Syntactic Complexity


How to calculate Mean Length of Communication Unit (MLCU)

(PDF version)

How to calculate Complexity Index (CI)

(PDF version)

Normative tables for MLCU, CI, NDW, and TNW measures

Note: These tables are in Excel format.  You can use the Excel file to obtain standard scores by entering the raw scores in the red box; the adjoining boxes will show the z-score and standard score for the raw score you entered. 

Mean Length of Communication in Words (MLCU), Number of Different Words (NDW) and Total Number of Words (TNW) were obtained using SALT with the transcript of all 6 stories.  SALT calls these MLU-W, No. of Different Word Roots, and No. Main Body Words (the last two are under TTR for the whole transcript -- not TTR for 50 or 100 utterances). The entire transcript was used for these measures.  The Analysis Set was set to exclude the following:  abandoned and interrupted utterances, unintelligible utterances, nonverbal utterances, and yes/no responses to questions. 

Although there is a table for 6 year olds, MLCU and CI do not discriminate well between children with and without language impairments at 6 years, and thus scores should be used with particular caution.

Please note:  Be sure to change only the numbers in the red boxes. 


Mean Length of Communication Unit (MLCU)


MLCU is calculated using all utterances, including incomplete sentences, with the following exceptions:


Responses to questions and sentences that are interrupted or broken off.


Maze words (false starts, repeated words, fillers such as ‘um,’ ‘uh,’ etc.) (See the Transcription section of the ENNI Manual for the conventions for transcribing mazes in the SALT and CLAN programs.)


Utterances that are not story-related (e.g., unrelated statements or questions such as, "I dropped my barrette," or "I went to the beach last week."


Story-enders, such as "the end" or "that's finished". These are not counted because they typically are short and may shorten the MLCU, and since not all children use them, the effect will not be uniform.


MLCU is calculated based on words, not morphemes. To calculate by hand, count all the words in the transcript except for maze words and words in excluded utterances, and divide the total by the number of included utterances. Remember to count both words in contracted utterances such as I'm, you're, don't.


To use SALT: Mark any line that you want to exclude from the count by added an equal sign (=) at the beginning of the line, before the letter indicating the speaker (e.g.: = C what about^) To get MLCU, select Analyses on the toolbar, and then Word and Morpheme Summary. MLU in Words and MLU in Morphemes will be listed on the chart. If you have used / to divide contractions such as I/’m, you/’re, and so forth, then you must use MLU-Morphemes rather than words. Note that this will actually give you MLU in Words, counting each word in contractions, if you marked contractions but did not mark morpheme boundaries in the transcript. You should not have used / to mark bound morphemes, since these are not counted in MLCU. If you have transcribed contractions as two separate words (e.g., I am, you are, do not), you can use either MLU in Morphemes or in Words, since these will be identical.

To use CLAN: Again, you should not have marked morpheme boundaries in the transcript, and contractions should have been entered as two words, as described in the Transcription section of the ENNI Manual. Mark any line that you do not want to include in the count by added [+ bch] to the end of the line beforethe punctuation. The command for MLU without maze words is:


MLU –s”[+ bch]” ______.cha [the name of the transcript file goes in the blank]


PDF version




Complexity Index (CI)

Complex sentences are those that contain an independent clause (sometimes called a main clause) plus one or more dependent clauses.  In order to calculate the Complexity Index (CI), each C-unit in the transcript must be examined for dependent clauses.  (A C-unit is essentially a sentence; each is entered on a separate line in a transcript.  Partial sentences are also C-units, but partial sentences due to ellipsis when answering a question or those that are cut off before the child finished them are not analyzed.  See the Transcription section for how to determine utterance boundaries.) 

 Dependent clauses are units with verbs.  There are two basic types: subordinate and nonfinite.  Dependent clauses include the following types.  It is not necessary to identify the type, just to identify the dependent clauses.  To identify dependent clauses, first look for ‘extra’ verbs in the sentence (those not in the main verb clause).  Not all ‘extra’ verbs will count – see the exceptions at the end of the list. 

 Note: As noted above, some partial sentences will be included; however, do not count any dependent clauses that are in partial sentences.  For example, "not happy because it fell in" would be counted as a main clause but the dependent clause would not be counted.

 Complex sentences:  Main clause plus any of the following types of dependent clauses.

 Subordinate clauses:

  • Adverbial clause

      He was angry because she had dropped it.

Although it was an accident she ran away.

When she tried to fly it, the plane fell in the water.

She grabbed it before he could stop her.

After she grabbed it, he was really mad. 

  • Relative clause (wherever it occurs) 

The elephant, who was very large, could move very fast.

The lifeguard used the net that was by the pool. 

  • Sentences in which full sentences serve as noun phrases

She thought the plane was cool.

She believed (that) she could do it. 

  • Sentences with wh-clauses (wh- words as well as how, if, and like)

She doesn't know where she's going.

That's how it happened.

She wanted to see if she could fly it.

Do it like he does (it). 

  • Direct or indirect quotations

  • Sentences containing quotations followed by at least a clause

She said, ADon=t do that.@

Look!” he shouted.

He said, “Thank you” / “You’re welcome.”

She explained (that) she had dropped it in the pool.

      NOT: She said, AHello.@ or He said, ANot that.@

  • Appositive clause

The answer, whether or not we like it, is complex.

Note:  An appositive is only a clause if it contains a verb.  Otherwise, it is an appositive phrase and not a clause, as in "The answer, short and sweet...".  Appositive phrases are not counted.


Nonfinite clauses:

  •  Infinitive clause

He really wanted to get the plane.

He got the ball for her to have.

She had to ask for help.

They wanna play [= want to play].

It's gonna fall in the water [= going to fall].

[Note: it is better to transcribe gonna and wanna as going to and want to, so that both words are counted and it is easier to see the infinitives when counting clauses.]

  •  Unmarked infinitive clause

She let the ball fall in the water.

She made the plane fall in.

  • Wh- infinitive clauses (count as one dependent clause, not two)

Tell me when to start.

I know how to do that. 

  • Gerund clause

She made a big mistake by dropping the plane.

She made a mistake trying that.

Trying to fly the plane was a mistake.  

[note: 'she was almost hit by the falling plane= would not contain a gerund clause, since 'falling' functions as an adjective here, but 'she was almost hit by the plane falling' would contain a gerund clause]

  •  Past participles

She had a wagon with a balloon tied to it.

the boy did not want his sandcastle ruined.

He had no more balloons left.


Double embeddings

 C-units may contain more than one dependent clause.  Count each dependent clause. 

Main clause + 2 dependent clauses:

She was sorry 1) when she realized 2) that the plane had fallen in. 

I think 1) I know 2) what this is.   

I'm gonna [= going 1) to] let it 2) go now. 


Units that are NOT counted as dependent clauses

  • Conjoined phrases, either nouns or verbs (also called coordinated phrases) 

She had coffee, toast and cereal for breakfast.  (1 main clause only) 

She sat down and ate her breakfast.  (Second subject was deleted--thus 1 main clause.  This can be confusing because there are two verbs, but no dependent clause)

  • Auxiliary and main verbs in the same clause (these are considered one verb phrase)

She should have been thinking about his feelings.

  •  Prepositional phrases, however long or numerous they may be

She was tired from the top of her head to the tips of her toes. (1 main clause only)

 To calculate the Complexity Index 

1. Add up all the independent and dependent clauses in the transcript.  (Note: count partial sentences as independent clauses except for any you have excluded according to the rules above.)

 2. Divide by the number of independent clauses.

 Example:  In a transcript with 20 C-units that are countable, there are 20 independent clauses.  If there were 5 dependent clauses in the transcript, the CI would be:

             20 + 5 / 20 = 1.25.