It has been known that speech articulation declines with age¹. Impairment in speech articulation could lead to shame and embarrassment, and may result in social isolation as speech is one of the primary methods to communicate with others. This functional magnetic imaging (fMRI) study addresses the question whether the elderly’s speech articulation is trainable or not. Speech articulation refers to the process to translate conscious speech plans into motor plans. The dual-route cascaded model² anticipates that visual word recognition and reading aloud can be realized via either lexical semantic route or lexical non-semantic route. A word stored in the mental lexicon (i.e., real word) activates its semantic network, which can facilitate conversion from orthography to phonology. To read aloud a word not stored in the mental lexicon (i.e., pseudo word), the pronunciation is assembled according to the rule of grapheme-to-phoneme conversion. Following the speech plans, the core motor plans of the sequence of phonological units are recalled from the sensorimotor memory, and then the movements necessary for producing the sounds in the planned units are organized³. We attempted to explore the elderly’s neural changes between before and after four weeks of articulatory training, by comparing real and pseudo words (i.e., the difference in speech plan), and hard and easy consonants to articulate (i.e., the difference in motor plan).

Participants: Twenty healthy right-handed elderly native Japanese speakers (12 males, age 69.7±4.2) participated in the experiment (3T MR Siemens Trio/Tim) in accordance with the local ethics regulations.

Procedure: The training study design comprised fMRI test 1 (Day 1), training sessions 1-21 (Day 2-28), and fMRI test 2 (Day 29). During the fMRI test on Day 1, the participants read aloud Japanese phrases which consisted of real and pseudo words with hard and easy consonants, while being scanned in the MR scanner. After scanning, they were tested for accuracy and reaction times of reading whole sentences. In each of the 21 training sessions, they were required to read aloud the whole sentences 8 times each as quickly, as accurately, and as loudly as possible. On Day 29, they took the second fMRI and behavioral test.

fMRI data acquisition: Functional images employed a T2-weighted GRE-EPI sequences with the following parameters: TE = 30 ms, TR = 3000 ms, flip angle = 90°, matrix 64 × 64, field of view = 192 mm, 39 axial slices, slice thickness = 3mm, and distance factor = 25%. For each EPI run, 90 volumes were acquired. A high-resolution, three-dimensional MPRAGE T1-weighted image was obtained for anatomical detail.

fMRI data analysis: First-level contrasts for each condition of the task were entered into second-level, random effects analyses on the basis of the general linear model using SPM 12 (Wellcome Department of Cognitive Neurology, London, UK).

The elderly participants showed reduced activity in the right pars triangularis of the inferior frontal gyrus after four weeks articulatory training of real words with both of easy and hard consonants (Table 1 and Figure 1). After psuedo word training, decreased activity was shown in the right precentral gyrus, the left middle occipital gyrus, the right supramarginal gyrus, the left postcentral gyrus, the right pars opercularis of the inferior frontal gyrus, the left insula, and the right SMA in the both conditions of consonants (Table 2 and Figure 2). Especially for easy consonants of pseudo words, the training reduced activity in the bilateral precuneus, the left fusiform gyrus, the right inferior parietal lobule, the left hippocampus, and the right cerebellar lobule VI.For the elderly’s real word articulation training which allows conversion from orthography to phonology, we found the reduced activity only in the right pars triangularis of the Broca’s area, but not in any motor-related areas. However, their training of pseudo words, which require grapheme-to-phoneme conversion, exhibited significant changes in many different regions including language, motor, visual, and cerebellar areas. Although speech articulation has been considered to be poorer in the elderly, the current study revealed that training can enhance their neural plasticity to improve articulatory execution of newly-introduced speech sounds, especially with easy consonants which do not require complex articulatory movements.The elderly speakers have plastic-adaptive articulation learning ability of newly-introduced pseudo words with easy consonants. This finding supports effectiveness of this kind of training activities for the elderly speakers to facilitate their fruitful language communication.