TOEIC Link Speaking — Pause and Filler Management: How Silent-Pause Placement, Filler Substitution, and Rhythm Control Move the Fluency Band from 5 to 8

Pause and filler behavior is the most audible fluency discriminator on the TOEIC Link speaking module, and it is the one candidates most consistently misdiagnose. The intuition most test-takers carry into the exam is that silence is failure — that any gap between words signals a breakdown the rater will penalize — and so they fill every gap with um, uh, like, or a trailing so... to prove the machinery is still running. The result is the opposite of what they intend. A response saturated with filled pauses reads to a rater as less fluent than a response with fewer, well-placed silent pauses, because filled pauses cluster at exactly the points where the speaker has lost control, while silent pauses in a fluent speaker fall at clause and idea boundaries where they signal structure. The fluency band does not reward the absence of pauses. It rewards pauses that land where a competent speaker would put them.

This distinction — functional pause versus hesitation pause — is the entire game. A candidate stuck at band 5 typically pauses mid-clause, mid-phrase, and mid-word, and masks each gap with a filler. A candidate at band 8 pauses between clauses, between idea units, and before a deliberate emphasis, and leaves most of those pauses silent. The raw pause count may be similar. The placement is completely different, and placement is what the rater's ear tracks.

The two pause types raters distinguish

Functional pauses

A functional pause falls at a syntactic or rhetorical boundary: at the end of a clause, before a contrasting idea, or ahead of a word the speaker wants to stress. Functional pauses are short — typically 0.3 to 0.8 seconds — and they are silent. To a rater they are invisible as errors because they mirror the prosody of a confident native or advanced speaker. In fact, a response with no functional pauses reads as rushed and flat, which caps prosodic scoring even when word-per-minute rate is high. Functional pauses are not something to eliminate; they are something to place.

Hesitation pauses

A hesitation pause falls inside a structural unit: mid-clause, mid-phrase, between an article and its noun, or between a verb and its object. Hesitation pauses signal retrieval failure — the speaker is searching for the next word or repairing a grammatical choice — and they are the ones that depress the fluency band. Crucially, hesitation pauses are the ones candidates instinctively fill, because the silence feels intolerable. So the filled-pause problem and the hesitation problem are the same problem wearing two costumes: the fix is not to fill the pause better but to move retrieval earlier so the pause never lands mid-unit in the first place. For the underlying recovery mechanics, the speaking fluency and hesitation recovery guide covers how to re-enter a stalled clause without restarting it.

Why filled pauses cost more than silence

Filled pauses carry three penalties silent pauses do not. First, they are marked — um and uh draw the rater's attention to the gap, whereas a 0.5-second silence often passes unregistered. Second, they cluster at retrieval failures, so a filler is a reliable signal to the rater that the speaker was searching, not structuring. Third, they chain: one filler tends to trigger a run of them (um... so... like... yeah...) as the speaker stalls, and a chain reads as a mini-breakdown even when the eventual content is fine. The hesitation marker and filler reduction guide breaks down the specific markers raters weight most heavily.

The correction is not to suppress fillers by force of will — suppressed fillers usually convert into equally penalized lengthened sounds (theee, aaand) or hard glottal stops. The correction is substitution and displacement, covered next.

The four substitutions

1. Filler to silence. The primary move: allow the hesitation pause to be silent rather than filled. A 0.6-second silence is scored better than 0.6 seconds of um. Trainable by recording responses and deleting fillers in review to hear that the silence alone reads as more controlled.

2. Mid-unit filler to boundary pause. When a filler appears mid-clause, the underlying cause is that planning ran out before the clause did. The fix is to pause before starting the clause — at the boundary — so the whole clause can be produced in one uninterrupted run. This displaces the pause from inside the unit to the edge of it.

3. Filler to discourse marker. A small set of genuine discourse markers (Now, So, The key point is, What matters here is) can occupy the boundary slot a filler would otherwise take, buying planning time while adding structure rather than noise. Unlike um, these are content-bearing and score neutrally-to-positively. Use sparingly — three or four per two-minute response — or they become fillers themselves.

4. Filler to controlled lengthening of a stressed word. A deliberate, prosodically-motivated lengthening on a content word (that was a major problem) reads as emphasis, not hesitation, and covers a fraction of a second of planning. This is the most advanced substitution and only works when it lands on a genuinely stress-worthy word.

The four-week protocol

Week 1 — Diagnosis. Record ten one-minute responses to standard TOEIC Link prompts. Transcribe each and mark every pause as functional (F) or hesitation (H), and every filler. Count the H-to-F ratio and the fillers-per-minute rate. Most candidates start at 8 to 15 fillers per minute with an H-heavy ratio. The goal by week 4 is under 4 fillers per minute with an F-dominant ratio. Do not try to change delivery yet — only measure.

Week 2 — Boundary placement. Practice the substitution 2 drill: take the same prompts and force every pause to the start of a clause. Speak one clause fully, stop at the boundary, plan the next clause in silence, then produce it. It will feel unnaturally staccato at first. The point is to retrain where retrieval happens — before the clause, not inside it. Pacing mechanics that support this are in the response timing and fluency pacing guide.

Week 3 — Silence tolerance. Record responses and, in review, delete every filler you produced. Listen to the silent-pause version. This rewires the intuition that silence equals failure by proving the silent version sounds more fluent. Then re-record, consciously leaving hesitation pauses silent. Fillers-per-minute should drop sharply this week as the tolerance builds.

Week 4 — Integration under time pressure. Return to full two-minute responses with normal planning time. Track fillers-per-minute and H-to-F ratio across ten more recordings. The target is fewer than 4 fillers per minute, an F-dominant pause profile, and no filler chains longer than one. At this point pause management is no longer a conscious effort mid-response — it has become the default rhythm, which is the only state that survives the cognitive load of the real exam.

Common failure modes

• Over-correction to zero pauses. Eliminating all pauses produces a flat, rushed delivery that caps prosodic scoring. Functional pauses are required, not optional.
• Converting fillers to lengthened sounds. Suppressing um often produces aaand or theee, which score identically. The fix is boundary displacement, not suppression.
• Discourse markers becoming fillers. Overusing So or Now turns them into the very hesitation signals they were meant to replace. Cap them.
• Practicing only in isolation. Pause control learned at slow, low-load practice speed collapses under real-exam cognitive load unless week 4 integration is done at full length and full pressure.

Pause and filler management is not about speaking without stopping. It is about stopping in the right places, silently, so that every pause the rater hears reads as a competent speaker organizing thought rather than a struggling one running out of it. Move the pauses to the boundaries, let them be silent, and the fluency band moves with them.