "If the choice is acceptable, decoding is extended, meaning is assimilated with prior meaning and prior meaning is accommodated, if necessary" (p. 135).

21. "Then the cycle continues" (p. 135).

22. The above propositions enable one to see reading as a "psycholinguistic guessing game."

Rhetorical Devices and More Logical Fallacies in Goodman's Guessing Game

Goodman's new conception of reading is unsatisfactory in several ways.

It is speculation, not science.
        A defining feature of science (in contrast to metaphysics, opinion, fantasy, and madness) is that propositions, arguments, theories, and conceptual schemes are judged viable and scientific not because proponents say so, but on the basic of empirical evidence and sound reasoning.  Science also requires that writers define terms--especially when terms are new or may be misunderstood.  However, Goodman's version of science--at least in his article--appears not to require any empirical evidence or effort at clear definition.  He offers no data whatever to support his assertions that, for example,

1. Reading does not result from "precise perception and identification of all the elements."

2. Readers "select" "productive cues," and then guess at what words say and mean.

3. "Readers utilize not one, but three kinds of information simultaneously" (p. 131).

Nor does he explicate the meaning of "cue," "guess," "thought," "language," or even "reading."

        The absence of evidence and clear definition weakens Goodman's claim that he offers a viable scientific alternative conception of reading. Still, Goodman managed to help fashion a new definition of science--a science with neither data nor reasoning nor defined concepts, a science indistinguishable from speculation and wishful thinking.  However, this revisionist science well-served later whole language teachers, writers, researchers, and advocates who (guided by Goodman) no longer felt obliged to abide by--or even accept the legitimacy of--traditional scientific rules about external verification of claims via tests available to other persons (Moats, 2000).  In the absence of empirical evidence, we can only assess the adequacy of Goodman's new guessing game conception of reading by examining the logical adequacy of his propositions, as shown below.

Goodman's conception of reading commits the fallacy of hasty generalization, or converse accident.
        Goodman’s paper implies that his new conception embraces all of reading.  He does not say that only certain elements of reading, at some times, for some readers are part of a guessing game.  Rather, "(R)eading is a psycholinguistic guessing game." It is for all readers a process of selecting cues, and then guessing, confirming, rejecting, or refining tentative decisions about what sounds letters make, what a word says and means, what a period and comma imply, how words are spelled.  However, such guessing, cue selecting, and decision making arguably apply only to (1) beginning readers; (2) older readers who have not been taught to read and understand text based on solid knowledge (and the automatic application) of sound/symbol correspondence, punctuation, spelling, subject/predicate, cause/effect, and so forth; or (3) skilled readers who have run into a new and difficult word.  Consider propositions 13-21.  Is it reasonable to assert that these activities apply to all readers?  Is there any evidence that skilled readers guess at every word--as if reading (fluent reading) were a series of tentative choices?

Another example of hasty generalization  Goodman's use of reading errors--called "miscues"--as the only evidence that all reading is guessing.  Goodman’s paper does not provide samples of fluent reading to substantiate his propositions about selecting and guessing. This may be because fluent reading provides no evidence of guessing.  In summary, it is likely that Goodman's guessing game conception of reading applies only to poor readers, beginning readers, or good readers who are decoding unfamiliar words.  In other words, all that is new in Goodman’s new conception is the unwarranted generalization that all readers guess all the time.

The massive irony, here, is that Goodman's followers created a method of reading instruction--whole language--that reversed the polarity of guessing.  Rather than something to be overcome because it signified lack of skill, guessing was now considered a natural and good thing, and therefore was to be encouraged.  Systematic instruction on phonemic awareness, sound/symbol relationships (m says mmm), word attack, and spelling was now unnatural--a bad thing to be discouraged. Whole language teachers therefore explicitly and systematically taught new readers the guessing strategy used by poor readers for making errors, and called it fine.

Goodman's conception of reading as a guessing game commits the fallacy of reification, or hypostatization.
        In other words, Goodman treats abstract terms ("reconcile the anamolous [sic] situation," assimilation, accommodation) and metaphoric fictions ("searches his memory for cues," "he checks the recalled perceptual input") as if they were concrete objects or events (Thompson, 1995). Recall that Goodman's new formulation hinges on rejection of the "common sense" notions that (1) reading involves an almost instantaneous recognition of whole words, or (2) reading involves an almost automatic "perception and identification of letters, words..."  Note that whole word and phonic processes are ordinary, readily observable, mundane actions.  The reader sees and properly or improperly identifies letters and words.  Most observable identification errors have straightforward, ordinary, mundane implications for instruction; e.g., at sounding out words. But Goodman will offer nothing attractive to potential followers unless he conjures a radical shift of reading from the mundane to the esoteric.  Something as commonsensical as mere skill instruction will not do.  Henceforth, reading processes and reading instruction will no longer be easily seeable and teachable.  Instead, reading processes will be located in the mind: reading will involve "an interaction between thought and language."  Goodman now invents a mental apparatus to account for reading skill and error--the psycholinguistic guessing game--and it consists of selecting, deciding, guessing, confirming, rejecting, and refining.

There are two logical problems with Goodman's reified mental guessing game apparatus.  First, in contrast to what we ordinarily expect of a viable scientific account, there is no way to test whether Goodman's hypothesized mental apparatus exists at all---i.e., whether readers in fact perform the elaborate guessing routine--or whether the hypothesized apparatus operates just as Goodman proposes.  After all, many models of thought processes can be generated to account for the same reading behavior--just as demonic possession once provided a coherent account of psychiatric symptoms.

Second, Goodman transforms similes and metaphors (as-if) into objects--thought processes. However, all anyone (with a scientific orientation) can reasonably say about a fluent reader's performance is, "Her eyes scan the words and she speaks them as written."  And all anyone can say about a struggling reader's halting, error-filled performance is , "It  as if she  guessing."  Yet, Goodman's "scientific" formulation would have us believe that readers (skilled and unskilled) actually see words, select cues, make a guess, check the guess, reject the guess, make another guess, confirm the guess, and then say the word correctly or incorrectly.  If the guessing game is not a convenient fiction enabling Goodman to make sense of reading, but is considered a reality--something really happening--then a reader enacting the psycholinguistic guessing algorithm (propositions 5-20 above) would be carrying on an internal dialogue, as follows.

"James said...Hmmm, that t h looks like it might be there.  Okay, I'll say there....There lion...Wait... That doesn't work.  Okay, I'll try them...Them lion.... Nope...Maybe it’s this...This lion...Yeah, that sounds right.  This lion..."

But we rarely see anything like this guessing process.  Even when readers make a high rate of errors, reading is so fast it  hard to imagine that somewhere in their subvocal thinking they perform the mental guess work.  The only thing available to the observer of the above reading sample is the reader saying, "James said, (three-second pause) This lion."  Which is the more reasonable account of the three-second gap between "said" and "This" (and every other error or pause in a passage)?  (1) The reader naturally (with no instruction) repeatedly enacts multi-step guessing routines in milli-seconds, or (2) The reader simply needs someone to tell her, "That word  this...Spell this.... t h i s...What word?.... this.. Good.  Start the line again...James said, This lion is big."

In other words, Goodman's psycholinguistic apparatus (which, for science, would be considered reified fictions, or hypothetical constructs)  either: (1) incapable of any sort of test; and/or (2) simply impossible as an actual activity in real time.  At best, his psycholinguistic guessing game can only be treated as a metaphor---in which case one asks if a metaphor is the right foundation for actual reading assessment and instruction.

Whole language and upward mobility
        Goodman's hypothetical multi-step mental guessing apparatus had and continues to have strong appeal.  As mentioned, Goodman helped to move reading and reading instruction out of the mundane world of common, observable skills and into the world of esoterica.  Even simple decoding of text was now a complex mental activity involving higher order thought processes such as selecting, testing, confirming, and revising.  Reading instruction would now require special skills giving teachers access to the realm of thought where the hypothesized higher order guessing game was played.  Special courses, textbooks, conferences, and education professors would be needed.

In other words, Goodman was not merely offering an alternative to the phonic centered and word centered approaches.  He was creating an invidious status distinction.  He was offering prestige. This may have been appealing to education professors long known to occupy positions of low status and prestige in the university community, and to school teachers whose long hours, lack of appreciation, and low salary also connoted low status and prestige.  By making reading and reading instruction esoteric processes, Goodman's paper helped foster the idea that traditional reading instruction was only for commonsense-minded technicians interested in observable skill.  Whole language teachers and professors would be much more than this; they would be theoreticians--certainly a higher class of people.  This clarifies the facile denigration of systematic instruction, planned practice, teaching formats, field tested materials, scripted lesson plans, mastery tests, and in general accountability by whole language teachers and education professors.  Reading instruction was to be an art; and the reading teacher an artiste.

I do not like the agitators dressed up as heroes who wear the magic cap of ideals
on their straw heads; I do not like these ambitious artists who like to pose as
ascetics and priests but who are at bottom only tragic buffoons…
[Nietzsche. The genealogy of morals. Third essay, section 26]

Miscue analysis and the quasi-therapeutic
        As noted earlier, the only empirical evidence that Goodman presents in support of (as examples of) his guessing-game model are reading errors made by children.  Goodman calls these errors "miscues in order to avoid value implications" (p. 127).  For example, the story text reads,

"So, education was good!  I opened the dictionary and picked out a word that sounded good.  'Philosophical'.  I yelled.  Might as well study word meanings first.  'Philosophical: showing calmness and courage in the face of ill fortune. "

What the child read was,

" So, education was good!  I hoped a dictionary and picked out a word that sounds good.  PH  He yelled.  Might as well study what it meansPhizo  Phiso/soophical : showing calmness and courage in his face of ill fort  future  futshion."


Goodman states, "His expected (i.e., correct.  MK) responses mask the process of their attainment (That is, how he read correctly.  MK), but his unexpected responses (i.e., errors, or miscues  MK) have been achieved through the same process, albeit less successfully applied" (p. 127).  This is a very interesting statement.  Goodman is saying that when readers are fluent, we do not see how they do it; i.e., we do not see any guessing game.  It  is only when they err that we can make a case for guessing.  And then, with no rationale at all, Goodman states that reading well and making errors are done via the same process.  How could he possibly know that?

But as to incorrect reading itself, Goodman still has no direct, empirical evidence of guessing or any other activity in the elaborate guessing game apparatus.  He does not ask readers to, for example, say outloud what they are doing as they try to read.  All he has are interpretations of alleged covert guessing processes.  Goodman's interpretations (miscue analysis) reveal that he is willing to avoid the most obvious interpretation of errors in favor of the guessing hypothesis.  For example, Goodman says, "The substitution of hoped for opened could again be regarded as careless or imprecise identification of letters.  But if we dig beyond (emphasis mine.  MK) this common sense explanation, we find (a) both are verbs (b) the words have key graphic similarities.  Further, there may be evidence of the reader's bilingual French-Canadian background here, as there is in subsequent miscues (harms for arms; shuckled for chuckled, shoose for choose, shair for chair)" (p. 128).

It  clear that despite what Goodman makes of them, these errors are by definition examples of the "imprecise identification of letters"--and this imprecision rests very much on the child's lack of sufficient instruction on how to sound out familiar and unfamiliar words based on knowledge of sound/symbol correspondence.  It seems that Goodman goes out of his way to avoid the obvious account of reading errors--the child has not been taught word attack skills--so that Goodman can "dig beyond" the obvious and provide a more interesting guessing game interpretation for which there is not a shred of direct evidence--not when persons read well and not even when they make errors.

In summary, Goodman uses miscues as a resource for making interpretations about thought processes in a way that suits his guessing game model. There is nothing in the miscues themselves that suggests anything about thought processes.  But there is everything in the miscues that points directly at poor instruction.  Ironically, if Goodman's approach were in fact scientific, he would provide a panel of impartial observers with a set of miscue examples and ask the panel to make sense of each error or miscue, and then compare his interpretation with theirs.  In this way he could determine the reliability of his interpretations.

Goodman's entire guessing game model commits the fallacy of affirming the consequent.
        Goodman began his paper with the claim that his model would be an example of science--not mere common sense.  However, his argument commits perhaps the most fundamental error that the scientific method is devised to avoid; namely, the fallacy of affirming the consequent.  This fallacy can be depicted as follows.

          If P happens, then  Q will happen.
          Q happened. (Affirming the consequent)
          Therefore P happened.

For example,

          If there is frustration, then there will be aggression.
          There is aggression.
          Therefore, there is frustration.

The logical problem is that aggression may be the result of many things besides frustration.  That is why scientific researchers try to identify alternative explanations (e.g., models of aggressive behavior, reinforcement for aggressive behavior, a history of physical abuse) and see if these alternatives can be disproved--leaving the original proposition (If frustration, then aggression) intact for the time being.

Goodman's argument can be summarized as follows.

           If reading  is a psycholinguistic guessing game, then readers will make certain kinds of errors--miscues.
           Readers do make these kinds of errors--miscues.
          Therefore, reading is a psycholinguistic guessing game.

I have pointed out that miscues themselves are not direct evidence of any mental guessing game activity.  Goodman has simply interpreted them that way.  And there is no way to "dig" into anyone's thought processes to determine whether Goodman  right or wrong.  Even so, there are other explanations for these miscues besides an hypothesized mental guessing game.  The strongest candidate alternative is  poor instruction.  At least that is a plausible rival explanation (Hempenstall, 1999).  A student makes half a dozen errors trying to sound out "philosophical" because he was not taught exactly how to sound it out.  He is not firm on each letter/sound combination; he is not firm on sounding out a letter or blend, holding the sound and scanning the word for the next letter or blend.  He says "hoped" instead of "opened" because, again, he is not firm on the sounding out strategy, and because he has not had a teacher who systematically juxtaposed similar looking words--hoped/opened--and demonstrated again and again that they are sounded out differently.

In summary, it may be that many reading errors are not the result of guessing--as some sort of natural process--but are taught.  A student reads a passage and says "fort" rather than "fortune."  The teacher or tutor simply (and improperly) tells the student, "fortune."  The student repeats "fortune" and goes on with the passage--never really learning to sound out the difficult word.  Predictably, when the student sees "fortune" again, she says "fort"--because that is what she has "practiced" so many times before.  Or, when the student says "fort" rather than "fortune," the whole language teacher tells the student to think of a word that might go there--in other words, the teacher encourages guessing.  The student casts about and tries "future" and "futshion."  Predictably, when the student runs into "philosophical," the student will not sound out the word, but will do as she was taught--she will cast about for likely possibilities--"phizzo," "physical," "physicacol."  In other words, the student's errors do not reflect a natural guessing game apparatus.  They are direct effects of explicit (mal)instruction on guessing and failure to receive proper instruction on how to sound out words.

The scientific test of the above rival hypothesis--Errors represent how students are mistaught; they do not represent an innate guessing game--is relatively easy to perform.  Identify the sorts of errors made by students taught with whole language vs. the sorts of errors made by students taught with more focused instruction in each reading skill, in which errors are not corrected by having students guess but by firming up the sound-it-out strategy.  The prediction is that students who are taught to guess (and who do not know when a guess is correct), will make many more errors.


Kenneth Goodman's 1967 article helped to foster the whole language movement, which for several decades was the predominant approach to reading instruction in many schools of education, school districts, and states.  However, recent experimental research has shown that many of the defining (and allegedly revolutionary) design features of whole language (e.g., attempting to teach elemental reading skills--such as phonemic awareness, sound/symbol correspondence, word identification, and spelling--in the context of complex reading and writing activities that require these very skills) are at odds with what is known about effective instruction.  In addition, evaluation research shows that whole language is often less effective than its advocates claim, and is specifically less effective than field-tested curricula that provide systematic, explicit, comprehensive, precisely planned and logically progressive instruction on all of the elemental and complex skills in reading.

This paper examined the "viable" and "scientific" model of reading proposed by Kenneth Goodman--a model that has guided both the methods used in whole language (e.g., implicit, as-needed instruction; miscue analysis) and the ways whole language advocates legitimize and valorize their actions.  The examination of Goodman's "psycholinguistic guessing game" model revealed that Goodman:

1. Provides no data that adequately support his presumption that there is any such guessing game apparatus.  This may be because the guessing game is merely a metaphor.

2. Uses a small and selective sample of reading behavior (errors, or "miscues") as evidence that readers use the psycholinguistic guessing game.

3. Interprets these errors in a way that supports the guessing game model, but fails to consider plausible alternative interpretations and offers no evidence of inter-observer reliability of his interpretations.  [See Hempenstall (1999) for a reasoned and extensive critique of miscue analysis.]

4. Commits the fallacy of hasty generalization by asserting that his interpretations of some readers' guessing errors imply that all readers use the guessing apparatus.

5. Commits the fallacy of affirming the consequent when he reasons that errors signify the existence of a psycholinguistic guessing apparatus, when (and more reasonably) errors signify poor instruction.

In summary, it appears that the whole language movement--with all of its publications, assessment instruments and devices, conferences and organizations, college courses, classroom methods, and consequences for young readers--rests on a mere metaphor (the psycholinguistic guessing game) supported by assorted logical fallacies.  An interesting sociological question is, What cultural circumstances disposed so many education students, administrators, college professors, boards of education, and veteran teachers to so easily and so thoroughly accept Goodman's psycholinguistic guess game as a premise for their reading curricula?


Daniels, H., Zemelman, S., & Bizar, M. (2000). Whole language works: Sixty years of research.  Educational Leadership, 57, 2, 32-37.

Chapman, J.W., Tunmer, W.E., & Prochnow, J.E. (1999).  Success in Reading Recovery depends on the development of phonological processing skills.  The Plains, VA:  The National Right to Read Foundation.  Online at http://www.nrrf.org/rr_study_chapman.htm

Fletcher, J.M., & Lyon, G.R. (1998).  Reading: A research-based approach.  In W.M. Evers (Ed.), What's gone wrong in America's classrooms (p. 40-90). Stanford, CA: Hoover Institution Press.

Foorman, B.R., Francis, D.J., Fletcher, J.M., Schatschneider, C., and Mehta, P.  (1998). The role of instruction in learning to read:  Preventing reading failure in at-risk children.  Journal of Educational Psychology, 90, 37-55.

Goodman, K. (1967).  Reading: A psycholinguistic guess game.  Journal of the Reading Specialist, May, 126-135.

Goodman, K. (1976).  Manifesto for a reading revolution.  In F.V. Gollasch.  Language and literacy: The selected writings of Kenneth S. Goodman (1986, pp. 231-241).  LondonRoutledge & Kegan Paul.

Goodman, K. (1986).  What's whole in whole language.  Portsmouth, NH:  Heinemann Educational Books.

Gough, P. (1993). The beginning of decoding.  Reading and Writing:  An Interdisciplinary Journal, 5, 181-192.

Hempenstall, K. (1999). Miscue analysis: A critique. Effective School Practices, 17(3), 85-93.

Hiebert, E.H. (1995).  Reading Recovery in the United States:  What differences does it make to an age cohort?  Educational Researcher, 23, 9, 15-25.

Liberman, A.M. (1999).   The reading researcher and the reading teacher need the right theory of speech.  Scientific Studies of Reading, 3, 95-111.

Moats, L.C. (2000).  Whole language lives on:  The illusion of "balanced" reading instruction.  Washington, DC:  Thomas B. Fordham Foundation.  Online at http://www.edexcellence.net/library/wholelang/moats.html#foreword

National Reading Panel. (2000).  Teaching children to read.  National Institute of Child Health and Development.  Washington, DC: US Department of Health and Human Services.

Pappas, C.C., & Pettegrew, B.S. (1998).  The role of genre in the psycholinguistic guessing game of reading.  Language Arts, 75, 1, 36-44.

Powell, D., & Hornsby, D. (1993).  Learning phonics and spelling in a whole language classroom.  New York:  Scholastic Professional Books.

San Diego Unified School District (1999).  Reading Recovery Research Project.  Office of the Board of Education:  San Diego, CA.  Online at http://ww.nrrf.org_sd/rrrp.htm

Smith, F. (1985).  Reading without nonsense. New York: Teachers College Press.

Stahl, S., McKenna, M.C., Pagnucco, J.R. (1994).  The effects of whole instruction: An update and reappraisal.  Educational Psychologist, 29, 4, 175-185.

Thompson, L.J. (1995).  Habits of the mind.  New York:  University Press of America.