EnglishBlitz

TOEIC Link Speaking — Hesitation Marker and Filler Reduction for Fluency Scoring: How Replacing Um-and-Uh With Structured Silence Lifts the Fluency Sub-Score From 22 to 27

Fillers are not the cause of low fluency scores — they are the symptom of unplanned speech, and cutting them by suppression alone makes speech worse, not better. This guide separates the four filler functions, the planned-pause substitution that replaces them without introducing dead air, and the six-week routine that converts anxious hesitation into rubric-scoring controlled delivery.

EnglishBlitz Editorial Team·

TOEIC Link Speaking — Hesitation Marker and Filler Reduction for Fluency Scoring: How Replacing Um-and-Uh With Structured Silence Lifts the Fluency Sub-Score From 22 to 27

Hesitation markers — um, uh, er, and the stretched soooo and liiike — are the most visible fluency problem in TOEIC Link speaking responses, and the most misunderstood. Candidates who know they overuse fillers usually try to fix the problem by suppression: they consciously clamp down on every um. This almost always makes the response worse. The filler was doing a job — buying planning time — and when it is removed without a replacement, the planning time is removed with it. The result is either a broken-off sentence or a longer, more damaging silence with visible panic in it. Suppression treats the symptom and amputates the function.

The fluency sub-score carries substantial weight in the TOEIC Link speaking rubric, and raters are trained to distinguish two very different kinds of pause. A planning pause — a brief, silent, controlled break at a clause boundary — reads as fluent, even confident. A hesitation pause — a filled um in the middle of a phrase, or a stalled silence mid-clause — reads as disfluent. The two pauses can be the exact same length. What separates them is placement and whether they are filled. The entire skill of filler reduction is learning to move your pauses to clause boundaries and let them be silent, rather than fighting to eliminate pausing altogether. For how pausing interacts with connected discourse, see the discourse cohesion and transition signal deployment guide.

Why suppression backfires

When a candidate suppresses fillers without changing anything else, one of two things happens. Either the planning pressure that produced the filler now produces a mid-phrase freeze — a silence in the worst possible place, right in the middle of a noun phrase, which reads as a memory failure — or the candidate rushes to avoid the freeze and produces a run-on that loses grammatical control. Both are lower-scoring than the original filled pause. The filler, for all its ugliness, was at least keeping the floor and signaling "I am still constructing this." Removing it removes that signal.

The correct model is not "eliminate pauses." It is "convert filled hesitation pauses into silent planning pauses, and move them to boundaries." You will pause roughly the same total amount. The pauses will simply be in different places and shaped differently, and that difference is the entire fluency gain.

The four functions fillers serve

To replace a filler you have to know what it was doing. There are four functions, and each has a different fix.

Function 1 — Lexical retrieval

The filler buys time to retrieve a word. "The main benefit is, um, efficiency." The fix is not to remove the um but to move the pause earlier, to the clause boundary before is: "The main benefit — [silent pause] — is efficiency." The retrieval still happens; it just happens during a silence at a natural break rather than during a filled break mid-phrase.

Function 2 — Syntactic planning

The filler buys time to plan the shape of the upcoming clause. "And so, uh, what I would do is..." The fix is a planned transition phrase that legitimately occupies the planning time: "And so what I would do — " with a brief silent pause after do. The transition itself gives you the planning window without a filler.

Function 3 — Floor-holding

The filler signals "I am not finished, do not think I have stopped." In a live conversation this matters; in a timed monologue response it does not — nobody is going to interrupt you. Floor-holding fillers are pure waste in the exam context and can be dropped entirely, because there is no floor to defend.

Function 4 — Anxiety discharge

The filler is a nervous tic with no linguistic function at all — it fires on every clause boundary regardless of whether you need planning time. This is the most trainable function, because it is not tied to any real cognitive load. It responds directly to boundary-pause drilling: once you have a trained silent pause to occupy the boundary, the anxiety filler has nowhere to fire.

The planned-pause substitution

The core technique is to install a silent boundary pause as your default clause transition. Instead of clause + um + clause, you train clause + [silent pause] + clause. The silent pause does everything the filler did — buys retrieval and planning time, marks the boundary — but reads as control rather than hesitation because it is silent and correctly placed.

Three rules make the substitution work:

  1. Pause at boundaries, never inside phrases. A silent pause between two clauses reads as fluent. The same silent pause inside a noun phrase reads as a freeze. Placement is everything.
  2. Keep the pause short and even. A boundary pause of roughly half a second reads as natural rhythm. A pause that stretches past a second, or that varies wildly in length, reads as struggle. Train an even, brief boundary pause, not a dramatic one.
  3. Do not fill the pause with breath noise. An audible in-breath or a stretched final vowel (soooo) reconverts your silent pause back into a filled hesitation. The pause must be genuinely silent.

The six-week routine

Filler reduction is motor retraining, not knowledge acquisition. You cannot think your way out of a filler in real time; you have to install a competing habit.

  • Weeks 1–2 — Awareness. Record five practice responses and transcribe them, marking every filler and every mid-phrase freeze. Count them. The goal here is not reduction yet — it is to make the fillers audible to yourself, because most candidates genuinely do not hear their own ums until they transcribe.
  • Weeks 3–4 — Boundary drilling. Practice speaking in deliberately short clauses with an exaggerated silent pause at every boundary. It will feel unnaturally choppy. That is correct — you are overcorrecting to install the silent-pause reflex, and it will relax into a natural rhythm later. Record and confirm the pauses are silent and at boundaries.
  • Weeks 5–6 — Integration. Speak at normal pace and let the boundary pause fire on its own. Record, transcribe, and count fillers again. The target is a filler rate cut by more than half, with the remaining pauses silent and boundary-placed. Do not aim for zero pauses — aim for zero filled pauses inside phrases.

The recovery move for when a filler slips out

You will still produce fillers under pressure, and the difference between a strong and a weak candidate is what happens next. The weak move is to react to the filler — flinch, apologize, restart. That compounds one filler into a visible disfluency cascade. The strong move is to treat the filler as if it were a boundary pause: complete the current clause cleanly and move on as though nothing happened. A single um in an otherwise controlled response costs almost nothing. A single um that triggers panic costs the whole fluency sub-score. Training the non-reaction is as important as training the pause itself. For how listeners decode spontaneous hesitation and self-repair — useful for understanding what raters are actually hearing — see the disfluency marker and self-repair decoding guide, and for connecting clauses cleanly once the pauses are controlled, the discourse markers and cohesion guide.

Fluency is not the absence of pausing. It is pausing in the right places, silently, at a controlled length — and the candidates who understand that stop fighting their pauses and start placing them.