Still thinking about spelling (I do this more than is altogether sane), here’s the full quotation from Mark Twain:

I don’t see any use in having a uniform and arbitrary way of spelling words. We might as well make all clothes alike and cook all dishes alike. Sameness is tiresome; variety is pleasing. I have a correspondent whose letters are always a refreshment to me, there is such a breezy unfettered originality about his orthography. He always spells “Kow” with a large “K.” Now that is just as good as to spell it with a small one. It is better. It gives the imagination a broader field, a wider scope. It suggests to the mind a grand, vague, impressive new kind of a cow.
(Mark Twain, reported in the Hartford Courant, May 13, 1875)

An article, Widening our Perceptions of Reading and Writing Difficulties, describes research in Italy to show that,

even in Italian, in which it is relatively straightforward to convert sounds into letters, children still have difficulties in spelling. Younger children with dyslexia generally performed worse than proficient readers; however, the older ones showed a more selective impairment when spelling words, suggesting that knowledge of vocabulary may be more important in spelling than previously thought.

The same link explains the relatively common phenomenon called, in the UK at least, Spoonerisms:

Another study, from Tel Aviv University, Israel, provided the first systematic description of a type of reading disorder called “attentional dyslexia” in which children identify letters correctly, but the letters jump between words on the page, e.g., “kind wing” is read as “wind king”. Teachers and neuropsychologists often notice that children substitute letters when reading, but in this type of dyslexia the substitutions are not caused by inability [sic] to identify letters or convert them to sounds; they result from migrations of letters between words. The findings showed that letters would mostly migrate to the same position in another word, so the first letter of one word would switch places with the first letter of another word.

The article goes on to recommend an intervention to help this letter migration: present a ‎single word at a time, e.g., with the help of a word-sized window cut in a piece of cardboard.

The trouble with such a tactic is that the context disappears: the reader is no longer aware of all the cues other than single letters within the text that lead to understanding; s/he is solely dependent on decoding. At times this may well be an appropriate technique but it needs to be used cautiously I would think, especially as there is a strong coneection between effective encoding and knowledge of vocabulary, as cited in the first study.