Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

A speech planning network for interactive language use


During conversation, people take turns speaking by rapidly responding to their partners while simultaneously avoiding interruption1,2. Such interactions display a remarkable degree of coordination, as gaps between turns are typically about 200 milliseconds3—approximately the duration of an eyeblink4. These latencies are considerably shorter than those observed in simple word-production tasks, which indicates that speakers often plan their responses while listening to their partners2. Although a distributed network of brain regions has been implicated in speech planning5,6,7,8,9, the neural dynamics underlying the specific preparatory processes that enable rapid turn-taking are poorly understood. Here we use intracranial electrocorticography to precisely measure neural activity as participants perform interactive tasks, and we observe a functionally and anatomically distinct class of planning-related cortical dynamics. We localize these responses to a frontotemporal circuit centred on the language-critical caudal inferior frontal cortex10 (Broca’s region) and the caudal middle frontal gyrus—a region not normally implicated in speech planning11,12,13. Using a series of motor tasks, we then show that this planning network is more active when preparing speech as opposed to non-linguistic actions. Finally, we delineate planning-related circuitry during natural conversation that is nearly identical to the network mapped with our interactive tasks, and we find this circuit to be most active before participant speech during unconstrained turn-taking. Therefore, we have identified a speech planning network that is central to natural language generation during social interaction.

This is a preview of subscription content, access via your institution

Access options

Buy article

Get time limited or full article access on ReadCube.


All prices are NET prices.

Fig. 1: Cortical responses during interactive speech.
Fig. 2: Spatial organization of speech planning responses.
Fig. 3: Linguistic selectivity of planning responses.
Fig. 4: Planning activity during natural conversation.

Data availability

The data used in these analyses are not publicly available owing to concerns regarding patient privacy; however, the corresponding author will provide deidentified primary data upon request.

Code availability

The corresponding author will provide the MATLAB code used in this study for analysis of ECoG and behavioural data upon request.


  1. Sacks, H., Schegloff, E. A. & Jefferson, G. A simplest systematics for the organization of turn-taking for conversation. Language 50, 696–735 (1974).

    Google Scholar 

  2. Levinson, S. C. & Torreira, F. Timing in turn-taking and its implications for processing models of language. Front. Psychol. 6, 731 (2015).

    PubMed  PubMed Central  Google Scholar 

  3. Stivers, T. et al. Universals and cultural variation in turn-taking in conversation. Proc. Natl Acad. Sci. USA 106, 10587–10592 (2009).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  4. Schiffman, H. R. Sensation and Perception: An Integrated Approach (Wiley, 2001).

  5. Flinker, A. et al. Redefining the role of Broca’s area in speech. Proc. Natl Acad. Sci. USA 112, 2871–2875 (2015).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  6. Basilakos, A., Smith, K. G., Fillmore, P., Fridriksson, J. & Fedorenko, E. Functional characterization of the human speech articulation network. Cereb. Cortex 28, 1816–1830 (2018).

    PubMed  Google Scholar 

  7. Mirman, D., Kraft, A. E., Harvey, D. Y., Brecher, A. R. & Schwartz, M. F. Mapping articulatory and grammatical subcomponents of fluency deficits in post-stroke aphasia. Cogn. Affect. Behav. Neurosci. 19, 1286–1298 (2019).

    PubMed  PubMed Central  Google Scholar 

  8. Guenther, F. H. Neural Control of Speech (MIT, 2016).

  9. Sahin, N. T., Pinker, S., Cash, S. S., Schomer, D. & Halgren, E. Sequential processing of lexical, grammatical, and phonological information within Broca’s area. Science 326, 445–449 (2009).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  10. Broca, P. Remarques sur le siege de la faculté du langage articulé, suivies d’une observation d’aphémie (perte de la parole). Bull. Mem. Soc. Anat. Paris 36, 330–356 (1861).

    Google Scholar 

  11. Chang, E. F. et al. Pure apraxia of speech after resection based in the posterior middle frontal gyrus. Neurosurgery 87, E383–E389 (2020).

    PubMed  PubMed Central  Google Scholar 

  12. Brass, M. & von Cramon, D. Y. The role of the frontal cortex in task preparation. Cereb. Cortex 12, 908–914 (2002).

    PubMed  Google Scholar 

  13. Sierpowska, J. et al. Involvement of the middle frontal gyrus in language switching as revealed by electrical stimulation mapping and functional magnetic resonance imaging in bilingual brain tumor patients. Cortex 99, 78–92 (2018).

    PubMed  Google Scholar 

  14. Levinson, S. C. Turn-taking in human communication-origins and implications for language processing. Trends Cogn. Sci. 20, 6–14 (2016).

    PubMed  Google Scholar 

  15. Indefrey, P. The spatial and temporal signatures of word production components: a critical update. Front. Psychol. 2, 255 (2011).

    PubMed  PubMed Central  Google Scholar 

  16. Schuhmann, T., Schiller, N. O., Goebel, R. & Sack, A. T. The temporal characteristics of functional activation in Broca’s area during overt picture naming. Cortex 45, 1111–1116 (2009).

    PubMed  Google Scholar 

  17. Ferpozzi, V. et al. Broca’s area as a pre-articulatory phonetic encoder: gating the motor program. Front. Hum. Neurosci. 12, 64 (2018).

    PubMed  PubMed Central  Google Scholar 

  18. Alario, F. X., Chainay, H., Lehericy, S. & Cohen, L. The role of the supplementary motor area (SMA) in word production. Brain Res. 1076, 129–143 (2006).

    CAS  PubMed  Google Scholar 

  19. Ramanarayanan, V., Goldstein, L., Byrd, D. & Narayanan, S. S. An investigation of articulatory setting using real-time magnetic resonance imaging. J. Acoust. Soc. Am. 134, 510–519 (2013).

    ADS  PubMed  PubMed Central  Google Scholar 

  20. Bogels, S., Magyari, L. & Levinson, S. C. Neural signatures of response planning occur midway through an incoming question in conversation. Sci Rep. 5, 12881 (2015).

    ADS  PubMed  PubMed Central  Google Scholar 

  21. Ferreira, F. & Swets, B. How incremental is language production? Evidence from the production of utterances requiring the computation of arithmetic sums. J. Mem. Lang. 46, 57–84 (2002).

    Google Scholar 

  22. Wagner, V., Jescheniak, J. D. & Schriefers, H. On the flexibility of grammatical advance planning during sentence production: effects of cognitive load on multiple lexical access. J. Exp. Psychol. Learn. Mem. Cogn. 36, 423–440 (2010).

    PubMed  Google Scholar 

  23. Dubey, A. & Ray, S. Cortical electrocorticogram (ECoG) is a local signal. J. Neurosci. 39, 4299–4311 (2019).

    CAS  PubMed  PubMed Central  Google Scholar 

  24. Cheung, C., Hamiton, L. S., Johnson, K. & Chang, E. F. The auditory representation of speech sounds in human motor cortex. eLife 5, e12577 (2016).

    PubMed  PubMed Central  Google Scholar 

  25. Glanz Iljina, O. et al. Real-life speech production and perception have a shared premotor-cortical substrate. Sci. Rep. 8, 8898 (2018).

    ADS  PubMed  PubMed Central  Google Scholar 

  26. Cisek, P. & Kalaska, J. F. Neural mechanisms for interacting with a world full of action choices. Annu. Rev. Neurosci. 33, 269–298 (2010).

    CAS  PubMed  Google Scholar 

  27. Ray, S. & Maunsell, J. H. Different origins of gamma rhythm and high-gamma activity in macaque visual cortex. PLoS Biol. 9, e1000610 (2011).

    CAS  PubMed  PubMed Central  Google Scholar 

  28. Flinker, A., Chang, E. F., Barbaro, N. M., Berger, M. S. & Knight, R. T. Sub-centimeter language organization in the human temporal lobe. Brain Lang. 117, 103–109 (2011).

    CAS  PubMed  Google Scholar 

  29. Bouchard, K. E., Mesgarani, N., Johnson, K. & Chang, E. F. Functional organization of human sensorimotor cortex for speech articulation. Nature 495, 327–332 (2013).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  30. Cogan, G. B. et al. Sensory-motor transformations for speech occur bilaterally. Nature 507, 94–98 (2014).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  31. Kotz, S. A. et al. Lexicality drives audio-motor transformations in Broca’s area. Brain Lang. 112, 3–11 (2010).

    CAS  PubMed  Google Scholar 

  32. Fadiga, L. & Craighero, L. Hand actions and speech representation in Broca’s area. Cortex 42, 486–490 (2006).

    PubMed  Google Scholar 

  33. Knudsen, B., Creemers, A. & Meyer, A. S. Forgotten little words: how backchannels and particles may facilitate speech planning in conversation? Front. Psychol. 11, 593671 (2020).

    PubMed  PubMed Central  Google Scholar 

  34. Long, M. A. et al. Functional segregation of cortical regions underlying speech timing and articulation. Neuron 89, 1187–1193 (2016).

    CAS  PubMed  PubMed Central  Google Scholar 

  35. Tate, M. C., Herbet, G., Moritz-Gasser, S., Tate, J. E. & Duffau, H. Probabilistic map of critical functional regions of the human cerebral cortex: Broca’s area revisited. Brain 137, 2773–2782 (2014).

    PubMed  Google Scholar 

  36. Long, M. A. & Fee, M. S. Using temperature to analyse temporal dynamics in the songbird motor pathway. Nature 456, 189–194 (2008).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  37. Okobi, D. E., Jr, Banerjee, A., Matheson, A. M. M., Phelps, S. M. & Long, M. A. Motor cortical control of vocal interaction in neotropical singing mice. Science 363, 983–988 (2019).

    ADS  CAS  PubMed  Google Scholar 

  38. Tremblay, P. & Dick, A. S. Broca and Wernicke are dead, or moving past the classic model of language neurobiology. Brain Lang. 162, 60–71 (2016).

    PubMed  Google Scholar 

  39. Hosman, T. et al. Auditory cues reveal intended movement information in middle frontal gyrus neuronal ensemble activity of a person with tetraplegia. Sci Rep. 11, 98 (2021).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  40. Catani, M. et al. Short frontal lobe connections of the human brain. Cortex 48, 273–291 (2012).

    PubMed  Google Scholar 

  41. Glasser, M. F. et al. A multi-modal parcellation of human cerebral cortex. Nature 536, 171–178 (2016).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  42. Mathis, A. et al. DeepLabCut: markerless pose estimation of user-defined body parts with deep learning. Nat. Neurosci. 21, 1281–1289 (2018).

    CAS  PubMed  Google Scholar 

  43. Deger, K. & Ziegler, W. Speech motor programming in apraxia of speech. J. Phon. 30, 321–335 (2002).

    Google Scholar 

  44. Jackson, E. S. et al. A fNIRS investigation of speech planning and execution in adults who stutter. Neuroscience 406, 73–85 (2019).

    CAS  PubMed  Google Scholar 

  45. Bogels, S., Casillas, M. & Levinson, S. C. Planning versus comprehension in turn-taking: fast responders show reduced anticipatory processing of the question. Neuropsychologia 109, 295–310 (2018).

    PubMed  Google Scholar 

  46. Dale, A. M., Fischl, B. & Sereno, M. I. Cortical surface-based analysis. I. Segmentation and surface reconstruction. Neuroimage 9, 179–194 (1999).

    CAS  PubMed  Google Scholar 

  47. Fischl, B. et al. Automatically parcellating the human cerebral cortex. Cereb. Cortex 14, 11–22 (2004).

    PubMed  Google Scholar 

  48. Klein, A. & Tourville, J. 101 labeled brain images and a consistent human cortical labeling protocol. Front. Neurosci. 6, 171 (2012).

    PubMed  PubMed Central  Google Scholar 

  49. Desikan, R. S. et al. An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest. Neuroimage 31, 968–980 (2006).

    PubMed  Google Scholar 

  50. Avants, B. B. et al. A reproducible evaluation of ANTs similarity metric performance in brain image registration. Neuroimage 54, 2033–2044 (2011).

    PubMed  Google Scholar 

  51. Tyszka, J. M. & Pauli, W. M. In vivo delineation of subdivisions of the human amygdaloid complex in a high-resolution group template. Hum. Brain Mapp. 37, 3979–3998 (2016).

    PubMed  PubMed Central  Google Scholar 

  52. Kovach, C. K. & Gander, P. E. The demodulated band transform. J. Neurosci. Methods 261, 135–154 (2016).

    PubMed  Google Scholar 

  53. Liu, Y., Coon, W. G., Pesters, A., de, B. P. & Schalk, G. The effects of spatial filtering and artifacts on electrocorticographic signals. J. Neural Eng. 12, 056008 (2015).

    ADS  CAS  PubMed  PubMed Central  Google Scholar 

  54. Friston, K. J. et al. Statistical parametric maps in functional imaging: a general linear approach. Hum. Brain Mapp. 2, 189–210 (1995).

    Google Scholar 

  55. Qian, T., Wu, W., Zhou, W., Gao, S. & Hong, B. in Annual International Conference of the IEEE Engineering in Medicine and Biology Society 2347–2350.

  56. Tilsen, S. et al. Anticipatory posturing of the vocal tract reveals dissociation of speech movement plans from linguistic units. PLoS ONE 11, e0146813 (2016).

    PubMed  PubMed Central  Google Scholar 

Download references


We thank A. Flinker, E. Jackson, J. Krivokapić, D. Schneider, N. Tritsch and members of the Long laboratory for comments on earlier versions of this manuscript; A. Ramirez-Cardenas, H. Chen, K. Ibayashi, H. Kawasaki, K. Nourski, H. Oya, A. Rhone and B. Snoad for help with data collection; and F. Guenther and N. Majaj for helpful conversations. This research was supported by R01 DC019354 (M.A.L.), R01 DC015260 (J.D.W.G.) and Simons Collaboration on the Global Brain (M.A.L.).

Author information

Authors and Affiliations



G.A.C. and M.A.L. conceived the study and designed the experiments; G.A.C., C.K.K., J.D.W.G. and M.A.L. conducted the research; G.A.C., C.K.K. and M.A.L. performed data analyses; G.A.C., C.K.K. and M.A.L. created the figures; G.A.C. and M.A.L. wrote the initial draft of the manuscript; G.A.C., C.K.K., M.A.H., J.D.W.G. and M.A.L. edited and reviewed the final manuscript. J.D.W.G. and M.A.L. acquired funding; J.D.W.G., M.A.H. and M.A.L. supervised the project.

Corresponding author

Correspondence to Michael A. Long.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review information

Nature thanks Gregory Cogan, Uri Hasson and Frederic Theunissen for their contribution to the peer review of this work. Peer reviewer reports are available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data figures and tables

Extended Data Fig. 1 Behaviour during the CI task.

a, Description of subprocesses assumed to occur during the perception, planning, and production windows of the CI task. b, Histograms of reaction times (RT) in early and late CI trials for all participants. c, Median RT values for early and late CI trials for all participants. d, e, Histograms depicting the distribution of average peak-to-trough response amplitudes for all electrodes displaying planning-related responses when aligned to CI onset in early and late trials (d) and different CI question types (e); median values for each distribution are indicated. Observed data (in black) are compared with a null distribution (in grey) consisting of randomly chosen timepoints (Methods). f, Schematics displaying GLM regressor structure for an early (top) and a late (bottom) variant of an example CI task question.

Extended Data Fig. 2 GLM temporal jittering analysis.

a, Full model R values for GLM fits of jittered high gamma activity from participant 436; each line represents data from an individual electrode. b, Example distribution of pooled D values with the fit of two Gaussians overlaid (black). The Gaussian distributions corresponding to well fit (blue) and poorly fit electrodes (red) as well as the 95th percentile of the D distribution for poorly fit electrodes (dashed line) are indicated. D values above the 95th percentile of the pooled distribution were deemed outliers (white bars) and not fitted. c, Table summarizing the number of electrodes rejected by the jittering analysis in each participant. d, Table reporting the anatomical locations of electrodes rejected by the jittering analysis and electrodes displaying significant activity in the CI task. e, Scatterplot depicting the proportion of rejected electrodes within a region as a function of the proportion of responsive electrodes in a region.

Extended Data Fig. 3 Analysis of neural activity in the CI task.

a, Scatterplot depicting the distribution of all simulated task-responsive electrodes from the continuum model in three-dimensional GLM weight space; cluster membership indicated by greyscale colour. b, c, Distribution of simulated electrodes from the continuum model displaying responses in one window (i.e., unmixed) of the CI task (b) or multiple windows (c); response class indicated by colour in b and c and unmixed electrodes denoted by small black points in c. In b, simulated unmixed electrodes located outside the cluster primarily containing electrodes of the same type (i.e., ‘misclustered’) are indicated with an ‘X’. d, e, Histograms depicting the distribution of the proportion of misclustered electrodes responsive during a single task window (i.e., unmixed electrodes) (d), and the proportion of electrodes displaying more than one significant positive weight (i.e., mixed electrodes) (e) across 100,000 iterations of the continuum model simulation. The median of each distribution as well as the values observed in the actual data (dashed line) are indicated. Gold arrows indicate the bin of each distribution containing the measurements corresponding to the example iteration depicted in panels p, r, and t of Fig. 1. f, Table reporting the number of electrodes displaying perception-related responses using either the full model or the reduced GLM lacking a planning regressor. g, h, Scatterplots depicting perception (g) and planning (h) GLM weights in the full model and reduced models lacking a planning regressor or perception regressor, respectively. Significant positive weights are denoted with filled points and nonsignificant or significant negative weights are denoted with unfilled points; the x-coordinates of each point are randomly jittered by 25% to better visualize filled versus unfilled status. No planning electrodes displayed significant perception responses in the reduced GLM lacking a planning regressor, and no perception electrodes displayed significant planning responses in the reduced GLM lacking a perception regressor.

Extended Data Fig. 4 Additional analyses of task-related activity changes.

a, Table reporting the number of perception, planning, and production-related electrodes displaying significant positive and negative weights for each GLM regressor. b, Histogram depicting mean high gamma amplitude in the first 500 ms of CI questions for all unmixed perception, planning, and production electrodes. c, d, Canonical cortical surfaces displaying electrodes with significant positive (coloured) or negative (black) GLM weights in the perception (c), production (d), and planning (e) windows of the CI task across all participants. Electrode diameter is scaled to the absolute magnitude of the GLM weight, and electrodes not displaying a significant weight for a given regressor are indicated with small white circles.

Extended Data Fig. 5 Anatomical analysis of responses.

a, Cortical reconstructions for all participants displaying the location of all electrodes; the size of each electrode depicts the actual size of its recording area on the cortical surface. GLM classification is indicated by electrode colour. b, Canonical cortical surfaces showing electrode locations from all participants as standard-sized white circles. c, Number of electrodes sampling each area of the canonical cortical surface (1 cm diameter spatial smoothing) after pooling electrodes from all participants. d, Proportion of electrodes displaying significant production-related responses in the CI task (1-cm-diameter spatial smoothing). e, Canonical cortical surfaces displaying electrodes with significant responses related to speech perception, production, and planning in patients with tumour (top) and patients with epilepsy (bottom) separately; electrode diameter scaled to GLM regressor weight. Electrodes not displaying a significant response for a process are depicted as small white circles.

Extended Data Fig. 6 Additional conversation-related analyses.

a, Table reporting additional turn-taking behavioural measures for each participant. b, Histograms of gap durations (time between experimenter turn offset and participant turn onset) during unconstrained conversation for each participant; bins are centred on 100 ms increments with a width of 100 ms. c, Scree plots for the PCA analysis of high gamma signals in the task (left) and conversation (right) periods of the recordings; data from each participant are represented by thin lines and the average across participants is denoted with a thick black line. The 95% confidence interval of the linear decay phase across participants (Methods) is also indicated. d, The observed number of electrodes whose cluster membership was not stable (i.e., switched clusters) between the task and conversation with a histogram depicting the distribution of electrode cluster switches expected by chance. e, The observed percentage of electrodes in perception, planning, and production clusters (in conversation-derived PC coefficient space) displaying significant perception, planning, and production responses (per the GLM), respectively, with histograms depicting the percentages expected by chance for each cluster type. f, Canonical cortical surfaces displaying the locations of all electrodes in perception, planning, and production clusters across participants (n = 6) in the task (left) and conversation (right). g, Table reporting summary statistics for PC activity (i.e., time-varying PC score) during unconstrained conversation for each participant.

Extended Data Fig. 7 PCA results for individual participants.

af, For 6 participants possessing sufficient numbers of electrodes belonging to multiple GLM classes (Methods): scatterplots depicting electrode distributions in PC coefficient space in the task and conversation periods (top row). Bar graphs depicting the PC coefficients for all electrodes in perception, planning, or production clusters from the PCA performed on task data and conversation data (bottom rows). Participant number given at top of each panel. g, h, For 2 participants possessing mainly planning electrodes (Methods, Extended Data Table 1): bar graphs depicting the PC coefficients for all planning-related electrodes from the PCA performed on task data and conversation data. In the bar graphs, the functional categorization of PCs is indicated by filled bars coloured either green (perception), blue (planning), or red (production). Any clusters rejected due to a high proportion (50%) of mixed electrodes are indicated with grey filled bars.

Extended Data Table 1 Participant information
Extended Data Table 2 DKT parcellation of all CI task-responsive electrodes

Supplementary information

Reporting Summary

Peer Review File

Supplementary Data 1

List of all task stimuli.

Supplementary Data 2

All electrode locations and GLM classifications.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Castellucci, G.A., Kovach, C.K., Howard, M.A. et al. A speech planning network for interactive language use. Nature 602, 117–122 (2022).

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI:


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing