I Write Like....

[Chairman Risus]

Dan Brown?

DAN FUCKING BROWN?

This is a joke, right?  They have about 5 authors listed, and they are assigned at random.  Cram, you slick bastard.

I'm getting the feeling there are a very limited option of authors you can be assigned to. If there were a wider selection and an explanation of why you match up with certain authors, I'd be interested in this.

[Golden Applesauce]

Now, that post is not the paragon of legibility, but run-on sentences do not require grade 80+ to decipher.  That's moronic.

yeah, but your example is unrealistic. Nobody writes 80+ word sentences.


In the textbook industry, we use the grade index as a tool to hone writing to a certain difficulty level. For example, a third grade reading book will go from reading level 2.5 to 4 over the course of the year.


My company publishes ESL books, which are geared to be somewhat easier to read than the mainstream textbooks. Our level 5 book (for example) has roughly the same grade reading level as a 2nd grade mainstream book.

I will admit to looking at the formulas and going for the crazy edge case where they break down horribly.

I just don't see how the grade level formulas work.  I agree with the vague notion of a "reading level," but don't really see how to go about quantifying it without having a bunch of students read the books and rate them.  A sentence that's too long and/or with too many clauses is certainly more difficult to understand, but is that a reading skill problem?  I'll freely admit to not knowing much about the process of learning to read, but it seems to me that dealing with overlong sentences is more of a working-memory problem - remembering which clauses go to which word, etc - than a reading skill.  I would think that the hardest sentences to read would be the ones with the most possible interpretations, where the reader has to figure out which one the author actually meant, or garden path sentences.

The factors I'm seeing in making reading a given text more difficult are:
-vocabulary (including common words used in uncommon or archaic ways.  "Lusty men" means something totally different if the author wrote a century or three ago.)
-the length of a sentence (if you forget how the sentence started before you get to the end, it's not going to make sense.)
-the clearness of the sentence (which clauses / descriptors go to which other words, which words belong to which clauses, which noun is the subject - anything that makes constructing a parse tree more difficult.)
-inherently confusing words (vague pronouns, words which can be different parts of speech in different circumstances, homonyms.)
-subject matter (ever go back an read a book you loved as a kid, only to be discover that they were having sex?)
-symbolism and metaphors (can be tricky to pick up on, especially if you're not familiar with the culture of the author.)
-deception (seeing through attempts to mislead the reader, discovering the bias of an author attempting to appear neutral, noting absent relevant information)

Heuristics that look at word / sentence length get indirect measures on the first two (arguably #3 as well) but not much else.  I'm surprised that they work at all.  Of course, if you're writing a textbook, you don't use much symbolism and the subject matter is carefully chosen to match the grade level, so I guess that helps.

[Mesozoic Mister Nigel]

Dan Brown?

DAN FUCKING BROWN?

This is a joke, right?  They have about 5 authors listed, and they are assigned at random.  Cram, you slick bastard.

I got a different author for every piece, including different pieces within series, so it's obviously not all that sophisticated. I think it's working primarily on vocabulary and sentence length.

[LMNO]

I don't even think they look at the vocabulary.  I'm guessing they have a program that looks for conjunctions, punctuation, and (as Nigel said) sentence/word length.  They then took texts by each author, analyzed it, and made that a base score in a quadrant on a grid.  Your piece is then analyzed the same way and placed somewhere on that grid.  Whoever's pattern you're closest to, that's "who you write like".

Since most of us post in a conversational manner, Dan Brown and Stephen King would turn up the most, because that's the style they try to employ.


Ooh!  Be right back.

[LMNO]

Back.

Apparently, Hirly0 writes like Douglas Adams and... Dan Brown.


Fail analysis is fail.

[Jenne]

  But still, it was fun.

[Rumckle]

I decided to put some random song lyrics in:

Paparazzi sounds like Chuck Palahniuk

Sympathy for the Devil sounds like Ian Fleming

[Eater of Clowns]

Back.

Apparently, Hirly0 writes like Douglas Adams and... Dan Brown.


Fail analysis is fail.

Brilliant.   

[Cramulus]

Back.

Apparently, Hirly0 writes like Douglas Adams and... Dan Brown.


Fail analysis is fail.

what author would have been a more appropriate answer for hirley0?

[LMNO]

R2D2?

[AFK]

Okay, I had to do these too:

Hugh:  Dan Brown
Horab:  Kurt Vonnegut
LHX:  Jane Austen
Purple Eris:  Bobcat Goldthwait

[Dysfunctional Cunt]

I used the piece I did on my Dad.  It's at EB&G.  It told me I write like Chuck Palahniuk.

YAY?

[Jenne]

@ LHX and PurpleEris...that's fucking sweet.

[Triple Zero]

I just don't see how the grade level formulas work.  I agree with the vague notion of a "reading level," but don't really see how to go about quantifying it without having a bunch of students read the books and rate them.  

I'm pretty sure this is described on the wikipedia page about them, but I might have read about it some place else. Go look it up if you like.

Anyway, they did pretty much exactly that. Got a bunch of students to read books and rate them. Then they made up a formula and found that with the correct coefficients it has a pretty high correlation with grade level.

Check the wikipedia page or the related links to see how high that correlation is, because they did also test for that.

Therefore there are multiple grading formulas, because there's more than one way to measure this effect.



The important point to realize about the grade level formulas is that they only work one way. If you write a text in a normal manner, then you can check its grade level with pretty high confidence by applying a very simple sentence length / syllable length type of formula. Because the average correlation with all those other (very reasonable) criteria such as vocabulary etc is very high. That is, someone that uses long sentences and lots of syllables tends to write using a larger vocabulary, use more metaphores, confusing words, etc etc.

Except that those things are all pretty hard to measure, so they tried to find a formula that is able to estimate those variables up to a certain amount of accuracy.

The other way around, it doesn't work. If you have a text that is too high a grade level, it's not going to be magically easier to read for kids if you make the sentences shorter. Even though the formula would tell you it is a lower grade, if you use the same obscure wordings and metaphores in shorter sentences, it's not going to be that much easier for kids, of course

[AFK]

At the risk of sounding too DK-ish...

It is really, really hard to write for a lower grade level or reading comprehension.  Back at the old job I had to write a curriculum for an on-line parenting course and I was constantly having to revise my work because I was making it too hard for the average parent to read.  I think we were supposed to shoot for a 5th or 6th grade level if I remember correctly.  Hats off to those who can do it easily.