The following is an analysis of the Benchmark Assessment pages 4 and 5 that is on the DIBELS website.
While I agree 100% with the concept of using nonsense words for diagnostic testing, I do believe that either the words chosen should follow the patterns used in English spelling or they should follow the patterns used in producing dictionary diacritical type spellings. And certainly 28% correctly spelled is unacceptable, especially since they are nonsense words!
For example, in our language the /k/ sound in words of a one syllable base is spelled ick as in pick and nick, picking, and nicking. However, the /ik/ sound when it occurs in words of a polysyllabic base is /ic/ as in picnic, panic, and scenic. No single syllable based word ends -ac, ec, ic, oc, or uc. Polysyllabic words such as maniac, Toltec, panic, and manioc can
The ending /j/ sound in monosyllabic base words using short vowels is dge as in badge, edge, ridge, dodge, and fudge. To my knowledge the only word ending aj are haj and Taj but then the ending j is pronounced /zh/. As you probably are gearing this to the teaching only of single phonemes that are not represented by digraphs or trigraphs with the possible except of the ending ck, I made more changes in the corrected version below.
The ending /s/ sound in monosyllabic base words using short vowels is -ss as in grass, dress, kiss, toss, and cuss.
The ending /v/ is practically nonexistent in English spelling. Perhaps, the monks had problems with their script and felt they just had to add the silent e to make the word look pretty. Who knows? Who cares? So /hav/ is spelled have and to avoid misspelling /sav/ as save, we stick in a silent l to spell /sav/ salve.
The ending /l/ sound is almost always ll in monosyllabic base words having the letter vowels e, i, o, and u as in will and willingness. In polysyllabic base words we have the single l as in peril, compel, numeral, Errol, and mogul. The same is true with the ending /f/ and /z/ sounds. Only al as in pal and gal has the single l. The "awl" words have the double l as in all, ball, etc.
It is my opinion that in any assessments using nonsense words, misspellings of real words should not occur as does hiz for his, kub for cub, huf for huff, dok for dock, sok for sock, kab for cab, wil for will, duz for does, and pac for pack. Nor should real words appear as nonsense words as do tic ad et doc id "hon" el wan. It is crucial, I believe, that no errors in phonics should occur as it does in the choice of the word wan. That cannot rhyme with any other -an words such as tan, man, ran, etc. The w- control prevents that. In English spelling the letter a when it follows the consonant w (or its twin u as in squander) changes the pronunciation to "ah" as in wan, wand, wander, squander, etc.
For a more complete discussion, see "The Mechanics of English Spelling," in The Teaching of Reading and Spelling: a Continuum from Kindergarten through College as well as The W- and -R Controlled Clans, Volume 4 of The Patterns of English Spelling.
My corrections would make your test of nonsense words of two or three phonemes (with no digraphs except the ck and the double letter digraphs ss, ll, and zz.) look something like this: