ToolsConductScience tool

PGU %Free in-browser calculator

PGU Calculator.

Percent Grammatical Utterances from a pasted language sample. Tap G or U per utterance and the calculator returns PGU live, plus the Eisenberg & Guo (2013) severity band once you have scored 25+ utterances. Data stays in your browser.

Run calculator Request a Quote

PrivateData stays in your browser

LiveNo sign-up required

Open methodFormula shown on this page

CitableMethods and citation included

Calculator

Results update in place

Paste your language sample

One utterance per line. Mark each utterance as grammatical or ungrammatical below; the calculator returns Percent Grammatical Utterances across the scored subset.

Paste a transcript above, then score each utterance as grammatical or not to compute Percent Grammatical Utterances.

Get the full analysis

Unlock normative comparison and error analysis

This free tool covers the basic case. ConductSpeech adds normative comparison, error categorisation, and a parent-ready report.

Get the full analysis with ConductSpeech

When to use

Computing PGU from a transcribed language sample for an evaluation or progress report
Tracking grammaticality growth across sessions for a child with Developmental Language Disorder
Reporting a grammaticality index alongside MLU and NDW for a complete LSA description
Teaching graduate SLP students how PGU complements MLU and lexical diversity
Screening a caseload child for whom MLU alone looks normal but grammatical errors are frequent

Do not use for

Scoring samples below 25 utterances — the severity bands are unstable
Diagnosing DLD from PGU alone — combine with MLU, lexical diversity, and narrative measures
Judging AAE or other dialectal productions against mainstream American English grammar rules
Auto-flagging ungrammatical utterances — that requires clinical context (use ConductSpeech for error categorisation)
Reporting PGU across bilingual samples without a parallel score for each language

Score each utterance in isolation

Judge the utterance against adult grammar without trying to reconstruct what the child meant. "Him go home" contains a case error (him for he) and a tense error (go for went / goes) — mark it ungrammatical, do not mentally repair it.

Omissions of required morphemes count as errors

"The dog run" omits third-person -s ("runs"), so it is ungrammatical. "He eat lunch" omits the third-person -s and the past tense ("He ate lunch" or "He is eating lunch"). Omission errors are the most common driver of low PGU in children with DLD.

Phonological errors do not reduce PGU

If a child says "wabbit" for "rabbit" but the sentence is otherwise well-formed ("The wabbit is hopping"), score the utterance grammatical. PGU measures morphosyntax, not articulation. Score articulation separately with the PCC Calculator.

Judge dialectal productions against the dialect

Features of African American English (habitual "be", zero copula, remote past "been") are grammatical within AAE. Do not penalise them on PGU. If you are unfamiliar with the dialect, consult a dialect-aware reference before scoring.

PGU is the single most sensitive LSA marker of DLD

Children with Developmental Language Disorder consistently score 10-15 percentage points lower on PGU than typically developing peers, even when MLU is age-appropriate. If MLU looks fine but the parents report concerns, compute PGU before closing the evaluation.

Method

Utterances are parsed by the shared SLP utterance parser (src/lib/slp/utterance-parser.ts) — newline-delimited first, sentence-ending punctuation as a fallback, SALT-style line prefixes stripped. Each utterance is rendered in a scoring table; the clinician taps Grammatical or Ungrammatical to score it (tapping the same button again un-scores). PGU = scored grammatical utterances / total scored utterances * 100. Severity classification follows Eisenberg & Guo (2013): >= 93% typical, 78-92% borderline, < 78% clinical concern. Severity is reported only once at least 25 utterances are scored, because the bands are unstable below that threshold.

Published

Published 2026-04-06. Calculations support planning and documentation; verify procurement decisions against manufacturer specifications or institutional SOPs.

How to cite

How to Cite

ConductScience PGU Calculator. ConductScience, Inc. 2026. Available at: https://conductscience.com/tools/percent-grammatical-utterances-calculator

Eisenberg SL, Guo L-Y. Differentiating children with and without language impairment based on grammaticality. Language, Speech, and Hearing Services in Schools. 2013;44(1):20-31. doi:10.1044/0161-1461(2012/11-0089)

Pavelko SL, Owens RE. Sampling Utterances and Grammatical Analysis Revised (SUGAR): New normative values for language sample analysis measures. LSHSS. 2017;48(3):197-215. doi:10.1044/2017_LSHSS-17-0022

Leonard LB. Children with Specific Language Impairment. 2nd ed. MIT Press; 2014.

Rice ML, Wexler K. Toward tense as a clinical marker of specific language impairment in English-speaking children. JSLHR. 1996;39(6):1239-1257. doi:10.1044/jshr.3906.1239

What Is Percent Grammatical Utterances?

Percent Grammatical Utterances (PGU) is the percentage of a child's utterances that contain no grammatical errors. It is one of the three core descriptive dimensions of a language sample alongside length (MLU) and lexical diversity (TTR / NDW). PGU captures grammaticality — whether the child marks tense, agreement, plurals, articles, and pronouns correctly — which is a different axis from either sentence length or vocabulary variety.

Why it matters. PGU is the single most sensitive language-sample index of Developmental Language Disorder (DLD). Children with DLD consistently score 10-15 percentage points lower on PGU than typically developing peers, even when their MLU is age-appropriate. A child who produces long utterances that are riddled with morpheme omissions will look fine on MLU and low on PGU — a pattern that only emerges when you report both measures.

How it is computed. Score each utterance as grammatical (G) or ungrammatical (U). The numerator is the count of G utterances; the denominator is the total number of utterances you scored. Utterances you leave unscored (unintelligible, imitations, maze behaviour) do not enter either the numerator or the denominator.

Interpreting PGU — Eisenberg & Guo (2013) Bands

Eisenberg & Guo (2013) established the most widely cited reference for PGU severity bands, based on a sample of preschool and school-age children with and without language disorder:

>= 93% grammatical — Typical range. Matches the mean PGU of typically developing preschool and school-age children.
78% - 92% grammatical — Borderline / clinical concern. Warrants closer evaluation with additional measures (MLU, lexical diversity, narrative).
< 78% grammatical — Clinical concern. Consistent with Developmental Language Disorder. Does not substitute for a full diagnostic workup.

The calculator assigns a severity band only once you have scored at least 25 utterances. Below that threshold, PGU is unstable and can swing by 10+ percentage points with a single reclassification. Always score 50+ utterances before including PGU in an evaluation report.

How This Calculator Works

Paste your language sample into the textarea, one utterance per line. The calculator:

Splits the sample into utterances using the shared SLP utterance parser (newlines first, sentence-ending punctuation as a fallback).
Renders a scoring table with a Grammatical / Ungrammatical toggle per utterance and a 44px touch target for mobile use.
Reports PGU percentage, scored count, ungrammaticality count, total utterances, and total words as you score.
Assigns an Eisenberg & Guo (2013) severity band once you have scored at least 25 utterances.
Stores your scores in component state — nothing is uploaded, nothing is persisted across a refresh.

Tap the G / U buttons to toggle a score; tap the same button again to un-score an utterance. Unscored utterances stay out of the PGU denominator.

Frequently asked

What is Percent Grammatical Utterances (PGU)?

Percent Grammatical Utterances is the proportion of a language sample that contains no grammatical errors, expressed as a percentage. You judge each utterance as grammatical (no morphosyntactic errors) or ungrammatical (at least one error), then divide the grammatical count by the total scored utterances. PGU captures the grammaticality dimension of expressive language — a separate axis from MLU (length) and lexical diversity (vocabulary).

How is PGU different from MLU?

MLU measures the average length of a child's utterances; PGU measures how many of them are grammatically well-formed. A child can have an age-appropriate MLU but low PGU (long sentences riddled with errors) or a low MLU and high PGU (short but correct utterances). Reporting both gives a fuller picture of expressive language than either metric alone. Children with Developmental Language Disorder (DLD) typically show low PGU even when their MLU is in the normal range.

What counts as an ungrammatical utterance?

Any utterance with at least one morphosyntactic error — omission of a required morpheme (he go instead of he goes), substitution of an incorrect form (me want instead of I want), incorrect tense marking, missing articles, or any other departure from standard English grammar. Phonological errors and word-finding pauses do not make an utterance ungrammatical for PGU purposes. Utterances in a non-standard dialect (e.g., AAE) are judged against that dialect, not against mainstream American English.

What are typical PGU values by age?

Eisenberg & Guo (2013) reported mean PGU values around 93% for typically developing preschool and school-age children. Children with Developmental Language Disorder typically score below 78%. The 78-92% range is a borderline / clinical concern zone that warrants closer evaluation. PGU rises across development — a 3-year-old averages lower than a 6-year-old — so always compare against an age-matched reference when interpreting an individual score.

How many utterances do I need for a valid PGU?

At least 25 scored utterances before the Eisenberg & Guo severity bands are interpretable, and ideally 50 (the SUGAR sample floor) before reporting PGU in an evaluation. The calculator refuses to assign a severity band below 25 scored utterances and flags the sample as too short.

Does the calculator auto-detect grammaticality?

No. Grammaticality judgment requires clinical context — what is ungrammatical in mainstream American English may be perfectly grammatical in African American English, and a child on your caseload may have a known substitution pattern that the tool does not know about. Manual scoring keeps the clinician in charge of the decision. For auto-detection plus error categorisation, upload the sample to ConductSpeech.

Is my data private?

Yes. The calculator runs entirely in your browser. The transcript you paste and the grammaticality scores you record never leave your device, and nothing is saved on our servers. Close the tab and the sample is gone.

325

Free tools

1,200+

Institutions

100%

Client-side

Uploads required

Next steps

Protocol