I write this post to another board a few years ago to explain a major cause of these errors as in Catherine's example:I'm going to look through my notes from Morningside re: "close-in non-examples." I remember them having a distinct procedure for teaching two mixed-up words.
In fact there has been some solid research on this phenomenon. I can't cite chapter and verse off the top of my head, but it was done by Engelmann or perhaps others in the DI community, and investigated just that pattern you mention: children who consistently misread when for then, of for from, where for there, and similar errors of words that are usually both syntactically and visually similar. [Catherine here, interrupting for a moment: the problem I was asking about mostly involved substituting one preposition for another.]
I may not be summarizing the research 100% correctly, but the gist of it is that what we are seeing is the result of the learner having been presented with two very similar things at the same time (or almost the same time), and not having been taught either one to mastery, so that the two became fused in memory in an indistinct way, and one would be randomly substituted for the other. In the case of most of these word substitution errors, the child is correct some of the time and incorrect some of the time -- so s/he doesn't ALWAYS read "when" for "then" or vice-versa. S/he randomly says one word or the other whenever s/he is presented with either of the duo. Over time, this habit becomes a neural circuit and a learned response that is very resistant to change. It rarely is self-corrected, the way some other decoding errors are, because the misreading usually makes sense even if it is incorrect (it may not make the CORRECT sense, but it is not unintelligible). "Julia talked to Edith when she went to the mall" means something different from "Julia talked to Edith then she went to the mall", but the inaccurate reader is unlikely to notice. It's not like the horse/house instance, where if a child reads "The rider mounted his house" s/he will usually recognize that can't be right.
The DI research indicated that children easily learn what they called misrules (usually due to poor or misleading initial presentation) , and these word substitution errors fall into that category. The more the error is practiced, the harder it is to supplant it and engrave a correct response. It may take hundreds of corrections of when/then errors before the reader reliably gets those words right (has a new neural pathway). Many examples of b/d confusion are the same thing - they are not visual discrimination errors in most cases, but are simply instances of children not having learned one of the pair thoroughly before the other was introduced. When words *are* visually similar, the likelihood of confusion is increased, but the visual similarity is not the cause of the confusion. Some children, for instance, repeatedly confuse other letter pairs, or words, that bear no resemblance to each other -- r and g, for instance, and further inquiry usually finds these were introduced about the same time, the child was not provided with adequate practice to mastery, DID remember that it was either one sound or the other, and would randomly say either /g/ or /r/ when presented with either stimulus (leading teachers to say "he knew it yesterday!" when they don't appreciate the randomness of the response -- but randomness within a very narrow range of options, usually only 2 or 3). One student I had consistently confused the words "and" and "said" which had no relation or resemblance to each other, but were introduced the same week on his classroom "word wall."
One effective remedial strategy -- not the only one -- is to teach these pairs explicitly, but first one word at a time. With when/then, for instance, practice writing the word, dictating "when" sentences, reading phrases and sentences with "when" and stories with "when" (reminding the student at the outset that /when/ is the word s/he will encounter, and no discrimination between when and then will be required); then do the same with the other in the pair, then mix them up -- but have the student slowly and carefully sound out each word if necessary. The solution to those misrule/misreading learned errors is careful and targeted practice to mastery.
One analogy I use (if someone thinks of a better one, please share) is to compare the student's experience confusing when/then to someone with a drawer full of socks -- let's say 30 pairs of socks, half of them grey and half of them black. If you pull two random socks out of the drawer, you will get a matched set half the time. It may not matter much that you get a mismatched pair the other half. Unless you have some reason to pay attention, you don't need to change anything. A grey sock and a black sock keep your feet warm, so what?
I had an experience a couple of years ago that made me aware of how this process works. I had to go around and do an attendance-taking chore first thing in the morning and check off two or three students in several classrooms (long story why -- not interesting). Most of the students I knew, and I would just look in the door and see if they were there and mark the attendance sheet. One Grade 4 class, however, had 2 girls on this list, both of whom were new to me. They did not look anything alike, really -- different ethnicities, one much taller, but both wore hijab and had very unusual first names. When I was first introduced to them, I failed to ensure I knew which one was which by solidifying identifying details in my mind. Usually, both were present, and I'd look in and say to myself, There's N. and there's F -- good -- but I could have N and F mixed up, I just noted they were both there. They would straighten me out when I met them in the hall, if I addressed them by the wrong name, but I continued to mix them up for months. I realized partway through the year that this was a similar error to the mislearning children sometimes do:
--I was introduced to 2 new items to learn at the same time, and they shared some common features but I did not spend time to adequately solidify recognition of the special features of each
-- I was frequently required to rehearse my mislearning but not in a situation where it mattered if I was correct or not, and where I rarely received corrective feedback so that I could stop myself and say, Wait a minute, N is the shorter one, or some other identifying factor.
-- The more times I repeated my random error (it was random because half the time I got their names right), the more likely I was to repeat it in the future.
-- the problem was solved by one girl moving, so after that I knew the remaining girl's name. Even so, a year later, I sometimes addressed that girl by the other girl's name. The experience did provide me with an insight into how these misrules/ mislearnings happen.
The DI research into this was part of their development of the Corrective Reading Program. They did a lot of empirical testing to find out how many times students needed to get those confusions corrected before they would master them and be error-free. The number of correction required varied with the age of the student -- the longer s/he had been making the error, the more correction required. Secondary students needed many more corrections than primary students. They have published some of this data, but it is not (probably) of that great an interest to most people. The principle of mislearning and misrules and how to correct them is of general interest, however.
So is the preventative measure, one of which is to introduce the items separately (sometimes separated by long periods, days or weeks or months), ensuring each is learned to mastery (in the case of word reading, by sounding-out all through the word, and segmenting all-through-the-word for spelling). At the upper grade levels, the DI programs also have students spell the words aloud in addition to blending and segmenting them. I've been meaning to look up the research on this (I know they included it because it improved results in large-scale field trials) but don't remember the details. Oral spelling of the words is not a feature of the programs for beginning readers.
Other programs I have heard of (not the DI ones) take the bull by the horns and DO introduce commonly confused items together -- but emphasize practice to mastery right at the start, so that the confusion and "practicing mistakes" does not have a chance to occur.
No doubt there are many other reasons why individual students get particular words wrong, but many of these common substitution errors do seem to share the same features and have similar origins. Many beginning readers are not expected to master correspondences or be accurate in their word reading (if they "make meaning" it is enough) so the Engelmann hypothesis and experimental data explain the phenomena I usually see.