Language sample analysis is really critical when initially evaluating and even monitoring progress, but they can be tedious and slow! If you read my previous post, you are going to love this update (12/2021) with even more helpful tips!!
I’ve got info on how to quickly get language samples through using hacks on Microsoft Word, Google docs, plus my Language Analysis FREEBIE and MLU/PGU Normative charts! Whew! So much good info!
Where are you getting your language sample?
First things first, best practices for language samples (especially for school-aged kids) are narrative or expository language samples, not conversation. After the preschool years, conversation language is typically less advanced that narratives or expository. So make sure you’re getting your students’ best efforts.
Microsoft Word for Language Sample Analysis
In the past, I have done my language sample analysis using Microsoft Word. Yes, you read that correctly! You can do so much using tools you likely already have! And it’s fast! Check out the steps below to complete one of your own:
Step One: Type in 50 utterances. Put each utterance on their own line. I like to use numbered bullets to keep track of how many I have. See an example below:
Make sure to break apart all morphemes with a space between them as shown below:
Step Two: Calculate total words. To do this, get rid of the numbered bullets first!! One you have a document with just the child’s utterances, go to Tools —> Word Count.
Step Three: Divide the number next to Words (113 in this example) by 50 (the total number of utterances). This calculation gives you an MLU of 2.26. Easy, right?!
Step Four: Fill out a Language Sample Analysis Checklist to analyze grammar, vocabulary, pragmatics, narratives, and more! I have a free one in my Teachers Pay Teachers store. Click here to check it out!
Google Docs for Language Sample Analysis
If you’re more of a Google docs fan, you can do a pretty similar process. I can never remember where to find the word count, but if click on Help, then type in “Word count,” it will pop up.
Are you ready to make this process even easier???
Google docs has a speech-to-text option called “Voice Typing.” (Again, if I can’t find it in the menus, I just click “Help” and search for “Voice Typing.”) It might ask you for permission to use the microphone. Then a microphone icon will appear. When you are ready to use the feature, click the microphone and it will turn red. That means that it’s recording.
Here’s the amazing thing: a recent study showed that used Google’s voice typing was actually MORE accurate than seasoned professionals doing transcription when recording narrative samples from 7-11 year old students. The study actually used a pre-recorded sample and they recommend that you let Google do its job, then review the sample once after, just to make sure that you agree with Google. Speech-to-text has come a long way!
So you can let Google transcribe it, then divide into utterances and morphemes to run your word count and figure out your MLU.
MLU – What Does It All Mean?
So you have an MLU calculation, but what does that even mean?
MLU is not necessarily a good benchmark to form goals by, but it does help when making diagnostic decisions.
The most updated information on MLU numbers I could find was in this article. If you look at Table 4, they have a great chart outlining MLU in words and in morphemes among kids aged 2;6-8;11 in 6 month increments with standard deviations, in both SLI groups (labelled “Affected”) and typical groups (labelled “Unaffected”).
For kids older than that, MLU hasn’t really shown to be a good measure, but the sentence structures mentioned above in my checklist might be a better place to start!
One More Language Sample Measure: Percentage Grammatical Utterances (PGU)
A more recent entry on the Language Sample Analysis front is Percentage of Grammatical Utterances, or PGU. MLU tells us if a child’s utterances are generally as long as we would expect them to be. But once children are putting words together, we really need to figure out why they aren’t as long or complicated as we want them to be. PGU gives you a measure of grammaticality and can help you zero in on actual goals for treatment.
Here’s how you figure out PGU (from this recent study):
- Make sure your language sample is divided into utterances. Include only intelligible, complete, and spontaneous C-units (independent clause with dependent clauses, as well as nonclausal statements that were a complete thought)
- Count how many contain at least one grammatical error (You can look at the article for all their specific errors, but think verb tenses, pronouns, morphemes like plural -s, etc.)
- PGU = (Total utterances – Ungrammatical utterances)/(Total utterances) * 100%
I took the information from the article and put it into this lovely chart if you want to see what kind of percentages you should be expecting from typical and atypical language development:
Hope that’s helpful and it saves you some time! There are definitely more in depth ways of doing this, but let’s be honest, we are working under some serious time constraints!
{thanks for reading}
References
Guo, L., Eisenberg, S., Schneider, P., & Spencer, L. Percent grammatical utterances between 4 and 9 years of age for the Edmonton Narrative Norms Instrument: Reference data and psychometric properties. American Journal of Speech–Language Pathology. https://doi.org/10.1044/2019_AJSLP-18-0228/.
This is an awesome idea! Definitely a time saver! Thank you!
You’re welcome!! SO glad you found it useful! Thanks for commenting 🙂
Thank you so much! I’ve been using Microsoft Word for language sampling but didn’t know there was a word count feature. Also, your checklist is just what I’ve been looking for, quick to use but also comprehensive.
That makes me so happy! Thanks so much for commenting.
So simple, but never thought to use it and will save major time! Thanks!
You’re welcome!! It is so simple but something I use all the time! Thanks for commenting!
Thanks for this tip. I am a google docs user, and you prompted me to check and see if google docs has a similar feature. Yes, it does!
Thanks for the checklist also!
You’re welcome! What a great idea to look on Google Docs! I will have to try that!
Great idea to use Word to quickly measure MLU! And this checklist helps me ensure that I include everything on my analysis. I just used it today for a 3-year-old I am evaluating. Thanks so much for sharing!
Abby
Schoolhouse Talk
Love it. I don’t have the Word software so I am trying it with Pages. Seems to work in a similar way. Thank you. 😀
Language samples provide so much information and are, unfortunately, often overlooked. Thanks for your time saving hints. I just purchased SALT (Systematic Analysis of Language Transcripts) software and am loving it. I use it with my K through 12 students. It is so helpful.
Hi, I have been on a search for updated predicted MLU norms with standard deviations so that I can calculate the standard deviations from the mean on my language samples. Do you have anything you use?
Hi Nancy! The most updated information I could find was in this article: https://pubs.asha.org/doi/10.1044/1092-4388%282009/08-0183%29
Check out Table 4!
Can you include how to calculate the SD for each age?
The standard deviations for MLU can be found in the original article. Look for Table 4. Here is the link, available if you are an ASHA member: https://pubs.asha.org/doi/10.1044/1092-4388%282009/08-0183%29
Thank you so much for taking the time to make this post. You are very gifted, and I’m amazed at how you made something complex seem simple.
Hi. Thank you so much.
In non verbal children, uteering may be a few single words. How can we calculate MLU-w?
Thanks
For those children, a MLU-w measurement isn’t really informative, or really even appropriate. You would calculate it the same way, with the MLU-w being 1 if the children are truly only communicating with one word, or you could just informally say that the child speaks in single words. At any rate, MLU-w is merely an indicator that there could be a language impairment. The single word level (or MLU-w of 1) would only be appropriate for a 12-26 month old child.
For non-speaking children, I find it more useful to report on communication intents and functions (are they only requesting? labelling?), as well as parts of speech being used (are they only talking nouns? Do we hear any adjectives? verbs?)
I have a question about the MLU chart. Is it possible that the headers are reversed for the MLU norms: i.e., the “MLU In Words” and “MLU in Morphemes” columns? I am looking at the publication and think that the numbers are different. I am trying to make a chart for myself and don’t feel confident with this — which is why I want to thank you for this article, it was extremely helpful!
Yes! Good catch! I’m fixing it ASAP. Thanks!
Hi!
I was wondering if the headers have been corrected yet. I had just copy and pasted the image when I saw the comment about it being switched. Thanks!
Yes I updated the headers! When in doubt, MLU in morphemes will always be higher than MLU in words. Thanks for checking!
I am confused. You are counting the words, but I believe you are supposed to count the morphemes. Am I forgetting something? I haven’t done this in a while so it’s possible.
You can count words or morphemes, it just depends on what you are trying to find out! If you want to have a quick measure of words per utterance, you can just count words. If you are trying to get an idea of the level of complexity of the language sample, you would want to count morphemes – if the morpheme count is basically the same as the word count, your student isn’t using morphemes.
The most important thing is that if you are using any charts for comparison, you are comparing words to words and morphemes to morphemes. Some research will use MLU in words and some will use MLU in morphemes and some will use both! Brown’s morphemes chart uses MLU in morphemes, or sometimes written as MLUm, so if you’re using that as a reference, you’ll need to count morphemes.
Hopefully that’s a little clearer!
And, just to clear it up further, when I’m doing a language sample, I put a space between morphemes so the word and the morpheme are each counted by the program. So if my student says “I playing,” (2 words) I would type out “I play ing” which is counted as 3 in the word count of the program.
Thanks for asking!