Lexical tone is an important phonological property in tonal languages, but its encoding process in speech production remains unclear. We conducted two electroencephalographic (EEG) experiments to investigate the time course of tonal encoding relative to that of syllabic encoding in Mandarin Chinese disyllabic and monosyllabic word production respectively. We employed a phonologically primed picture naming task and orthogonally manipulated the tonal and syllabic overlap between the prime and the target word. In both experiments, the ERP data revealed that the main effect of tonal relatedness began to emerge alone in an early time window before that of syllabic relatedness, indicating an early independent retrieval process for lexical tone. Moreover, we observed a significant interaction between tonal and syllabic relatedness in later ERP time window(s) and onset latencies, indicating a later tone-to-syllable integration process. These results support the two-stage model of tonal encoding in Mandarin word production and offer implications for current speech production models.
Sam: When a Mandarin speaker names a picture, the tone of the word may be ready first. Then it has to wait for the right syllable before it does any good. That's the core claim of a pair of brain-recording experiments on how tone gets planned in speech.
Alex: Place it for me before we go further. Whose work is this, and what kind of study are we talking about?
Sam: It's from Xiaocong Chen and Caicai Zhang at The Hong Kong Polytechnic University, published in the Journal of Memory and Language. It's two EEG experiments, recording scalp electrical activity while people named pictures out loud.
Alex: I work near production research but not on tone. Why should someone like me care when this seems like a niche question about one language?
Sam: Because the main models of speaking were built mostly on non-tonal European languages, and they barely say how tone is planned. This study offers timing evidence for where tone fits in that sequence. So we'll cover what people believed, the design trick that makes it work, what they found, and how far to trust it.
Alex: Start with the background. What's lexical tone, and what was the disagreement?
Sam: Lexical tone is a pitch pattern that changes word meaning. In Mandarin, the same syllable "ma" means mother, hemp, horse or scold depending on the tone. One camp says tone is retrieved on its own, like a consonant or vowel. The other says it behaves more like stress in English. It's part of a rhythmic frame that gets attached to the syllable.
Alex: And there's a timing question layered on top of that, I assume. When does tone show up relative to the syllable?
Sam: Right, and the evidence pointed both ways. Some reaction-time studies hinted tone is available early. But earlier EEG work suggested it comes late, after the syllable. Those EEG studies used a decision task, though, not actual speaking. And a recent overt-naming EEG study in Cantonese also found tone came late, but the authors note its stimuli weren't perfectly controlled.
Alex: So the gap is a clean test in real, overt Mandarin speech. How did they set it up?
Sam: The one idea is to cross two kinds of overlap. Before each picture, people saw and heard a one-syllable prime, a word they were told to ignore. The prime could share the tone with the picture name, share the syllable, share both, or share neither. Take the picture of corn, "yu4 mi3." The prime "yu4" shares both syllable and tone with the first syllable. "Yu2" shares only the syllable. "Xiang4" shares only the tone. And "ge2" shares nothing.
Alex: So the tone-only prime is the interesting one. If tone is retrieved independently, that prime should show some effect on its own.
Sam: Exactly what they tested. And if tone has to be fused with the syllable, the effects of tone and syllable should depend on each other. Statistically, that's an interaction. They ran this once with two-syllable words, then repeated it with one-syllable words.
Alex: What did the naming times show? That's the part people outside EEG will trust most.
Sam: The behaviour was the same in both experiments. Sharing the syllable always helped. Sharing the tone helped too, but only when the syllable was also shared. The full match beat the syllable-only prime by about 52 milliseconds with two-syllable words. With one-syllable words, the gap was about 38.
Alex: And when only the tone matched? I'd guess no effect.
Sam: It was worse than nothing. The tone-only prime made people slightly slower than a completely unrelated prime, in both experiments. So tone overlap helps or hurts depending on whether the syllable matches.
Alex: That sounds like the integration camp wins. Tone can't do anything alone. Where does the "tone first" claim come from?
Sam: From the timing in the brain signal. They looked at event-related potentials, which are averaged voltage changes locked to the moment the picture appears. In both experiments, a tone effect appeared on its own before any syllable effect. With one-syllable words, it showed up around 180 to 250 milliseconds after the picture. With two-syllable words, it came a little later. The syllable effect appeared mostly from about 320 milliseconds.
Alex: To put that in context, how long did people take to actually start speaking?
Sam: On average, roughly three quarters of a second. So the tone effect is very early in the planning. Then, in later windows, tone and syllable started to interact, mirroring the naming times. The authors read that as tone-to-syllable integration: the tone is chosen early, then attached to the syllable later.
Alex: So the story reconciles both camps. Early independent retrieval, then fusion. How do they explain the tone-only prime slowing people down?
Sam: They offer it as a possible account. The tone-only prime is a real word, so it may activate a competing syllable. Then the speaker has to suppress that syllable before attaching the tone to the right one. They suggest this is why another study, using nonsense syllables as distractors, found tone-only overlap could help early.
Alex: Now the trust question. Two experiments, about forty and about fifty people. Did the brain results actually replicate?
Sam: Partly. The timing pattern did: tone first, interaction later, in both. But the shape didn't. In the two-syllable experiment, the early tone effect was negative and sat only in the left front of the scalp. In the one-syllable experiment, it was positive and spread across front and central areas.
Alex: Opposite polarity is not a small difference. Do the authors take that seriously?
Sam: They do. They suggest two-syllable words may recruit different brain areas, but they say scalp EEG can't settle it. They also say the link between behaviour and brain data wasn't transparent. And they flag that tone coming before the syllable is a new finding needing replication. They also note priming tasks themselves can invite strategies.
Alex: Does tone-first clash with the established idea that the bare syllable is Mandarin's first planning unit?
Sam: They argue it doesn't have to. They keep the atonal syllable, meaning the sounds without the tone, as the main planning unit. Tone, in their view, travels a separate route as part of a frame. It can be activated early but can't finish until the syllable is ready. But they admit the leading models assume tone and syllable start in parallel, not tone first.
Alex: Would this hold for Cantonese, given that earlier study found the opposite?
Sam: They don't know, and say so. They speculate that Mandarin speakers learn Pinyin, the romanization that marks tones explicitly, which might make tone categories more separable. They also point to a Mandarin rule where one tone changes before another, which could reward planning tone early. Both are hypotheses for future comparisons.
Alex: So who should actually sit down with the full paper, and where should they start?
Sam: Anyone studying speech production, especially in tonal languages. Start with the section comparing the two experiments' brain results. That's where the agreement and the mismatch are laid out honestly. Then read the theoretical implications, where they fit tone-first into existing models.
Alex: And for everyone else, the one line to remember?
Sam: In Mandarin, tone may be fetched first, on its own. But it only speeds you up once it has the right syllable to land on.
Alex: Tone first, syllable second, and a call for replication.