Assessments: English Proficiency

Status: Current | Last reviewed: 2026-10-19

The Assessments app brings Behavioral and English Proficiency together at https://assessments.dev.trueserve.deloitte.com. Enter the Name, Email and Code supplied by your administrator, or select Log in with Deloitte SSO to use your corporate profile. Code users see their assigned assessment types; Deloitte SSO users can access both. See sign-in and invitations. The previous langtest address redirects here.

For presentations, follow the demo walkthrough. View AI review opens your latest completed result directly on the reviewer's AI estimate tab; select Reveal AI estimate when it is hidden. Start new assessment begins a separate attempt, while Continue assessment resumes an unfinished one. Earlier results stay available on the card and in history. The configured daily attempt limit still applies.

Take an assessment

For the five-question Behavioral Assessment, see BPaaS candidate screening. It uses the same sign-in session and has its own scoring rubric.

  1. On My assessments, choose Get started under English Proficiency.
  2. Read the recording notice and research explanation. Every admitted Deloitte user has reviewer access to submitted evidence in this demo. Recordings are retained for the period shown. Optional coaching, assessor-training and model-improvement choices start unchecked and do not affect your score.
  3. Agree to the recorded assessment and select Continue to device check.
  4. Select Record a microphone check, speak for four seconds, and listen back. Confirm that you can hear yourself clearly. This check stays in your browser.
  5. Start the conversation. Listen to the instructions, then speak when the microphone opens. Avoid disclosing confidential project or client details.
  6. After your answer, pause or select Answer complete. Listen and respond to the interviewer’s follow-up questions. The progress indicator shows completed exchanges (normally two). Select Finish section when available, wait for Conversation saved, then continue. End section early saves available evidence but may leave the speaking level undetermined. Keep the tab open while recording saves.
  7. Open each reading section and select your answers. Its timer starts when you open it, and your choices save as you select them.
  8. Submit the final section. You can leave the page while your saved evidence is processed and return from My assessments.

Use a quiet room and headphones where possible. Allow microphone access in your browser. A connection interruption is recorded separately from language performance. Reconnect conversation is available when enough time remains; it does not restart the timer. If an answer fails to save, retry the selection before continuing.

Understand your result

After submission, the candidate page says Assessment received and explains that feedback will follow later. AI scores stay hidden there, even if you have already revealed them in the reviewer workspace. A reviewer can release a separate reviewed profile and feedback. Speaking and reading are reported separately. A missing skill level means the configured requirement cannot yet be determined.

The reviewed profile shows Configured research requirement as met, not met or not determined, alongside the speaking and reading thresholds. This reports the configured language requirement; it is not an employment decision.

B2 is the initial requirement for each skill; an assessment can require C1 or C2. Below B2 is reported as a single band in this professional-range pilot. Reading initially shows marks without a CEFR level because its score boundaries have not yet been established.

If the reviewer workspace shows criterion observations but speaking is undetermined, processing has finished and the evidence is incomplete for a full speaking estimate. Waiting will not fill that gap. A reviewer can inspect the observations and arrange further assessment.

This is a research assessment, not a validated CEFR certificate, IELTS, TOEFL or ACTFL result. It does not assess writing or issue an overall four-skill English qualification.

After release you can download the report, request another review with a reason, change optional reuse choices, or request deletion. Personal practice feedback is available after a reviewed result when you enable the coaching choice. Deletion removes retained evidence and report text; minimal audit metadata remains. Downloads already held by others cannot be recalled.

Review a candidate

Every admitted Deloitte SSO user sees English reviews and Behavioral reviews. You can reveal AI for your own completed submission in the reviewer workspace. You cannot save or release your own human assessment; another reviewer must do that.

  1. Open English reviews, then select an assessment.
  2. Use Recordings & responses to choose a section, listen to the full response, inspect Conversation transcript · grouped by speaker and examine reading answers. The transcript preserves repairs and overlapping acknowledgements. Its timestamps are within each voice connection; check uncertain words against the recording.
  3. Use Select and explain an excerpt to mark start/end positions, add a label and explain what the passage demonstrates. Use player position copies the current playback time. Preview it, then Cite excerpt. Use Add an evidence annotation for a correction; original recordings and transcript fragments remain intact.
  4. Enter speaking and reading levels only when supported, a candidate-facing summary, strengths, improvement priorities and your reason.
  5. Save an independent rating before opening the AI result. Under AI estimate, Reveal AI estimate is audited; subsequent ratings cannot be marked independent. Use AI draft in my assessment copies the proposed levels and feedback into your editable form. Confirm or change them, listen to the evidence and add your own reason. This does not save or release anything.
  6. Select Release this reviewed result to the candidate and release it when the evidence supports your judgment. You cannot release your own assessment.

Review history shows rating revisions, reasons and provenance. A new scoring run adds AI evidence without replacing an already released human result. Resolve an open candidate review request when releasing the revised result.

While a new scoring run is queued or running, the page says so and keeps the last completed result visible. The page refreshes its status automatically.

After AI reveal, AI evidence suggestions shows labeled passages with a criterion and an explanation, including counterevidence where available. Select Cite this passage, Hear surrounding context, or Adjust excerpt. These passages span original storage chunks; expand their recording references for provenance. Boundaries can be approximate, so adjust them to include a complete thought. AI descriptions are interpretations, not verified quotations.

The AI panel lists Suggested evidence passages as direct links to their sections. When you have not chosen a section, the evidence view opens the first suggested passage’s section.

A Provisional speaking suggestion is a draft from limited evidence. Its missing criteria remain explicit, and it is not a complete assessment. If interaction was not recorded, another speaking exchange may be needed. Reading has no AI CEFR suggestion until documented score boundaries exist. For older results, Request a new scoring run adds the new draft and passage suggestions without changing the original evidence.

Expand Interviewer audio and opening instructions to hear the prompts and generated interviewer speech. Generated speech can include audio suppressed during instructions or finalization. Inspect interaction and playback events (JSON) opens the corresponding delivery records; long logs use the returned nextCursor as the next request's after value. Saved evidence annotations appear with their source section.

Configure and examine the pilot

These controls require administrator permission, held by erigolden@deloitte.com and other explicitly elevated users. General reviewer access does not grant configuration access.

The main navigation uses concise page titles; the header labels, descriptive tab subtitles and footer have been removed.

See also