Step 1: Write the governing formulas.
For a whiskbroom sensor at altitude $H$, the optical footprint (GIFOV) is $\text{GIFOV} = \beta H$, where $\beta$ is the detector's angular instantaneous field of view (a fixed optical/detector property). The ground sample spacing, i.e. the pixel size, is instead $\text{Pixel size} = \omega \, \Delta T \, H$, where $\omega$ is the mirror's angular scan rate and $\Delta T$ is the time between successive electronic samples.
Step 2: Compare the two formulas.
$\text{GIFOV} = \beta H$ depends only on the fixed optical parameter $\beta$ and the altitude $H$; it says nothing about how often the detector output is digitized. $\text{Pixel size} = \omega \Delta T H$ depends on the scan rate $\omega$, the altitude $H$, and crucially the sampling interval $\Delta T$, which is the only variable in this expression that is not shared with the GIFOV formula.
Step 3: Isolate the controlling factor.
For a given sensor design, $\omega$, $\beta$, and $H$ are essentially fixed by the hardware and orbit, leaving $\Delta T$ as the parameter an engineer tunes (through the digitizer's clock rate) to make the pixels finer or coarser. Doubling $\Delta T$ doubles the pixel size directly, while GIFOV stays unchanged because it never appears in that relationship.
Step 4: Confirm GFOV is a separate, swath-level quantity.
GFOV corresponds to the full scan angle $\theta_{\text{max}}$ times $H$, describing swath width, an entirely different geometric quantity from the spacing between adjacent pixels, so it cannot influence pixel size.
Step 5: Conclude from the formulas.
Since pixel size is a function of $\Delta T$ and GIFOV is not, and since GFOV is unrelated to pixel spacing altogether, only the sampling interval $\Delta T$ governs the pixel size.
\[ \boxed{\text{Pixel size} \propto \Delta T} \]