Abstract:

Speech imagery is an internal perceptual experience of one's own speech or others' speech. It not only plays an important role in the pre-processing mechanism of the brain, but it is also the latest technology in the field of brain-computer interface (BCI) research. Firstly, the theoretical model, activation of brain regions, and neural conduction pathways of speech imagery have many similarities with speech production, but there are still some controversies. Theoretical models of speech production, such as the Directions Into Velocities of Articulators (DIVA) model and the State Feedback Control (SFC) model, show that the process of speech imagery and speech production are highly overlapping in the motor planning of articulatory organs and the prediction of somatosensory and auditory results. The difference is that speech imagery does not activate the last step in speech production, that is, the execution of articulatory movement. There are many similarities between the brain regions activated by speech imagery and speech production, which are mainly reflected in the brain regions of the speech motor planning and auditory center. However, the signal strength of the brain regions activated by speech imagery is weaker. The neural pathways of speech imagery and speech production are also similar. However, current research is only limited to the connections between some brain regions, such as the connection and integration of auditory cortex and motor cortex. Whether or not there are direct or indirect conduction pathways in speech imagery consistent with speech production needs further study. Secondly, for people with speech disorders, the severity of speech imagery ability is not completely positively related to the severity of speech production impairment. For example, some patients with aphasia or stuttering have no limitation in their ability of speech imagery. Due to the similarities between neural mechanisms on speech imagery and speech production, speech imagery therapy is considered a new technology for the rehabilitation of people with speech impairments caused by brain injury. However, there is still a lack of mechanism and application of speech imagery for dysarthria caused by the damage to speech motor planning and auditory cortex and its neural conduction pathways. Thirdly, when imagining meaningful words and sentences, the EEG signals are different from those in speech production. The complexity, length, and meaning of the speech samples will affect the brain activation during speech imagery. Moreover, the identification of brain regions with weak signals, stimulation methods, and brain signal detection techniques will affect the identification of cranial nerve signals. The neural mechanism of speech imagery plays an important role in both the brain preprocessing mechanism and BCI technology. Given the complexity of the speech system, research on the neural mechanism of speech imagery still faces a series of challenges. Further research can focus on speech imagery quality evaluation tools, neural decoding paradigms, brain control circuits, activation pathways, and speech imagery mechanisms in speech disorders. Further exploration of the cranial nerve signals of word and sentence imagery will help provide a basis for effectively improving the recognition rate of BCI and to facilitate the communication for people with speech disorders.

Key words: speech imagery, speech motor planning, brain-computer interface, neural mechanism