For the last couple of days, I’ve been running into a lot of online phonics apologia about the use of nonwords, nonsense words, pseudowords, word-attack words, phonemic decoding items, and/or so-called “detached syllables,” in instruction, intervention, and assessment. For starters, the fact that these things have so many different names should cue us in that they are not an actual thing, not a scientific thing, anyhow (just like so-called “sight words”). They are not an actual category, if for no other reason than that many of the examples I’ve encountered over the years are actually real words in use in English, like cam, pate, lander, din, rayed, oft, knap, sedge, bi, [P]og, ta, lat, lum, barchan, and a lot more. Some people collect stamps; I collect linguistic scat from literacy educators and I study it.
People like to argue that nonwords are an effective means of teaching or assessing a student’s knowledge of what they call “grapheme-phoneme correspondences,” or GPC. But every single one of these nonword materials and studies misapprehends both a G and a P, as evidenced by such fabricated baloney as the “quadgraphs” [sic] like *<ough> and *<eigh>, and by the failure to even consider the difference between phones and phonemes. The fact of the matter in our writing system is that no G has a C to a P outside of an M, and M stands for morpheme. Once you remove phonemes and graphemes from a meaningful context, they’re no longer phonemes or graphemes.
To a resource, they all erroneously assume phonological primacy; that is, they remove orthographic phonology from its meaningful context because they wrongly assume that it’s primary to the meaningful constraints and influences of morphology and etymology. That very practice effectively means it’s no longer phonology, because phonology — including phonemes and the graphemes that spell them — is distinctive for meaning and it’s language-specific; nonwords are neither. It is noncontroversial that English orthographic phonology is delimited and constrained by meaning, structure, and history, regardless of how that fact makes people feel.
More than one person has suggested that nonwords were the only way to “break” a student of the habit of guessing at words, often in isolation. Well, you can break an overeating habit by taking up smoking, too, and you can kick a heroin habit by taking up methodone, but that doesn’t mean that the new habits have no harmful consequences. I’d rather focus my scholarship on what I can build than on what I can break.
I’ve also heard from a number of people working with “older” children who are called “treatment resisters” or “treatment fatigued” — kids who spend YEARS in Barton or Wilson and never get past so-called closed and open syllables [sic]. They may begin to “read” better (depending on what you think “read” means), but they continue to spell and write years behind their eulexic peers, largely spelling everything based on the way they pronounce it, because that’s exactly what they’ve been taught to do. I’ve heard from teachers and parents of children who read years ahead of their peers too, kindergarteners who read 3rd grade chapter books with ease, but have no idea how to spell or how to “decode” unfamiliar words, so they’re subjected to nonword drills in order to “measure” their “knowledge” of “alphabetics” or of GPCs.
So here is my analysis of all of the nonwords featured on a publicly available assessment called The Nonword Reading Test. The test instructions say “Either a regular or an irregular pronunciation is acceptable,” but no definition of “regular” or “irregular” is offered beyond that for <soser>, “soaser” would be “regular and rhyming with “loser” would be irregular.
First of all, there is NO ENGLISH WORD spelled with the sequence <oaser>, or even with an <oase> to which we could add an <er>. So how on God’s Grapheme Earth is that “regular”? Moreover, they do NOT specify how the <oa> or the medial <s> in “soaser” would be pronounced. Is the <oa> pronounced as in boat or as in broad or as in oasis? Is the <s> pronounced as in wiser, eraser, or pleasure? And how is the child or the teacher supposed to know or understand that?
You know why <loser> is spelled with an <o>? Because <looser> is a different word, and <lose> is cognate to <loss> and <lost>. What’s “irregular” about that? Just because teachers and researchers and psychometricians are generally ignorant to that breathtaking fact makes it no less a breathtaking fact. Context matters to so-called GPCs. Otherwise they’re neither Gs nor Ps, and any Cs you think are there are not real.
While we’re talking about <soser>, we may as well take a closer look at, um, <closer>: in“this street is closer than that street, the medial consonant is [s]. In he’s the best sales closer of the month” it’s a [z] — those two examples have two different suffixes that happen to spelled and pronounced the same, but don’t share a meaning! The ONLY way you know how to pronounce that word is if you know what it means. And that’s not even considering the pronunciation of the <s> in <closure>.
My analysis provides incontrovertible evidence against the motivating characteristics of all nonword resources: That dusty old crooked Assumption of Phonological Primacy.
Here’s the list from this test, along with English words I provide that share (some of) the same sequences of letters. If it feels like some kind of shameful hell for you to read through these, just imagine you’re a 12-year-old dyslexic with an IQ of 138. Or really, anyone.
1. plood: food, good, blood
2. aund: auberge, auto, Auschwitz — and <aunt> can rhyme with pant [ænt], haunt [ɔnt], or font [ɑnt], depending on your dialect.
3. wolt: colt, but also wolf, wolverine, woman, word, work, worm
4. jint: pint, lint — in many dialects lint and lent rhyme.
5. hign: sign, malign, benign, but signal, malignant; also hour, honor, and herb.
6. pove: shove, move, stove
7. wamp: ramp, swamp, swam
8. cread: bread, bead — for crying out loud, <read> is both [riːd] and [rɛd] — and how about create, or triad?
9. slove: glove, stove, prove — haven’t we been here before?
10. fongue: tongue, fondue, wrong, humongous, segue
11. nowl: bowl, fowl, snow, now, lowly, bowlegged
12. swad: swan, swam, swamp (is there an echo in here?)
13. chove: choir, cholera, chop, chef, and see pove and slove
14. duede: suede, due, clued, cued, swede, educate
15. sworf: sword, swollen, sworn, swore, word, work
16. jase: base, phase — vase, for crying out even louder, can be [veɪs], [veɪz], or [vɑz]
17. freath: breath, wreath, great, smooth
18: warg: war, warm, forward, wary, argue (there is no English word that ends in <arg> — if it’s a detached syllable, then what about larger?)
19. choiy: the graphemes <oi> and <y> are never, ever in sequence. Even <iy> is tightly constrained: that sequence is either across a morpheme boundary (as in multiyear) or in a non-English word, like teriyaki or aliyah. Consider joy and soy and bok choy.
Two Syllable (so much for that ‘detached syllable’ rationale)
1. louble: double, rouble, boucle, tousle, loud
2. hausage: sausage, usage, garage, stage, courageous, also hour and honor and herb again.
3. soser: loser, poser
4. pettuce: lettuce, induce; petty has a double <t>; petting has a doubled <t>; flattop has neither.
5. kolice: police, policy (some people say POlice), malice, preslice. And why does this have a <k> before an <o>?
6. skeady: steady, beady, skean
7. dever: clever, fever — hell, lever can be both [‘lɛvɚ] and [liːvɚ]!
8. biter: This is not a nonword. It’s a word: “My new puppy is a biter.” Nonetheless, if it were, say, <piter> instead, notice writer, whiter (note the different <er> suffixes), liter, arbiter
9. islank: island, mislay, Islam, mankiller, and anyhow, vowel pronunciation is often disputed before [ŋ], but the orthographic phonology is revealed by the graphemes.
10. polonel: colonel, colony, colon, polish, police, Polish — what in the hell can *polonel tell you about anything at all? Someone please make it stop.
11. narine: This is actually a word; it means “pertaining to the nostrils” or the same as “narial.” Criminy, is your google broken? But also, marine, margarine, alkaline, urine, line, incline…
12. kiscuit: biscuit, intuit, circuit, circuitous, recruit, and how about Jesuit? The Jesuits have always valued knowledge and evidence.
Why 19 monosyllables? Why 12 disyllables? Why 31 total? Only the <shade + ow> <know + s>.
This “test” features the following rough distribution of graphemes, depending, for example, on whether the <s> in <islank> and the <g> in <hign> are supposed to be graphemes or markers, or on whether the <<ui> in <kiscuit> is one grapheme (bruise) or two (intuit). Those are just a few examples of the ascientific foolishness embedded in here that makes a real scientific analysis challenging:
c (4, including both [k] and [s])
g (2-3, [g] and [ʤ] and [∅])
h (0-2, initial only, which could be French markers)
k (4, including the unconventional *kolice)
l (9-10, including *polonel. Honestly.)
s (8-9, most of which have multiple possible pronunciations)
v (4, of which 3 are in an <ove> rime)
w (3-4, initial or following <s>)
gue (1, or maybe it’s a <g> followed by a <ue>, as in argue, or a <g> followed by a <u> and an <e>, as in segue. Who knows?)
That’s 17 of 20 single-letter consonant graphemes (x, y, and z didn’t rank), two digraphs (out of more than two dozen), and whatever the heck <gue> is supposed to be. Why are <n> and <l> — which have a single phonemic association — as important as <s>, or more important than <c> or <ch>, which all have multiple pronunciations?
I so want to cuss right now. FFS: the middle F stands for Fonics, though.
o (8-9, including whatever the hell is up with *polonel)
u (0-2, depending on whether the <u> in *duede or in *fongue is a grapheme or not)
Final non-syllabic <e> (10, of which 3 are in an <ove> rime)
This includes 5 or all 6 of the single-letter vowel graphemes, but <i> and <o> are featured 2-3 times as much as <a> and <e>. It also includes three of many rhotic vowel spellings (why <or> but not <oar>, <ore>, <oor>, or <our>, which can all spell [ɔɹ]?) It also includes 6-8 vowel digraphs (out of around 30) and zero vowel trigraphs (we have two). This doesn’t even include half of the orthography’s vowel graphemes, the vast majority of which are digraphs. You know why <feat> has an <ea> and <feet> has a <ee>? I can give you at least two good reasons for each word. And they make total sense.
How is this nonword GPC inventory in any way representative of any kind of coherent “knowledge” about graphemes, phonemes, or their alleged correspondences? It’s just not. Whoever slapped it together — as with every single nonword resource I’ve ever seen, used, or recently investigated — has no idea that <w> can mark the phonology of a subsequent <a> or <o>, or that an <ove> rime has multiple possible pronunciations. I can think of at least three good reasons why <move> is spelled with an <o>; nonwords can’t think of a single one.
As my good and wise friend and colleague says, if a child writes *<dun> instead of <done>, you have all the information you need that he already owns the CrAPP concept of GPCs, and that it’s already doing its damage.
Can anyone offer any explanation that makes this kind of nonsense anything other than a sadistic but nonlethal method of collecting meaningless data about meaningless “knowledge” about meaningless “patterns”? I welcome any and all nonsense word measures. I guarantee you I can find you massive problems with any one of them.
Ighm aul ierse. Doar’z oapon.