ProductHow it worksPricingSecurityFor districtsResourcesFAQ
Log InGet EASI at Northern Speech

Blog post

How Long Should a Language Sample Be? What Research Says

How many utterances is enough? Plan on at least 50, and closer to 100 for detailed grammar. Here is what the research says about length and context.

The EASI team · October 24, 2026

A speech therapist listens while a young child tells a story from a picture book.

How long should a language sample be? For most children, plan on at least 50 utterances, and closer to 100 when you need detailed grammar; an article in ASHA's journal for school clinicians suggests at least 50 utterances in most cases (Pezold, Imgrund, & Storkel, 2020). If you think in minutes, a study of 3 year olds found that 7 to 10 minutes of talk gave reliable results, and that 1 minute samples may mislead (Guo & Eisenberg, 2015).

Every speech therapist has asked it at the end of a long day: how much talk do I really need from this child? The honest answer depends on the measure you plan to use and the setting you choose. The good news is that the research gives us real numbers to work with. Here is what it says, where it disagrees, and how to put it to work with the child sitting right in front of you.

How many utterances should a language sample have?

For most clinical work, plan on at least 50 utterances, and lean toward 100 when you need a fine grained look at grammar. Researchers have long described 50 to 100 utterances, gathered in roughly 10 to 15 minutes, as a balance between reliability and the time you actually have with the child (Guo & Eisenberg, 2015). The large SUGAR normative project collected 50 utterance conversational samples from 270 typically developing children ages 3 to 7 (Pavelko & Owens, 2017).

Some measures ask for more talk. The Index of Productive Syntax (IPSyn) was built around 100 utterances (Scarborough, 1990). A large study of 1,051 conversational play samples found that 50 utterance samples showed most of the same psychometric properties as longer ones, which points toward a shorter IPSyn in the future (Yang, MacWhinney, & Bernstein Ratner, 2022).

Is time a better target than utterance count?

Often, yes. Minutes are easier to plan around with a child, and the research measured both.

  • Heilmann, Nockerts and Miller compared 1, 3 and 7 minute cuts of conversation and narrative samples. Measures were quite consistent across the cuts, and productivity, lexical diversity and utterance length were the most reliable in short samples (Heilmann et al., 2010).
  • Guo and Eisenberg cut 22 minute parent and child play samples from sixty 3 year olds. Total number of words (TNW) reached a reliability of .90 at 7 minutes. Number of different words (NDW) and mean length of utterance in morphemes reached .90 at 10 minutes. They concluded that 7 to 10 minutes is desirable, and warned that 1 minute or 10 utterance samples may lead to misdiagnosis (Guo & Eisenberg, 2015).
  • An older study of preschoolers sampled twice, days apart, found NDW and MLU stayed above .92 only with 175 or more complete, intelligible utterances (Gavin & Giles, 1996).

So the studies do not fully agree. Short samples can give a fair snapshot of utterance length, but vocabulary counts and repeat testing need more talk from the child. When the stakes are high, as with eligibility, more is safer.

There is a gentle, human reason to wait, too. In the first minute, many children are still warming up to you. Their talk can look shorter and simpler than it really is. Give them time to settle in before you judge what you hear.

Which context should I use for which child?

Context changes what you hear, so choose it on purpose for the child in front of you.

  • Play works well for toddlers and preschoolers. In one study, free play gave more utterances, but a smaller share of complex sentences than conversation or story generation (Southwood & Russell, 2004).
  • Conversation suits preschool through school age. Ask about things absent in time and space, like a holiday or a family visit, and avoid strings of yes or no questions (SALT conversation protocol).
  • Narrative, a story retell or a story the child makes up, tends to draw out longer, more complex sentences than conversation in young school age children (Westerveld, Gillon, & Miller, 2004).
  • Expository talk, like explaining how to play a favorite game, was more syntactically complex than conversation at every age from 7 to 49 (Nippold et al., 2005). It is a strong choice for older students and teens.

If you compare a child to norms, use the same context and procedure the norms used. That point comes up again and again in the research. If you are weighing tools for this kind of comparison, our guide to EASI vs SALT for language sample analysis walks through the differences.

What about bilingual children?

Sample both languages when you can, so you see the whole child. In a study of 170 Spanish and English bilingual kindergarteners who told stories in both languages, English MLU, English grammaticality and Spanish grammaticality best predicted overall language ability (Bedore et al., 2010). Another study found that a story retell classified bilingual children more accurately than a story the child told from pictures (Kapantzoglou, Fergadiotis, & Restrepo, 2017).

Practical tips for a sample you can trust

  1. Plan for minutes, count for utterances. Aim for 10 to 15 minutes of talk and check that you reach at least 50 complete, intelligible utterances.
  2. Match the measure to the length. If you need IPSyn or a precise MLU, keep going toward 100.
  3. Pick the context for the question you are asking. Play for everyday talk with little ones, narrative or expository for complex sentences.
  4. Talk less, wait more. Open ended questions and comfortable pauses give the child room. Try "Tell me more about that." or "What do you think will happen next?" In one study, a clinician interview drew out the same profile as free play for school age children with language impairment, and most behaviors showed up more often in the interview (Evans & Craig, 1992).
  5. Record it. In a national survey, most school based SLPs transcribed in real time, and many did not report recording their samples (Pavelko et al., 2016). A recording lets you check your transcript later.
  6. Keep the audio clean. A quiet room and a close microphone mean fewer unintelligible utterances to set aside. For more on this, watch How to Record a Clear Language Sample.

Where EASI fits

EASI lets you record live, using the interactive backgrounds and high interest pictures built right in, or upload audio, or bring in a transcript. You review and correct the words and speaker labels, and attest each line of the transcript. Then, when the sample supports it, EASI calculates MLU, number of different words, vocabulary diversity, the Index of Productive Syntax (IPSyn), percent consonants correct, and intelligibility and phonological information for you to review. Learn more about EASI language sample analysis.

The numbers are a starting point, never the final word. You still decide whether the sample is long enough and what it means for this child. If a child is talking freely, let the sample run toward 50 utterances or more, especially before an eligibility decision.

EASI is HIPAA and FERPA compliant, with a BAA and insurance. MySLP comes with every plan.

Watch the one minute answer

https://youtube.com/shorts/jKTH8Nsskmc

In this short Ask MySLP video, we walk through how many utterances to collect, how many minutes work for 3 year olds, and how to get more talk with play, stories and open ended questions like "Tell me more about that." and "What do you think will happen next?"

EASI is available through Northern Speech Services: learn more or get EASI from Northern Speech Services. Plans are $39 a month (MySLP plus 4 report tokens) or $199 a year (MySLP plus 50 report tokens). The report generator is in beta. Questions or groups of 10 or more: info@northernspeech.com.

FAQ

How many utterances do I need for MLU? Most guidance says 50 to 100. One study of 3 year olds found MLU reached strong reliability at about 10 minutes (Guo & Eisenberg, 2015).

Is a 3 minute sample enough? For a quick look at utterance length, maybe (Heilmann et al., 2010). For vocabulary counts or eligibility, plan on 7 to 15 minutes.

Does the context really matter? Yes. Narrative and expository tasks usually draw out more complex sentences than conversation or play.

Should I count unintelligible utterances? Researchers disagree. Traditional analysis leaves out any partly unintelligible utterance when figuring MLU, while SUGAR keeps some (Guo et al., 2018). Follow the rules of the norms you compare to.

References

  • Bedore, L. M., Peña, E. D., Gillam, R. B., & Ho, T. H. (2010). Language sample measures and language ability in Spanish English bilingual kindergarteners. Journal of Communication Disorders, 43(6), 498 to 510. https://doi.org/10.1016/j.jcomdis.2010.05.002
  • Evans, J. L., & Craig, H. K. (1992). Language sample collection and analysis: Interview compared to freeplay assessment contexts. Journal of Speech and Hearing Research, 35(2), 343 to 353. https://doi.org/10.1044/jshr.3502.343
  • Gavin, W. J., & Giles, L. (1996). Sample size effects on temporal reliability of language sample measures of preschool children. Journal of Speech and Hearing Research, 39(6), 1258 to 1262. https://doi.org/10.1044/jshr.3906.1258
  • Guo, L. Y., & Eisenberg, S. (2015). Sample length affects the reliability of language sample measures in 3 year olds: Evidence from parent elicited conversational samples. Language, Speech, and Hearing Services in Schools, 46(2), 141 to 153. https://doi.org/10.1044/2015_LSHSS-14-0052
  • Guo, L. Y., Eisenberg, S., Bernstein Ratner, N., & MacWhinney, B. (2018). Is putting SUGAR into language sample analysis a good thing? A response to Pavelko and Owens (2017). Language, Speech, and Hearing Services in Schools, 49(3), 622 to 627. https://doi.org/10.1044/2018_LSHSS-17-0084
  • Heilmann, J., Nockerts, A., & Miller, J. F. (2010). Language sampling: Does the length of the transcript matter? Language, Speech, and Hearing Services in Schools, 41(4), 393 to 404. https://doi.org/10.1044/0161-1461%282009/09-0023%29
  • Kapantzoglou, M., Fergadiotis, G., & Restrepo, M. A. (2017). Language sample analysis and elicitation technique effects in bilingual children with and without language impairment. Journal of Speech, Language, and Hearing Research, 60(10), 2852 to 2864. https://doi.org/10.1044/2017_JSLHR-L-16-0335
  • Nippold, M. A., Hesketh, L. J., Duthie, J. K., & Mansfield, T. C. (2005). Conversational versus expository discourse: A study of syntactic development in children, adolescents, and adults. Journal of Speech, Language, and Hearing Research, 48(5), 1048 to 1064. https://doi.org/10.1044/1092-4388%282005/073%29
  • Pavelko, S. L., & Owens, R. E. (2017). Sampling Utterances and Grammatical Analysis Revised (SUGAR): New normative values for language sample analysis measures. Language, Speech, and Hearing Services in Schools, 48(3), 197 to 215. https://doi.org/10.1044/2017_LSHSS-17-0022
  • Pavelko, S. L., Owens, R. E., Ireland, M., & Hahs Vaughn, D. L. (2016). Use of language sample analysis by school based SLPs: Results of a nationwide survey. Language, Speech, and Hearing Services in Schools, 47(3), 246 to 258. https://doi.org/10.1044/2016_LSHSS-15-0044
  • Pezold, M. J., Imgrund, C. M., & Storkel, H. L. (2020). Using computer programs for language sample analysis. Language, Speech, and Hearing Services in Schools, 51(1), 103 to 114. https://doi.org/10.1044/2019_LSHSS-18-0148
  • SALT Software, LLC. (2023). Conversation elicitation protocol. https://www.saltsoftware.com/media/wysiwyg/reference_database/ConElicProtocol.pdf
  • Scarborough, H. S. (1990). Index of Productive Syntax. Applied Psycholinguistics, 11(1), 1 to 22. https://doi.org/10.1017/S0142716400008262
  • Southwood, F., & Russell, A. F. (2004). Comparison of conversation, freeplay, and story generation as methods of language sample elicitation. Journal of Speech, Language, and Hearing Research, 47(2), 366 to 376. https://doi.org/10.1044/1092-4388%282004/030%29
  • Westerveld, M. F., Gillon, G. T., & Miller, J. F. (2004). Spoken language samples of New Zealand children in conversation and narration. Advances in Speech Language Pathology, 6(4), 195 to 208. https://doi.org/10.1080/14417040400010140
  • Yang, J. S., MacWhinney, B., & Bernstein Ratner, N. (2022). The Index of Productive Syntax: Psychometric properties and suggested modifications. American Journal of Speech Language Pathology, 31(1), 239 to 256. https://doi.org/10.1044/2021_AJSLP-21-00084

Your caseload shouldn't take over your life.

EASI connects evaluations, reports, progress notes, planning and scheduling, with the SLP in control.

Get EASI at Northern Speech More resources