If we derive the momentum in Special Relativity we get :
$$ \dfrac{\mathrm{d}\boldsymbol{p}}{\mathrm{d}t} = \dfrac{\mathrm{d}m \gamma \boldsymbol{v}}{\mathrm{d}t} = m\dot{\gamma}\boldsymbol{v}+m\gamma\dot{\boldsymbol{v}} = \boldsymbol{f} $$
It's the relation between the 3-force and 3-acceleration in SR we can find on Wikipedia on other textbook. We can see that if it exist a "work" done by a force and the motion is not in the direction of that force (velocity is not longitudinal to the force) then, the "object" feel an acceleration in the transverse direction.
For example the object is a bullet with :
$$ \boldsymbol{v} = (v_x,v_y) \quad \boldsymbol{f}=(f_x,0) $$
$\boldsymbol{f}$ can be due to an uniform electric field or gravity. The work is not 0, and the velocity in $y$ direction is not 0 too. Even if there is no force in the $y$ direction we have :
$$ \dfrac{\dot{\gamma}v_y}{\gamma}=\color{red}{-}\dot{v_y} $$
And so, the acceleration is not parallel to the force like in Newtonian/Classical physics, where $\gamma$ does not exist. Here, the change of energy due to the force along $x$ imply an acceleration in $y$ direction.
I read several text where authors spoke about longitudinal and transverse mass, but it seems it's an old concept which is don't use anymore.
This phenomenon can be expressed with simple word ? I don't find an explanation in my own brain apart "The derivation chain rule"...
\dot{v_y}
and\dot{v}_y
typeset differently ($\dot{v_y}$ and $\dot{v}_y$ respectively). People (like me) who are hung up on typesetting generally prefer the latter. – dmckee --- ex-moderator kitten Jun 03 '19 at 19:10