Depends a bit how you define it - there isn't really a rule as far as I know. But the LS50, like most speakers with deepish waveguides, falls more towards the narrow directivity camp for me. I find this easier to assess from SPL/angle graphs and sometimes polar graphs than from the DI curves though, which more provide a general overview of DI per frequency than tell you how directivity with respect to angle.
With deep waveguides (and again, the LS50's midwoofer
is a waveguide to the tweeter), you tend to see a gradual tilting of the response at greater angles.
Here's the LS50 Wireless:
I tend to use the response at a certain angle out at a few frequencies to determine what I consider wide or narrow. Mainly I'm concerned with the region between 2-10K for soundstage purposes.
Just about every deep waveguide speaker I've tested follows this general tilting trend. (Note I only measured the LS50W out to 60 degrees as opposed to 75 on everything else). They are all down roughly 8-10 dB at 7K and 60 degrees, for instance.
Neumann KH80:
View attachment 48495
Buchardt S400:
View attachment 48496
KEF R3:
View attachment 48497
Compare that to, say, the Focal Chora. At 60 degrees, it's only down about 5dB at its furthest points until the tweeter becomes hyper-directional above 10Khz.
Granted, that's in part because of some bunching and unevenness - rarely do wide dispersion speakers have directivity as
controlled as those with a good, big ol' waveguide, but it's pretty easy to visualize in SPL graphs how much more energy they maintain throughout most of the frequency range. In the case of the Chora, it has more energy at 7KHz and
75 degrees than any of the above have at 60.
View attachment 48498
Similar story with the Q Acoustics Concept 20
View attachment 48499
Or the Devialet Phantom Reactor:
View attachment 48500
Or the Polk Legend L200:
View attachment 48501
Or the Sonus Faber Sonetto II:
View attachment 48502
Lots of wide dispersion speakers will tend to maintain more energy out throughout much of the presence and treble regions, before dropping off in the top octave or two. It effectively means the sound is more "similar" in level to the direct sound, and almost always sounds more expansive.
All of these speakers have either very shallow waveguides or none at all. Direct sound aside, it's actually pretty remarkable how similar these radiation patterns tend to be. within their respective "categories." The Phantom Reactor is the most unusual one because of its unusual shape, crossover, and choice of drivers.
One Toole study even suggests people prefer a wider directivity even if the response is less even, though it doesn't seem to be something with a ton of research. That's been my personal inclination though.
But again, it depends on your definition. A speaker with a big woofer will be more directive in the bass and mids, and that can have perceptual effects too. But since most modern speakers have similar directivity before the crossover, I tend to feel I think that 2K-10K region is most responsible for soundstage, that's what I mainly consider with regards to "wide" or narrow.
Here's one chart in Toole's book describing frequency regions for the perception of soundstage:
View attachment 48507