Most headphones require far less voltage and current than speakers to achieve loud listening levels. High impedance headphones typically need a bit more voltage than low impedance headphones but less current, but neither need anywhere the same voltage levels as speakers. Of course, what a speaker needs also depends on its efficiency and other factors.
Just to put things in perspective. The Beyer DT990 250 ohm and most Grado models are rated around 96-98dB SPL per milliwatt. 1mW into 250 ohms (Beyer) is about 0.5Vrms. 1mW into 32 ohms (Grado) is about 0.18Vrms. Contrast this with a typical bookshelf speaker, with about 88dB/W/m efficiency rating. This means that to achieve 98dB at 1m distance, you need 10x more power (dB = 10 * log(P1/P2)), so 10 * log(10) = 10dB. So 10W into 8 ohms is needed, and that is ~9Vrms. You can see in this ad hoc example, the speaker needs 18x the voltage swing to achieve the same SPL as the Beyer, or 50x the voltage swing of the Grado. In reality, you sit farther away than 1m from the speakers so you need even more power than the calculation.
As for current, you can also calculate this. We determined that at ~98dB there is 0.5Vrms on the Beyer, so I = 0.5V/250 = 2mA. On the Grado it's 0.18V / 32 = ~6mA. But on the speaker? 9V / 8 = 1.13A! We're talking several orders of magnitude more here.
As for adding resistance to reduce speaker level output to headphone levels, you need to attenuate by at least 20x if not more. This means some fairly high value series resistors (several hundred ohms typical). Even if the headphone load impedance varies by a few ohms it will introduce a non-trivial amount of frequency response change. Many headphones have rather non-linear impedance curves while others are reasonably flat, so the response change is headphone-dependent. It doesn't change the fact that this is simply not a good way to drive headphones.
