We present an investigation of the low-frequency radio and ultraviolet properties of a sample of $simeq$10,500 quasars from the Sloan Digital Sky Survey Data Release 14, observed as part of the first data release of the Low-Frequency-Array Two-metre Sky Survey. The quasars have redshifts $1.5 < z < 3.5$ and luminosities $44.6 < log(L_{text{bol}}/text{erg s}^{-1}) < 47.2$. We employ ultraviolet spectral reconstructions based on an independent component analysis to parametrize the CIV$lambda$1549-emission line that is used to infer the strength of accretion disc winds, and the HeII$lambda$1640 line, an indicator of the soft X-ray flux. We find that radio-detected quasars are found in the same region of CIV blueshift versus equivalent-width space as radio-undetected quasars, but that the loudest, most luminous and largest radio sources exist preferentially at low CIV blueshifts. Additionally, the radio-detection fraction increases with blueshift whereas the radio-loud fraction decreases. In the radio-quiet population, we observe a range of HeII equivalent widths as well as a Baldwin effect with bolometric luminosity, whilst the radio-loud population has mostly strong HeII, consistent with a stronger soft X-ray flux. The presence of strong HeII is a necessary but not sufficient condition to detect radio-loud emission suggesting some degree of stochasticity in jet formation. Using energetic arguments and Monte Carlo simulations, we explore the plausibility of winds, compact jets and star formation as sources of the radio quiet emission, ruling out none. The existence of quasars with similar ultraviolet properties but differing radio properties suggests, perhaps, that the radio and ultraviolet emission is tracing activity occurring on different timescales.