Formal Register Dictation: Kill the Contractions on the Pass

The first formal register dictation session produces a draft with a conversational accent, and the reason has nothing to do with your knowledge of the topic. Human speech contracts naturally. Nobody says I do not think when the thought is quick; the mouth produces I don't think, and it's, can't, won't, and the gap between the spoken and the written form is exactly the space every academic paper is supposed to cross. A dictated draft is a transcript of a mind at speaking speed, so it arrives carrying every contraction you used, and the rhythm of conversation along with it. This is not a failure of technique; it is the truthful output of the channel.

The surprise for most students is that the informal passages are not the ones they expect. Long, careful paragraphs dictated slowly sometimes emerge almost fully formal, while short asides, signals like well, honestly, kind of, and little hedges that speech uses to negotiate agreement, land in the text unfiltered. A dictation tool will happily write I guess that the model works if you say it, and the sentence will pass unnoticed during the session, because in the moment it sounds like the sentence you meant. Only the read-back reveals the register gap. This is why the fix must be a separate pass rather than a correction made in real time.

There is a second reason the draft drifts informal beyond mechanics: spoken language is social. When we talk, we build rapport, soften statements, and check how the listener is receiving us, and the residue of all that care lands on the page as contractions and hedges. Academic writing has different social rules, and a dictated draft sits in the awkward middle. The solution is not to abandon dictation, which is the fastest way to get the thinking down, but to formalize deliberately, in a phase you control. Think of the draft as a rehearsal and the expansion pass as the performance, each with a different instrument: recording for thought, and revision for tone.

Contractions are the visible signature, and they leak from everywhere at once. The word archive is forgiving, and the fix must be systematic because a manual scan for don't and can't will always miss the ones where the contraction is part of a longer phrase, like I'd say or it's worth noting. Less visible but just as damaging are the fillers: well, so, actually, basically, you know, and the stalling sounds that a recorder sometimes catches. These are not errors, they are rhythm, but in formal prose they read as hesitation. A useful practice is to scan the transcript for their cousins as well, words like stuff, things, a lot, and pretty, which are exact but unacademic.

The asides are the third leak and the most insidious, because they are the sentences that explain what the writer is doing rather than what the argument says. A spoken draft often contains phrases like let me move on to the next point, I think this is important, and the thing I wanted to add, which are stage directions of the mind. In conversation they are invisible; in prose they break the fourth wall. Some can be deleted outright. Others must be converted, so that this is worth emphasizing becomes the emphasis itself, a sentence that does the work the aside was announcing. Transcription rarely preserves the distinction, which is why the pass below is worth doing by ear.

Finally, the leak that students never notice until it is pointed out: impersonal constructions disappear. Spoken English loves a person, and the dictated draft therefore fills with I, we, and even you, where formal writing prefers it, one, the paper, or a noun phrase. The shift is more than pronoun substitution; it is a whole change of stance, from someone reporting an opinion to a text presenting a finding. Hear it in a sentence like when we look at the data, we see that the effect fades. De-contracting alone would not fix that sentence. The register pass must therefore work at four levels at once: word form, filler, aside, and subject.

The reliable fix is a second recording session, not a search and replace. Open the draft, read each sentence aloud, and when your mouth produces a contracted form, stop and say the sentence again with the words fully spelled: do not, cannot, it is, they would. Speak the corrected version in full before moving on, then transcribe from your own corrected reading, or simply edit the text directly the moment your ear flags the gap. The doubling matters, because hearing the formal version changes how the next sentence will be said, and a session transforms the register of the whole document rather than patching isolated instances.

The expansion pass needs a rule to keep it from becoming a hunt. Work sentence by sentence, top to bottom, and say each sentence once in its spoken form, waiting to judge it aloud. You are listening for three categories at once: contractions, which need their full forms; fillers, which need deletion; and asides, which need conversion. Make the edit in the same moment, and if a sentence contains more than one issue, fix the worst first and reread once more. Twenty sentences takes less than fifteen minutes of practice, and the second session is noticeably faster than the first, because your ear has been trained to notice the pattern it previously skipped.

The pass also lifts the register past the sentence level. A paragraph that was a chain of short conversational clauses reforms itself as you read, because the speech rhythm of formal prose is longer and slower, and your voice leads the writing toward it. You will find yourself combining sentences that dictation had split, replacing a series of quick statements with one main clause and its qualifiers. This is the real payoff. The result no longer sounds like someone who dictated an essay and then corrected it; it sounds like someone who wrote it carefully, because the writing was made in a voice that already knew the register it wanted.

The comparison is worth rehearsing on a real paragraph. Dictated, a discussion of housing policy might come out as: we don't really know why rents go up, but I think it's clear that supply isn't the whole story. Universities would say that it's more complicated than that. Move through the expansion pass and the same thought becomes: the causes of rising rents remain contested, although the evidence suggests that supply alone does not explain the trend. Policymakers contend that the issue resists a single explanation. The second version says the same thing with less noise, and notice that the words were not replaced with fancier ones; the register shifted through expansion, deletion, and a change of subject.

Three changes give the paragraph its new tone, and each is teachable. First, every contraction was expanded, so no spoken shorthand survives in the sentence. Second, the hedges and asides, I think, really, honestly, let me say, were cut, and their residue was folded into the claim itself. Third, the subject shifted from we and I to the causes, the evidence, and policymakers, so the sentence argues from a standpoint rather than from a person. The vocabulary is slightly more formal but stays plain, which is the strongest argument for the method: you do not need a thesaurus to sound academic, you only need to remove the sounds of chat.

Read the two versions back to back and the effect compounds across a long document. A single contraction is invisible; seventy of them, especially when paired with fillers and asides, produce a paragraph a reader describes as casual even if they cannot say why. The de-contracted paragraph, read aloud, feels different in the mouth as well, and that feeling is a useful quality check. After the pass, read the paragraph once more into the recorder. If the second reading is clean, the register holds. If it drifts, the remaining informal elements are still in the text, and your ear just told you which sentence to fix.

The cheapest fix beats the whole pass: set the register before you record. Open a session by saying a short formal framing aloud, not as content but as a threshold: this paragraph examines the evidence on X, and the argument advances in three stages. The wording is almost comically stiff, which is the point. It changes your speaking rhythm in about ten seconds, because the mouth, once it has produced a formal sentence, tends to stay there. Writers who use this opening ritual report fewer corrections in the expansion pass, and some report that the draft arrives nearly clean. The trick is not to sound academic; it is to warm up the muscle that produces it.

A second prompt takes the form of a one-line written instruction placed above the dictation area, visible while you speak: no contractions, no fillers, no first person unless the assignment allows it. Reading the line before each listening session is optional, but a simple reminder does more work than most writers expect, because the constraint moves from the back of your head to the front. If you dictate into a tool with an instruction pane or a custom text field, keep the line there permanently. And for writers who prepare with an outline, write the outline's briefs in full prose rather than bullet fragments; a formal prompt produces a formal reply, which is dictation's own version of consistency.

The final habit is the verify loop: after recording, read the transcript aloud verbatim, without editing, and mark as you go where the voice changed. Mark the first contraction you catch, the first filler, and the first aside. Three marks in a ten-minute session is normal; seven is a session worth a redo. Over two weeks the marks fall, and the interesting graph is that improvement happens fastest in the first session of the day, meaning the true asset being trained is not a rule but a switch. Once that switch is in your hands, dictation becomes a formal-writing instrument, and the expansion pass shrinks to a ten-minute confirmation instead of a repair job.