RULES FOR MLU Calculation

I. Identifying Utterances

A. When transcribing divide "speech stream" into utterances by using natural syntactic boundaries (i.e., you'd write them as separate "sentences"), falling inflections, and pauses.

B. Transcribe everything- what you say, what the child says, mazes (i.e., false starts, revisions, repetitions), incomplete utterances, etc. Everything gets used in one type of analysis or another.

C. Identify "countable utterances," (i.e., those that contain a subject and a verb).

1. There are two types of utterances that do not contain a subject and verb, yet are still countable for MLU calculation:

a) Utterances in which the copula verb is omitted ("She nice to me.")

b) Commands in which the "you" is understood ("Sit down and eat your dinner.")

2. Elliptical utterances (those that omit information that's included in a previous utterance. For example, "pancakes can toast" in response to "What did you have for breakfast?") are usually not counted because they don't contain a subject and verb. If one did include a subject and verb, it would be counted.

D. If an utterance contains more than TWO clauses (i.e., subject and verb) connected by the conjunctions and, but, or, divide it into two separate utterances. The rule is that you should have no more than two independent clauses in each utterance. This is done so that a child's MLU wouldn't be inflated by the following, "My mom fed me breakfast and she washed my hair and she made my bed and she read me a story and she gave me a dog and sheÖ."

E. Your sample should contain 100 countable utterances. Since you've transcribed everything, your sample will include far more than that. Number every utterance. Then go back and circle (or otherwise mark) the utterances that you'll be using for the MLU count (i.e., the "countable" ones).

F. Omit unintelligible utterances, but count unclear words as one word/one morpheme if the rest of the sentence is clear.

G. Exclude any utterance which is an exact repetition of another.

H. Do not count spontaneous imitations of the examiner (or mother) unless the child alters forms.


II. Tallying Morphemes

A. Tally the morphemes included in each "countable" or circled utterance, using the following guidelines

B. Mazes (i.e., false starts, interjections such as uh, oh; repetitions) are not counted as additional morphemes.

C. All compound words, proper names, and ritualized reduplications count as single words and single morphemes. E.g. "moo-cow, choo-choo, night-night, see-saw".

D. Count as one morpheme all diminutives, e.g., "doggie".

E. All irregular past tensed verbs and irregular plurals are counted as one morpheme.

F. All modals, auxiliaries, and catenatives (hafta, wanna, gonna) count as one word, one morpheme each.

G. All inflections count as separate morphemes (e.g. -ed, -s, -en, -ing).

H. In stuttering, count the word once in the most complete form produced.

I. Contractions count as "base + morpheme", i.e., two morphemes, one word.

J. For self-corrections and false starts, count the final version.

K. Contracted forms typically count as separate morphemes (e.g., donít, isnít, etc. count as two.)


III. Calculate the Mean Length of Utterance

A. Count the morphemes in each countable utterance.

B. Sum the talleys of each utterance

C. Divide by the number of countable utterances (this should be 100).

IV. Reference Data (taken from Miller and Leadholm, 1992)