Kinematics of lateral tongue-pushing movement in coordination with masticatory jaw movement: An anteroposterior projection videofluorographic study

Objective: During the mastication of solid food, the tongue pushes the bolus laterally to place it onto occlusal


Introduction
The tongue is an essential organ in human feeding; impairment in tongue function greatly affects masticatory performance. Cerebrovascular and neuromuscular diseases often cause impaired movement and/ or weakness of the tongue [1][2][3]. In affected patients, because the tongue cannot hold the bolus of food on its dorsal surface, it cannot sufficiently propel the ingesta from the oral cavity to the pharynx. Similarly, in patients with surgical removal of tongue cancer, a deficit in the tongue body can lead to severely impaired feeding that affects mastication, oral transport, and swallowing [4,5]. Understanding the kinematic mechanisms of tongue behavior in feeding would therefore be beneficial in developing rehabilitation programs for mastication and swallowing in such patients.
The tongue has a unique musculature: its extrinsic muscles connect to the mandible, craniofacial bones, and hyoid bone, while its intrinsic muscles comprise sets of intralingual muscle fiber bundles, which are not connected to bony tissue. The coordinated action of intrinsic and extrinsic muscles enables complicated three-dimensional (3D) movement of the tongue during feeding, which has been examined using various techniques, such as videofluorography (VFG), X-ray microbeam, magnetic resonance imaging (MRI), and ultrasonography [6][7][8][9][10]. However, due to technological limitations, the coordinated movements of the tongue and jaw during feeding have only been evaluated two-dimensionally, either in the sagittal or frontal projection [11][12][13][14][15].
The temporo-spatial coordination of the tongue and jaw has been studied quantitatively in the sagittal view using VFG and radiopaque markers attached to the tongue surface and teeth [11,12,14]. During chewing, the tongue moves anteriorly and inferiorly in the jaw opening phase and posteriorly and superiorly in the jaw closing phase [11]. This motion is produced by the kinematic linkage of the jaw and hyoid bone with the actions of the tongue muscles. The major shortcoming of these reports is that the tongue moves laterally and rotates during feeding, and sagittal (two-dimensional) VFG observation alone is unable to capture the mediolateral movements of these structures.
Lateral movement and rotation of the tongue are important for masticatory performance but these aspects have not been well studied due to limitations in visualization techniques. Abd el Malek recruited volunteers who had no incisors and observed their tongue movement directly during mastication, using small retractor forceps to hold the lips apart [16]. He reported that the tongue twisted over to one side to push the ingesta onto the surface of the lower grinding teeth. Subsequently, cinefluorography and VFG became available in clinical and research settings and was used to visualize lateral tongue movement [13,17,18]. Mioche et al. reported that the tongue pushed the food bolus laterally, rotated, and then placed the bolus onto the surfaces of mandibular post-canine teeth during mastication, and this phenomenon was named "tongue-pushing" (TP) [13]. These previous studies have qualitatively described lateral tongue movement and rotation during mastication, but the kinematics of this movement remains poorly understood due to the lack of empirical analyses.
Our group developed a technique to visualize and quantitatively assess the movement of the tongue, jaw, and hyoid during feeding sequences, using VFG and radiopaque metal markers [11,12,14,15]. In the present study, we applied VFG in the anteroposterior (AP) projection, to quantify the kinematics of lateral tongue and jaw movements during mastication.

Participants
All protocols in this study were approved by the applicable institutional review board. Fourteen healthy young adult volunteers (8 men and 6 women; age [mean ± SD]: 23.0 ± 4.0 yrs) participated in the trial, after providing oral and written informed consent. The subjects had no history of swallowing difficulties, speech or voice disorders, dental problems, gastroesophageal reflux disease, otolaryngeal pathology, pulmonary or neurological disease, or structural disorders. They exhibited no signs of abnormal swallowing in a diagnostic 10 mL liquid barium swallow in lateral and AP projection VFG. All subjects had Class I dental occlusion.

Data collection
Small lead discs (4 mm in diameter and 0.4 mm thick) were used as radiopaque markers to quantify tongue surface and jaw movement. Three radiopaque discs were glued to the dried dorsal surface of the tongue with topical skin adhesive (Dermabond, Ethicon, LLC, San Lorenzo, Puerto Rico). One disc was attached approximately 10 mm posterior to the tongue tip in the midline (anterior tongue marker; ATM). The two remaining discs were glued to the tongue surface near the lateral edge, on each side, approximately 20 mm posterior to the ATM and near the lower first molars (right and left tongue markers) (Figs. 1,2). Markers were also glued to the buccal surface of the right and left upper first molars and lower canines with dental adhesive cement. Previous studies have indicated that the effect of tongue markers on feeding behavior is negligible [12].
VFG in the anteroposterior (AP) projection was recorded on a digital video recorder at 30 frames/sec while subjects ate two test foods: 6 g each of meat (chiken breast; fibrous texture) and shortbread cookie (Pure Butter Shortbread Ⓡ , Walkers Shortbread Inc.; crunchy texture). Meat was served with the same size spoon each time. A time stamp was simultaneously recorded and overlaid onto each video frame. The recording of meat-chewing was performed before that of cookiechewing because tongue surface markers occasionally detach during feeding, especially when eating hard food [11]. Markers that detached were replaced and the recording was repeated, if the VFG recording time had not exceeded 5 min of total radiation exposure. Six subjects completed two sets of recordings, whereas the other eight subjects were unable to complete both sets (meat-and cookie-chewing) because the 5 min radiation exposure time had been exceeded; four subjects missed one trial for cookie-chewing, two subjects missed one trial each for cookie-and meat-chewing, and two subjects could complete only one trial for meat-chewing. Each subject performed two trials for each food type. Consequently, a total of 42 trials were acquired for analysis. We included all trials except for the marker detached trials because the  exclusion of individuals who ate slowly could have created an experimental bias. Additional trials for detached marker subjects were included in the data.

Data reduction
VFG recordings on digital video tapes were converted to digital video files (AVI format) with no image compression on a desktop computer. Jaw and tongue movements were first evaluated visually using the slow motion and stop-frame functions of video editing software (VirtualDub ver. 1.9.11, GNU General Public License). For each recording, the first masticatory sequence from food ingestion to the first swallow was used for analysis. Jaw motion was divided into cycles, starting at minimum gape (the end of visible upward movement of the jaw) and ending at the next. Because this study aimed to determine tongue movement during chewing, we excluded the first jaw cycle, which started just after ingestion, and the five cycles before the first swallow, because the movement of the tongue during food transport to the post-canine region in the first jaw cycle and during oro-pharyngeal food transport, which often occurs in later jaw cycles, were found to differ from tongue behavior during chewing [15,19]. Cycles with no tooth-food-tooth contact, which frequently appeared at the end of the masticatory sequence, were also omitted since food was only being "manipulated" during these cycles [19].

TP cycles
In their report, Mioche et al. defined TP as cycles in which the tongue pushes food onto lower occlusal surfaces by its lateral movement and rotation [13]. In the present study, a TP cycle was similarly established as a jaw motion cycle in which the tongue markers shifted and rotated toward the lower teeth on the working side. Non-TP cycles were defined as jaw motion cycles in which the tongue did not show the above-described motion. Two expert observers divided all masticatory jaw motion cycles into TP and Non-TP cycles by visual observation using video images. TP cycles were then checked by consensus of five expert observers. The frequency of TP cycles was defined as: where F TP = frequency of TP cycles; N TP = number of TP cycles; and N Non-TP = number of Non-TP cycles.

Tongue movement and rotation
For each VFG image, the raw X-Y coordinates of the tongue and tooth markers were manually acquired frame-by-frame, using an imageprocessing program (Image-J, NIH, Bethesda). Acquired X-Y positions for each marker were then calibrated and expressed as Cartesian coordinates relative to the occlusal plane of the upper teeth in spreadsheet data files (Fig. 1). The upper occlusal plane was defined by a line passing through both upper first molar markers, which served as the X-axis. The Y-axis was defined as a line which passed through the upper right molar and was perpendicular to the X-axis. If the location of a tongue marker was not visible in a video frame due to rapid tongue motion, the video frame was excluded from the analysis. A total of 1632 frames (4.4%) were excluded for this reason.
The working and balancing sides of the jaw were identified visually for each jaw motion cycle. The jaw side which compressed food particles was defined as the working side of the jaw, while the other side was termed the balancing side. Markers were classified for each cycle as canine markers on the working (WC) and balancing (BC) sides and tongue markers on the working (WTM) and balancing (BTM) sides.
To determine the displacement of the tongue and jaw in each jaw motion cycle, the range of motion of each marker was calculated by subtracting the minimum values from the maximum values of the Cartesian coordinates. To evaluate tongue rotation in each jaw motion cycle, tongue surface angle (TSA) was calculated as the angle between a line connecting the WC and BC and that connecting the WTM and BTM (Fig. 3). The maximum TSA value (maxTSA) was measured for each jaw cycle.

Movement direction
Movements are described with the following conventions: Horizontal displacement refers to the component of motion that is parallel to the mediolateral axis; horizontal displacements are generally directed towards the working side. Vertical displacement is the component of motion parallel to the superoinferior axis; vertical displacements are directed downward. The AP component of motion is poorly visualized with VFG in the frontal projection. The tongue rotates along its longitudinal (AP) axis.
The directions of horizontal ATM and right canine marker movement (from the onset of jaw opening to maximum gape) were identified for each jaw cycle on the working and balancing sides. The concordance rates of movement directions of the jaw and tongue markers were calculated for TP and Non-TP cycles.

Data analysis
Multivariate analysis of variance (MANOVA) was performed to test the hypothesis that horizontal and vertical displacements of the markers differed by marker (WC, BC, ATM, WTM, or BTM), jaw motion cycle (TP or Non-TP), and food type (cookie or meat). Univariate ANOVA was employed to test differences in maxTSA between jaw motion cycles and food type. We adopted Bonferroni's test for all post-hoc pairwise comparisons.
Data sets for TP frequency and concordance rates were analyzed for  statistical significance by binomial testing. The critical value for rejecting the null hypothesis was set as P < 0.05. SPSS software (ver. 21.0; IBM, Armonk, NY, USA) was used for statistical analyses.

Results
Overall, TP occurred in 58.8% (416/707) of jaw cycles. TP was significantly more frequent with meat-chewing than with cookiechewing (64.3% [283/440] vs. 49.8% [133/267]; P < 0.01). Typical TP movements from the frontal view are shown as drawings based on VFG image traces in Fig. 4 and as line graphs of marker motion and gradual changes in TSA in Fig. 5.
When the subjects chewed the test foods, large particles were compressed between the upper and lower teeth, often escaping to the lingual (medial) side. As the jaw moved downward during opening, the tongue blade moved laterally and rotated toward the working side, depositing the food on the occlusal surfaces of the lower teeth (Figs. 4B-F, 6). At its farthest lateral position, the tongue surface also rotated toward the lower occlusal surfaces (Fig. 4F). Just after the start of jaw closing, the tongue maintained its lateral excursion to retain the food between the upper and lower teeth until the time of tooth-food-tooth contact. After the initiation of this contact, the tongue returned to its initial position.

Horizontal displacement
In the horizontal (mediolateral) dimension, the average displacement of all jaw and tongue markers was significantly larger in TP cycles than in Non-TP cycles (P < 0.05, Fig. 7A).
Horizontal displacement was larger for tongue markers (ATM, BTM, and WTM) than for jaw markers (BC and WC) during TP and Non-TP cycles (P < 0.05). During TP cycles, the horizontal displacement was largest for the ATM (P < 0.05), followed by the BTM (P < 0.05), and was significantly larger with cookie-eating than with meat-eating (P < 0.05, Fig. 8A).
However, food type had no significant effect on the horizontal displacement of the WTM or jaw markers. The tongue and jaw markers typically moved in the same direction (i.e., toward the working side) during both TP and Non-TP cycles, but the concordance rate was significantly higher for TP cycles (concordance rates: 88.1% vs 70.8%; P < 0.05).

Vertical displacement
In the vertical (superoinferior) dimension, the average displacement was downward for all jaw and tongue markers during jaw opening, and was significantly larger in TP cycles than in Non-TP cycles (P < 0.05), (except for the BTM, Fig. 7B). In contrast to horizontal displacement, the downard displacement was significantly larger for jaw markers than for tongue markers in both TP and Non-TP cycles. Downward displacement was significantly larger for the WTM and ATM than for the BTM (P < 0.05). In Non-TP cycles, downward displacement was similarly largest for the ATM and lowest for the BTM (P < 0.05). Downward displacement of the tooth markers in TP cycles was significantly larger with meatchewing than with cookie-chewing (P < 0.05, Fig. 8B). In contrast, the displacement was significantly larger with cookie-chewing than with meat-chewing for the WTM (P < 0.05), and there was no difference in BTM between the food types.

Discussion
In the present study, we demonstrated the kinematic mechanisms of TP motion during the mastication of solid food, using VFG in the AP projection and radiopaque metal markers. Previous studies reported that the tongue pushes food onto the occlusal surface while the jaw opens [13,16]. We found that TP comprised inferior and lateral movements of the tongue on the working side and medial movement of the tongue on  D). Until the start of tooth-food-tooth contact, tongue markers move laterally toward the working side and the main part of the bolus is positioned on the occlusal surface (E). As tooth-food-tooth contact is initiated, the jaw is moved to the working side in coordination with the tongue movement. The tongue surface is displaced toward the working side (F). A', C', and F' illustrate the overall movement of the tongue surface during TP and these images correspond to A, C, and F, respectively. the balancing side that produced lateral tongue shift and rotation during jaw opening. In contrast, the movements of tongue markers during Non-TP cycles were less pronounced and were bilaterally symmetrical (as reported for oral-pharyngeal food transport [15]). We also witnessed that the vertical displacement of the tongue was significantly less than that of the jaw in TP cycles, indicating that the tongue moved upward only slightly while the jaw was opened. This upward motion may support the placement of food onto the occlusal surface and lateral tongue shift and rotation.

TP mechanisms
Unlike BTM, the ranges of vertical displacement of the WTM and ATM were greater in TP cycles than in Non-TP cycles. TSA was also greater in TP than in Non-TP cycles. TSA was mainly determined by the difference in the vertical positions (Y-coordinates) of the WTM and BTM. These findings demonstrate that tongue rotation toward the lower occlusal surfaces of the working side is produced by the downward movement of the tongue surface on the working side with the balancing side remaining relatively stable.
We observed the median value of maxTSA for TP cycles to be 33.2 • . However, this angle does not precisely represent the angle of the tongue surface, since the tongue is curved. Given that the WTM and BTM are located near the lateral edges of the tongue surface, the TSA can be considered an "average" angle of the tongue surface relative to the upper occlusal plane. Given the convex shape of the tongue's surface at MaxTSA, we believe that the working side of the tongue was nearly perpendicular to the occlusal surfaces of the adjacent lower teeth.
During TP cycles, the range of horizontal displacement toward the working side was highest for the ATM and lowest for the WTM. If TP was produced by simple rotation of the tongue body around its longitudinal axis, we would expect the BTM to have much larger horizontal displacement than the ATM, but the reverse was the case. These findings suggest that TP motion is not simple tongue rotation, but a vigorous lateral shift of the anterior tongue blade simultaneous to tongue rotation. We consider that this motion increases the pressure of the tongue surface against the food and the lingual surfaces of the adjacent lower teeth. This could allow TP to reposition the food more effectively.

Effect of food consistency on TP
The overall frequency of TP cycles was 58.8% in our cohort. We could not precisely identify the location of the bolus because it contained no radiopaque contrast. However, a previous study by Mioche et al. [13] suggested that unilateral chewing occurred more frequently in TP cycles, whereas bilateral chewing and oro-pharyngeal food transport were predominant in Non-TP cycles. The frequency of TP cycles was significantly higher with meat-chewing (63.8%) than with cookie-chewing (49.8%). This may reflect the textures of the test foods: meat is fibrous, thus, significant shearing force is required to crush it. This effect is readily accomplished with unilateral chewing. Conversely, cookie is more friable, thus, food particles can be crushed by the first few chewing strokes, with a higher likelihood of bilateral chewing. In bilateral chewing cycles, the food is on both sides of the mouth and motions are more symmetrical. This could explain the preponderance of TP cycles with meat-chewing than with cookie-chewing. On the other hand, the 14% change in TP cycle frequency between cookie and meat might be small compared to the difference in food texture. In the future study, we would like to examine the equivalence of the TP cycle frequency in different food.
The range of vertical displacement of the jaw was significantly larger with meat-chewing than with cookie-chewing. The hardness or physical properties of food strongly influence the vertical amplitude of jaw movement, as shown previously [20,21]. For cookie-chewing, we observed that the vertical amplitude of chewing cycles gradually decreased as the food was crushed by biting force. For meat, however, the fibrous texture prevented it from being rapidly crushed, resulting in a larger average amplitude of vertical jaw motion during chewing.

Link between the tongue and jaw during TP
Previous studies have reported complex patterns of temporospatial relationship between the tongue, jaw, and hyoid bone in the sagittal plane during eating and speech [11,12,14]. We report kinematic linkage of the tongue and jaw during TP cycles, with 88% concordance in the mediolateral dimension. This close temporospatial coordination is consistent with the presence of a central pattern generator that regulates motions of the jaws and the related soft tissues during mastication and is sensitive to food consistency. Further investigation is needed to elucidate this neural mechanism, that is, the coordination of the jaw and tongue musculature during TP.

Clinical significance
The tongue plays a key role in the feeding process, which includes mastication, oral-pharyngeal food transport, bolus aggregation, and swallowing [22]. Tongue movement impairment caused by neuromuscular or cerebrovascular disorders affect the placement of food on the occlusal surfaces during chewing and/or food bolus transport to the pharynx. Thus, food residue may remain in the lingual sulcus (between the teeth and tongue). The present study revealed several kinematic mechanisms for TP motion during chewing. Exercises to increase the strength of tongue protrusion and elevation are common in rehabilitation for swallowing disorders [23]. Our findings suggest that designing exercises to improve the strength of the lateral motion and rotation of the tongue body may be useful for individuals with impaired tongue function for eating and swallowing.

Limitations
This study had several limitations. First, food samples were not visualized with barium to avoid occluding the images of the radiopaque tongue markers. Thus, we could not precisely identify the movements of the food in the recordings. Mioche et al. reported that food is transferred from one side of the mouth to the other during chewing or is retained on one side [13]. We could not verify these hypotheses because the test foods did not contain a radiopaque contrast agent.
Another limitation is the limited visualization of the tongue surface. Although the three tongue markers indicated the trajectory of tongue surface movement at several loci, they could not fully represent the movement of the contoured tongue surface. Kinematic analyses using visible markers is quite useful for relatively rigid structures, such as the jaw or limbs [24,25]. However, the tongue is a muscular hydrostat; the distance between surface markers changes during food processing.
Another limitation is the young age of the study participants. It seems important to examine the robustness of our findings in a more heterogeneous sample. However, the novel design of the present study using a unique montage of radiopaque tongue surface markers yielded new valuable insights into the kinematics of the tongue that are associated with masticatory jaw movement.

Conclusions
This study suggests that: (1) TP comprises inferior and lateral movements of the tongue on the working side and the anterior tongue blade, with medial movement of the tongue on the balancing side. These movements produce lateral shift and rotation of the tongue surface during jaw opening; (2) TP frequency and displacement vary with food Box and whisker plots of the displacement of the lower canine and tongue markers in tongue-pushing (TP) and Non-TP cycles in the horizontal (A) and vertical (B) dimension. * indicates P < 0.05. ¶ indicates P < 0.05 between TP and Non-TP cycles. A. Downward displacement of all markers was significantly larger in TP cycles than in Non-TP cycles. Among TP cycles, horizontal displacement was largest for the anterior tongue marker (ATM), followed by the balancing tongue marker (BTM). Among Non-TP cycles, horizontal displacement was higher for the working side tongue marker (WTM) than for the BTM. B.
In TP cycles, downward displacement was larger for WTM and ATM than for BTM.
consistency; and (3) TP depends on tight kinematic coordination of the jaw and tongue. These findings have potentially important clinical implications for the understanding and treatment of impairments of tongue motion and control.

Funding
This study was supported by the National Institute on Deafness and Other Communication Disorders, Award R01-DC-02123, JSPS KAKENHI Grant Number 22K02860.

Declarations of Competing Interest
None.  Fig. 8. Comparison of the displacement of the lower canine and tongue markers between cookie-and meat-chewing in tongue-pushing (TP) and Non-TP cycles. Displacements are shown in box and whisker plots in the horizontal (A) and vertical (B) dimension. * indicates P < 0.05. A. In TP cycles, horizontal displacement for the anterior and balancing side tongue markers (ATM and BTM, respectively) was significantly larger with cookie-chewing than with meat-chewing. There were no significant differences between food types for the working side tongue marker (WTM) or lower canine markers. In Non-TP cycles, there was no significant difference in horizontal displacement between food types for any tongue or jaw marker, apart from the ATM. B. In TP cycles, downward displacement for both balancing and working side canine markers (BC and WC, respectively) and the ATM was significantly larger with meat-chewing than with cookie-chewing. Downward displacement for the WTM was significantly larger with cookie-chewing than with meat-chewing in TP cycles. There was no significant difference between foods for the BTM. In Non-TP cycles, there was no significant difference in vertical displacement between food types for any tongue or jaw marker, apart from the ATM.