1: \begin{abstract}

2:   % Context

3:   Radial velocity fluctuations on the plane of the sky

4:   are a powerful tool for studying the turbulent dynamics of emission line regions.

5:   % What we did

6:   We conduct a systematic statistical analysis

7:   of the \halpha{} velocity field for

8:   a diverse sample of 9 \hii{} regions,

9:   spanning two orders of magnitude in size and luminosity,

10:   located in the Milky Way and other Local Group galaxies.

11:   % How we did it

12:   By fitting a simple model to the second-order spatial structure function

13:   of velocity fluctuations, we extract three fundamental parameters:

14:   the velocity dispersion,

15:   the correlation length,

16:   and the power law slope.

17:   We determine credibility limits for these parameters in each region,

18:   accounting for observational limitations of noise,

19:   atmospheric seeing, and the finite map size.

20:   % What we found

21:   The plane-of-sky velocity dispersion is found to be a better diagnostic

22:   of turbulent motions than the line width, especially for lower

23:   luminosity regions where the turbulence is subsonic.

24:   The correlation length of velocity fluctuations is found to

25:   be always roughly 2\% of the \hii{} region diameter,

26:   implying that turbulence is driven on relatively small scales.

27:   No evidence is found for any steepening of the structure function

28:   in the transition from subsonic to supersonic turbulence,

29:   possibly due to the countervailing effect of projection smoothing.

30:   Ionized density fluctuations are too large to be explained by the action

31:   of the turbulence in any but the highest luminosity sources.

32:   A variety of behaviors are seen on scales larger than the correlation length,

33:   with only a minority of sources showing evidence for homogeneity on the largest scales.

34: \end{abstract}

35: