Context. The ambiguous origin of the [C ii] 158 µm line in the interstellar medium complicates its use for diagnostics concerning the star-formation rate and physical conditions in photodissociation regions. Aims. We investigate the origin of [C ii] in order to measure the total molecular gas content, the fraction of CO-dark H 2 gas, and how these parameters are impacted by environmental effects such as stellar feedback. Methods. We observed the giant H ii region N 11 in the Large Magellanic Cloud with SOFIA/GREAT. The [C ii] line is resolved in velocity and compared to H i and CO, using a Bayesian approach to decompose the line profiles. A simple model accounting for collisions in the neutral atomic and molecular gas was used in order to derive the H 2 column density traced by C + .Results. The profile of [C ii] most closely resembles that of CO, but the integrated [C ii] line width lies between that of CO and that of H i. Using various methods, we find that [C ii] mostly originates from the neutral gas. We show that [C ii] mostly traces the CO-dark H 2 gas but there is evidence of a weak contribution from neutral atomic gas preferentially in the faintest components (as opposed to components with low [C ii]/CO or low CO column density). Most of the molecular gas is CO-dark. The CO-dark H 2 gas, whose density is typically a few 100s cm −3 and thermal pressure in the range 10 3.5−5 K cm −3 , is not always in pressure equilibrium with the neutral atomic gas. The fraction of CO-dark H 2 gas decreases with increasing CO column density, with a slope that seems to depend on the impinging radiation field from nearby massive stars. Finally we extend previous measurements of the photoelectric-effect heating efficiency, which we find is constant across regions probed with Herschel, with [C ii] and [O i] being the main coolants in faint and diffuse, and bright and compact regions, respectively, and with polycyclic aromatic hydrocarbon emission tracing the CO-dark H 2 gas heating where [C ii] and [O i] emit. Conclusions. We present an innovative spectral decomposition method that allows statistical trends to be derived for the molecular gas content using CO, [C ii], and H i profiles. Our study highlights the importance of velocity-resolved photodissociation region (PDR) diagnostics and higher spatial resolution for H i observations as future steps.