Connected Speech and Assimilation

You can know every suffix and still not understand spoken Turkish — because natural speech systematically reduces the careful forms you learned. A native speaker does not pronounce every letter of geliyorum or carefully separate ne haber; they glide, drop, and merge in patterns that are entirely regular but unwritten. The gap between the written form (which you study) and the heard form (which you must decode) is the single biggest reason advanced learners still freeze in real conversation. This page maps the most important of those reductions so you can train your ear to expect them. Crucially, these forms are spoken, not standard orthography: you should recognise and produce them aloud, but you keep writing the full forms.

The flagship reduction: the dropped r of -(I)yor

The present-continuous suffix -(I)yor is the most frequent verb ending in conversational Turkish, and in casual speech its final r is routinely dropped. Geliyor "he's coming" becomes geliyo; ne yapıyorsun "what are you doing" becomes napıyosun (combining several reductions at once). This single pattern, once you internalise it, unlocks a huge proportion of spoken speech.

The reason this matters so much is statistical: because -(I)yor is on so many verbs, a learner who is waiting to hear a clean "...yor" sound will mishear nearly every present-tense sentence. Retrain the expectation: the default spoken shape of this suffix is -(I)yo, with the r resurfacing only in careful, slow, or formal delivery.

Vowel elision across word boundaries

When one word ends in a vowel and the next begins with one — or when a high vowel sits in an unstressed syllable between consonants — fast speech tends to swallow a vowel, fusing two words into one rhythmic unit. The most famous case is the everyday greeting itself.

The word bir "one / a(n)" deserves special mention: as the indefinite article it is almost always unstressed and almost always reduced to bi in speech. A learner reading bir kahve expects "bir," but hears "bi kahve" — and momentarily loses the word. Likewise the question word ne "what" elides relentlessly into whatever follows (naber, noldu, napıyosun, napalım), which is why so many of the most common spoken phrases start with an n that has no obvious source on the page.

Voicing and place assimilation across boundaries

Turkish suffixes already encode a lot of assimilation in the writing system itself — the past tense surfaces as -di or -ti, the locative as -da or -ta, depending on the voicing of the preceding consonant. That is regressive voicing assimilation baked into the orthography, and you have met it as the "D hardens to T" rule. In connected speech, similar adjustments happen across word boundaries, where they are never written.

These cross-boundary assimilations are subtler than the -(I)yor drop and you do not need to produce them deliberately — they happen naturally once your rhythm speeds up. The value in noticing them is comprehension: a phrase you "know" can sound unfamiliar simply because its consonants have softened into each other at the seams.

Reduction of common multi-word formulae

A whole inventory of high-frequency phrases has fused so thoroughly that the reduced form is the normal spoken form, and the full form sounds stiff. Recognising these as units — rather than parsing them letter by letter — is essential for following casual speech.

Register: when reductions are appropriate — and when they are not

This is where a B2 learner needs nuance. Reductions are not "incorrect Turkish"; they are register-marked. They belong to casual, spoken, informal contexts and are out of place in formal ones.

• (informal / colloquial): dropped -(I)yor r, naber, bi şey, hişşey — the default in friendly conversation, texting, social media.
• (formal / careful speech): full geliyorum, bir şey, ne haber — used in a job interview, a news broadcast, a speech, addressing someone with siz.
• (written standard): you always write the full form. Naber and geliyo appear in writing only to deliberately represent casual speech (in fiction dialogue, chat messages, song lyrics).

Why training your ear is the real task

The deep point of this page is that your bottleneck at B2 is rarely grammar — it is the mismatch between the form you encoded and the form you hear. You learned geliyorsun and stored it as eight clean sounds; the speaker delivers four or five. Your brain searches for the stored shape, fails to match, and you lose the thread while the conversation moves on. The cure is not more vocabulary but re-encoding the spoken shapes alongside the written ones: when you learn yapıyorsun, also learn that it is heard as yapıyosun and, with ne, as napıyosun. Listening to unscripted Turkish — podcasts, vlogs, dramas — with this expectation is what closes the gap. The patterns are finite and systematic; once you have heard each one a dozen times, the heard form and the written form fuse into a single entry, and spoken Turkish stops sounding like a different language from the one you studied.

Common mistakes

The two opposite traps: expecting full forms (and so mishearing fast speech), and using reduced forms in the wrong register (and so sounding too casual). Comprehend the reductions everywhere; produce them only where they fit.

Key takeaways

• Spoken Turkish systematically reduces the careful written forms; the gap between them is the main barrier to understanding fast speech.
• The flagship reduction is the dropped r of -(I)yor: geliyor → geliyo, yapıyorsun → yapıyosun — the spoken default, not an error.
• Vowel elision fuses words at boundaries, driven especially by ne "what" (naber, n'oldu, napıyosun) and bir "a" (→ bi).
• Cross-boundary voicing/place assimilation softens consonants at word seams; you needn't produce it deliberately, but it explains why known phrases sound unfamiliar.
• These forms are register-marked (informal/spoken): recognise them everywhere, produce them only casually, and always write the full form.