The theoretical study reported here is devoted to diluted GaInNAs nitride alloys, and focuses on correlation between local chemical atomic neighborhood and atomic distances. The model used is Valence Force Field approximation, and we model a periodic Ga1−xInxNyAs1−y alloy film deposited on a GaAs substrate. The surface is dimerised (for simplicity we consider 2×1 anion rich surfaces). First starting from a random film, first and second nearest neighbor distances are calculated, and the corresponding histogram drawn. Then a pseudo-annealing process is simulated by allowing the N atoms to choose their optimal location. This pseudo-annealing strongly enhances the number of In-N bonds, in agreement to experimental studies. From this statistical study, fine structure of each peak of the histograms is shown to be not directly related to a given surrounding chemical distribution up to 3rd nearest neighbors. The positions (in distance) of the peaks appear not to be modified by alloy composition nor alloy segregation, which only alter relative intensities and peak shapes. Last, we consider stripes of In-rich and In-poor zones: calculated energy variations show a strong tendency for N atoms to completely desert In-poor zones.