BIVR Exam: Mastering Verbatim Fidelity Under Speed

Verbatim reporting tests whether you can reproduce the spoken word exactly, not whether you can summarise it well. Build fidelity and speed together: transcribe untidy real speech, mark what you cannot verify, and self-audit against audio with an error rubric.

What 'verbatim' actually demands: the word-for-word standard

Verbatim means recording every spoken word as uttered, including repetitions, false starts and filler, not producing a tidied, readable summary of what the speaker meant.

A verbatim record answers the question 'what exactly was said?', while a summary or edited transcript answers 'what did the speaker mean?'. Legal proceedings, arbitrations, inquiries and Fitness to Practise hearings rely on the first, because a witness's hesitation, correction or choice of words can itself be evidence. When you practise, treat every 'er', restart and repeated phrase as content to capture, not noise to delete.

The discipline runs both ways. Deleting a disfluency changes the record, but so does inserting words the speaker never said — smoothing grammar, completing an unfinished sentence, or normalising dialect. A useful self-check after any practice passage: read your transcript aloud against the audio and ask whether each deviation was spoken aloud by someone. If not, it belongs in your error count.

• Capture spoken words exactly, including filler and restarts
• Do not tidy grammar, complete sentences, or substitute synonyms
• Treat wording itself as part of the record, not just meaning

Commonly confused terms: knowing which role and standard applies

BIVR distinguishes court reporters, speech-to-text reporters, scopists and audio transcribers; each works from different inputs and applies the verbatim standard differently.

The terms are easy to blur. A court reporter writes proceedings live from machine shorthand (stenographic or Palantype) or pen shorthand. A speech-to-text reporter captions live for D/deaf and hard-of-hearing audiences. A scopist edits a reporter's raw output, and an audio transcriber works from recordings after the event. Each category has its own accreditation path, so mixing them up in preparation means training the wrong skill.

The standards also differ in timing and feedback. A scopist or transcriber can replay audio repeatedly; a live reporter or captioner gets one pass. Studying 'Commonly Confused Terms' on the BIVR site pays off twice: it sharpens your vocabulary for the discipline, and it forces you to decide which working conditions your practice should simulate — one-pass live work or multi-pass recorded work.

Disfluencies, false starts and repetitions: a worked scenario

False starts and repetitions must appear in the record as spoken; the common mistake is silently merging a restarted sentence into its corrected version.

Scenario: a witness says, 'I left the office at — no, sorry, it would have been about six, I left at six.' A frequent learner error is to transcribe, 'I left the office at about six o'clock,' merging the correction into one clean sentence. That reads smoothly and destroys the record: the witness's uncertainty and self-correction are gone. The better decision is to render the interruption and restart exactly as spoken, following whatever false-start convention you have been taught, so a reader can see the speaker changed their answer mid-sentence.

This matters because examining counsel may probe precisely that correction: 'You first said you left at a different time, didn't you?' Practice this deliberately. Take any audio containing corrections — news interviews work well — and transcribe short bursts. Then mark every point where you were tempted to smooth the text, and check whether the smoothing deleted a spoken word or inserted an unspoken one. Both count as fidelity errors even when the meaning survives.

• Render restarts and interruptions as they occurred
• Do not merge a correction into the original statement
• Practise with naturally untidy speech, not scripted reading

Uncertain words and proper nouns: protecting the record against guessing

When a word, name or figure cannot be verified, the record is protected by marking the uncertainty, never by substituting a plausible guess.

Scenario: in an arbitration a party says, 'the consultant firm, er, McKinstrath — whatever the pronunciation is.' You are unsure of the spelling. The plausible mistake is to pick the nearest familiar name, because a confident-looking wrong spelling in a transcript is treated as fact by everyone who reads it. The better decision is to transcribe phonetically as best you can and apply your convention for uncertain matter, and, if working post-event, to check exhibits, name spellings agreed at the start of proceedings, or correspondence before finalising.

This principle, sometimes framed as protecting the record, is the professional habit that separates defensible work from convenient work. Build a personal checklist: how you will render an inaudible passage, an unintelligible overlap, a number you half-heard, a name nobody spelled. Decide these before you practise, so the decision under time pressure is already made. A guessed word that happens to be right teaches you nothing; a marked uncertainty you later resolve teaches you your own listening limits.

• Never substitute a plausible guess for an unverified word
• Use your uncertain-matter or inaudible convention consistently
• Check names and figures against exhibits or agreed spellings where available

Punctuation for sense and speaker identification in the finished transcript

Punctuation in a transcript records meaning as delivered, and every paragraph change must trace to an actual speaker change or clear topic shift by that same speaker.

Because verbatim text is untidy, punctuation carries unusual weight. A comma versus a full stop can change who asserted what. Train 'punctuation for sense' as a separate skill: take a clean paragraph of prose, remove its punctuation, and repunctuate it from audio at speed, asking at each mark whether the spoken delivery — pause, falling tone, interruption — supports the stop you chose. Where delivery is genuinely ambiguous, note that the mark is your interpretation, and be ready to justify it.

Speaker identification has the same logic. In multi-party proceedings — tribunals and arbitrations often feature rapid exchanges and interruptions — you must attribute every word to the right speaker or flag attribution as uncertain. Practise with a multi-voice recording: transcribe two minutes with speaker labels, then replay and audit each label. Note every place you attributed words by context rather than by evidence; those are the labels a challenging counsel would attack first.

• Punctuate from spoken delivery, not from grammar habit
• Audit speaker labels against the audio, not against expectation
• Flag ambiguous attributions rather than resolving them silently

A speed-and-fidelity drill with a self-check rubric

Practise in short timed bursts against real recorded speech, then score the transcript against the audio by error category, watching which category grows as speed rises.

Exercise: choose a two-minute recording of unscripted speech — a radio phone-in or courtroom footage. Transcribe it in a timed pass without stopping, then replay as many times as needed to build a reference version. Score your first pass against the reference in four categories: words dropped, words substituted, unspoken words inserted, and punctuation or speaker-label errors. Worked example of the scoring habit: on a 300-word passage, three drops, two substitutions, one insertion and five punctuation queries gives you a fidelity picture you can compare week to week.

Expected observations: as your writing speed approaches the speech rate, dropped words rise first, then substitutions on unfamiliar names; inserted words usually indicate you were completing sentences from habit. If drops dominate, slow the audio slightly and hold accuracy before pushing speed; if insertions dominate, drill transcribing deliberately messy speakers. If punctuation queries exceed a handful in two minutes, drill punctuation separately rather than chasing more speed. Rubric: aim to see total word-level errors fall across sessions while you deliberately keep the speed constant — that is a learning milestone, not a prediction of any exam result.

• Transcribe timed, then build a reference version on replay
• Score four error categories: drops, substitutions, insertions, punctuation/labels
• Hold speed constant across sessions so error trends are visible

A preparation sequence and readiness checks before the exam

Sequence your preparation from fidelity to speed to exam simulation: verbatim habits first, timing second, and full one-pass conditions only in the final weeks.

A workable adaptable sequence: weeks one and two, transcribe short untidy passages untimed, building your conventions for disfluencies, inaudibles and uncertain names, and study the role distinctions between reporter, STTR, scopist and transcriber. Weeks three and four, add timed bursts with the rubric, keeping speed fixed. Weeks five and six, work one-pass only — no replay before scoring — on multi-speaker material, and finish each session by auditing speaker labels and punctuation choices. Adjust the durations to your own schedule and current speed.

Readiness checks you can actually observe: you can transcribe unscripted multi-speaker speech in one pass without inserting unspoken words; you can state and apply your convention for every uncertain-matter situation before starting a passage; you can justify each punctuation mark from spoken delivery; and your rubric error counts are flat or falling across at least three consecutive sessions. For administrative details of BIVR's examinations and membership categories, use the institute's own site, bivr.org.uk, as the authoritative reference.

• Fix your conventions before adding speed pressure
• Move to one-pass, multi-speaker conditions in the final phase
• Readiness means flat-or-falling error counts plus justifiable punctuation

Scenario two: the post-event transcript where replay becomes a trap

When transcribing from a recording, unlimited replay invites over-editing; the mistake is polishing grammar across passes and losing the original delivery.

Scenario: you hold a two-hour hearing recording and no deadline pressure. On your third pass you notice a passage of stilted, ungrammatical evidence and rewrite it into clean English, because you can — the audio is right there. The mistake is subtle: with multi-pass work, the temptation shifts from guessing to over-correcting. The better decision is to treat each pass with a defined purpose — first pass for the words, second for names and figures, third for punctuation and speaker labels — and to verify against audio only, never against what 'reads better'.

Why it matters: in a transcribed record, the difference between 'I never said that' and 'I never, I never said that' can bear on credibility findings. Multi-pass work should converge on the audio, not on house style. Apply the same read-aloud check from earlier sections to your final version: every deviation from the recording must be traceable to a spoken word, a defined convention, or a marked uncertainty. If you cannot trace it, restore the original.

• Give each replay pass one defined purpose
• Verify against audio, never against what reads better
• Restore any change you cannot trace to the recording

References and further reading

Use these references to explore the concepts and check the latest information from the relevant organizations.

• bivr.org.uk

Continue your preparation

• British Institute Of Verbatim Reporters Member Bivr Uk
• Study Guides