MBRA: Reading

Data

Table          Chart

Basic Ideas

• It is possible to quantitatively measure at least some aspects of reading difficulty.

• Reading difficulty appears to consist of at least six separate components:
  • Lexical difficulty: Difficulty of the words used.
    This is pretty well correlated with word rarity — texts with more rare(r) words are more difficult to read. Donald Hayes' LEX measurement captures this well.

  • Syntactical difficulty: Difficulty of the sentence structures.
    It turns out that this can be very closely approximated by average sentence length. The longer the average sentence, the more complicated the text is to read (in general).

  • Semantic difficulty: Difficulty of the subject area.
    This is very tough to measure and I know of no examples of this sort of measurement. The basic idea here is that the text and grammar might be simple, but the underlying subject matter doesn't have to be. One good example would be to compare two stories of the same length, grammatical complexity and lexical complexity, but with one story having 100s of important characters, while the other story has only a handful.

    As another example, a text that assumes a lot of background material can be harder to read than one which provides all the background material as part of the text itself.

    Just because this is difficult to measure quantitatively doesn't mean we should ignore it. There is good evidence that knowing more background knowledge about a text makes one a better reader. The practical implication of this is that the texts read over time must cover a wide range of subject matter to ensure good reading mastery.

  • Font size.
    Smaller fonts are more difficult to read, partially because they require finer control over the eye muscles, which leads to fatigue. While it is worth building up over time a child's eye muscles to allow reading books with smaller fonts, the small fonts will make it harder for the child to focus on vocabulary and grammar.

  • Total length of the text
    More words are harder to read, partially because of fatigue, and partially because there is more to keep track of to understand what one is reading.

  • Literary Style
    Flashbacks, multiple threads of action and subtle symbolism can make for harder reading than a single thread of action told linearly with little or no symbolism.

  We can measure at least three of these components and account for one more! The last two are harder to manage quantitatively.

• Two scoring systems appear to be useful: LEX, by Dr. Hayes, and Lexile. The LEX program provides a score for lexical difficulty as well as an average sentence length. LEX is backed by a mathematical model that seems quite reasonable, and also seems to map fairly well to reality.

  Lexile combines lexical difficulty (although measured differently than LEX) and grammatical difficulty (using average sentence length as a proxy) into one score. This score can then be aligned with grade levels (see below). The primary expectation of the people behind Lexile is that it will be used to measure students' reading progress and that it will be used to select "next" books for them to read. Lexile has no theoretical model behind it.

• We can use LEX and average words/sentence to track reading progress and to order reading material. We can (and should) do this along both axes independently and can use Lexile as a check, but need not use Lexile (which aggregates the two concepts together) exclusively.

  The idea is that we want to be able to track reading ability at both a vocabulary level and at a grammatical complexity level.

• Finally, the quantitative measures are a guide only. The Lexile measure for "The Grapes of Wrath" (680L), for example, shows it as significantly easier to read than "D'Aulaires Book of Greek Myths" (1070L), "Little House in the Big Woods" (930L), or "The Lion, the Witch and the Wardrobe" (940L). This seems unlikely.

• Scientific books and articles have some of the most difficult vocabulary. A "normal" literature curriculum alone will leave a student unable to read science texts without struggling.

  As an example of this, The Illiad has significantly simpler vocabulary than a typical Scientific American article. Shakespeare (with archaics removed) has even simpler vocabulary than The Illiad.

  The upshot of this is that scientific texts must be part of the reading plan. Focusing only on literature is simply insufficient.

Curriculum Targets

Grade Age	End-of-Year Target		Example
	LEX	Tokens/sentence
KG 5½-6½	-30	10	David Adler "A Picture Book of ..."
1 6½-7½	-25	12	Hardy Boys Heinlein Juveniles
2 7½-8½	-20	14	Harry Potter and the Sorcerer's Stone Harry Potter and the Goblet of Fire
3 8½-9½	-17	17	Sherlock Holmes D'Aulaires' Book of Greek Myths The Hobbit Aesop's Fables (ISBN:0-448-05803-0)
4 9½-10½	-13	19	The Call of the Wild The Marvelous Land of Oz
5 10½-11½	-9	21
6 11½-12½	-5	24	The Iliad (Fagles translation) History's 100 Greatest Events The Doom that Came to Sarnath

Note that some of the example texts may not be readable due to content difficulty (e.g. Heinlein Juveniles at the end of 1st grade; Sherlock Holmes may be a stretch by the end of 3rd grade).