The ILR scale and the DLPT, explained

The Interagency Language Roundtable (ILR) scale is the U.S. Government's measure of language proficiency: an ordinal scale with six base levels, 0 to 5, and plus levels for 0 through 4. The May 2021 edition of the ILR reading descriptions says the scale runs from no communicative ability to mastery. The ILR does not test anyone. The roundtable's FAQ says there is no "ILR Test" and that agencies score their own tests against the descriptions. Most federal agencies rely on the Defense Language Proficiency Test (DLPT) and the Oral Proficiency Interview, according to the Defense Language Institute Foreign Language Center (DLIFLC).

Base levels and plus levels

The scale grew out of work the Foreign Service Institute began in the 1950s, according to a history by Dr. Martha Herzog on the ILR site. The first version was a single scale from 1 to 6 that rated "language" as a whole. It was later split into separate scales for each skill and standardized to six base levels. In 1968 several agencies wrote formal descriptions of the base levels for speaking, listening, reading, and writing. By 1985 the ILR had revised the document to include full descriptions of the plus levels.

The 2021 preface sets the rules. A plus level describes proficiency that substantially exceeds one base level without fully meeting the next. To earn it, a performance must match the plus description, not merely be strong at the base level. Each description covers four aspects: functional ability, precision of forms and meanings, content meaningfulness, and contextual appropriateness.

The FAQ gives names for two levels. Level 2 is "Limited Working Proficiency," and many agencies require at least Level 3, "General Professional Proficiency." The FAQ describes Level 5 as functionally equivalent to a highly educated native speaker. The 2021 reading edition describes Level 5 as mastery and makes no native-speaker comparison.

What each reading level means

The May 2021 edition, in summary:

Level	Reading ability
0	Essentially none beyond occasional isolated letters, characters, or words
0+	Some memorized words and phrases, such as numbers, names, and common signs
1	Simple texts with basic information, often with great difficulty
1+	Simple material beyond basic needs, though inaccurately
2	Straightforward everyday and routine work material; generally cannot draw inferences
2+	Many informal contexts; not always complete in formal and professional ones
3	Formal and informal language with accuracy on general and professional topics; can draw inferences
3+	A number of prose styles with accuracy
4	Writing tailored to many audiences, with considerable precision
4+	Precise, nuanced language in most contexts, with only rare mistakes
5	Clear and precise writing in almost any context

How the DLPT5 tests reading and listening

DLIFLC's familiarization guide for the multiple-choice DLPT5 says a lower-range test reports 0, 0+, 1, 1+, 2, 2+, or 3, and an upper-range test reports 3, 3+, or 4. Reading and listening get separate scores. A lower-range reading test has approximately 60 questions on about 36 authentic passages of up to approximately 400 words. Only the passage is in the target language; the questions and answers are in English. Each test allows 3 hours, and dictionaries and note-taking are not permitted. The guide says scores do not reflect speaking, writing, or job-related performance.

The DLPT guides page adds the rules behind the numbers. An examinee needs a 3 on the lower-range test to take the upper-range one. Multiple-choice tests need at least 100 validation examinees, preferably 200 or more. Languages with fewer get constructed-response tests, scored independently by two raters, and an examinee generally needs 75 percent of the questions at a level to be awarded it. The page says older DLPTs were calibrated so that errors leaned generous, and the DLPT5 so that errors lean strict.

Translation is a separate scale

The ILR publishes descriptions for translation performance that are separate from the four skills. A translator must read the source, write the target, and choose the target expression that best matches the source meaning, which the ILR calls congruity judgment. Proficiency testing serves as a screening tool, and assessing translation requires a translation test. The descriptions group levels into three bands. Translation is not possible at the minimal levels, 0+ to 1+. Level 2 and 2+ renditions are not professional translations and must go through quality control. At level 3 the skills begin to align for a reasonably accurate and reliable translation, still a draft subject to quality control. Gisting and summarizing are left out as distinct skills.

The audio translation descriptions, approved on September 28, 2012, cover rendering speech into written text. They list unfavorable conditions such as low volume, background noise, multiple speakers, and overlapping exchanges. Listening and writing tests are screening tools only, and job-related performance testing is called absolutely necessary.

What to check

• Which skill does a rating cover? A DLPT5 score covers reading or listening, not speaking, writing, or translation.
• Did a level 3 come from a lower-range test, which cannot report higher, or from an upper-range test?
• Which edition of the descriptions applies? The DLPT5 guide's appendix reproduces older wording, such as "Functionally Native Proficiency" for level 5.
• For translation work, was the person rated against the ILR translation descriptions, which treat proficiency testing as a screening tool?
• Is a gist or summary being accepted as a translation, although the translation descriptions exclude both?
• For recorded speech, did testing reflect unfavorable conditions such as background noise and overlapping speakers?

Azcuna's BABEL reads, translates, and searches foreign-language documents entirely on-device, with no cloud and no connection, and keeps every claim tied to the passage it came from.

Sources

1. 1Interagency Language Roundtable Skill Level Descriptions for Proficiency - Reading (May 2021 Edition) · Interagency Language Roundtable
2. 2Frequently Asked Questions · Interagency Language Roundtable
3. 3An overview of the history of the ILR Language proficiency skill level descriptions and scale · Interagency Language Roundtable (Dr. Martha Herzog)
4. 4DLPT Relevant Information and Guides · Defense Language Institute Foreign Language Center
5. 5Defense Language Proficiency Test 5 System Familiarization Guide for Multiple-Choice Format · Defense Language Institute Foreign Language Center
6. 6ILR Skill Level Descriptions for Translation Performance · Interagency Language Roundtable
7. 7ILR Skill Level Descriptions for Audio Translation Performance · Interagency Language Roundtable (approved September 28, 2012)

Where this shows up in our work

• BABEL, offline multilingual translation and search translates and searches foreign-language documents on-device, every fact tied to its source passage.