Language development follows predictable checkpoints. A toddler should be saying multiple single words by $18$ months and combining two words by $24$ months. At $20$ months a child limited to one word has clear expressive language delay, which is never something to simply watch.
The clinician's job is to find a cause. The leading and most readily correctable explanation for delayed speech is undetected hearing loss, so audiometry plus a structured speech-language assessment is the logical first investigation. Identifying and treating a hearing deficit early can restore normal language trajectory.
Reassurance, deferral to age $5$, or an outright autism label are all inappropriate: the first two ignore the delay and the third jumps to a diagnosis without excluding hearing impairment, especially when social and motor development are reportedly normal.
\[\boxed{\text{Next step} = \text{Speech and hearing evaluation}}\]