Which of the following correctly describes a feature of Simple LR (SLR) parsing?

SLR(1) parsing is a bottom-up (shift-reduce) LR technique: it builds an LR(0) automaton from the grammar's items and then places its reduce actions using FOLLOW sets computed from the grammar -- a reduce-by-A entry is added only under the terminals in FOLLOW(A) for a completed item A → α•. (If a shift-reduce or reduce-reduce conflict still remains after this, the grammar is not SLR(1).) This FOLLOW-set step is exactly what separates SLR from a plain LR(0) parser and from the stronger canonical-LR(1)/LALR(1) parsers, which instead compute per-state, context-sensitive lookaheads.

Building the SLR parsing table therefore means: for every LR(0) state containing a completed item A → α•, a reduce-by-that-production action is added in every ACTION-table column headed by a terminal in FOLLOW(A). This is precisely the feature that correctly describes SLR parsing among the options given.

Each of the other statements fails for a specific reason:

• Claiming parsing tables are not used contradicts the core SLR mechanism -- SLR is entirely table-driven, using ACTION and GOTO tables built from the LR(0) automaton.

• Claiming a leftmost derivation is constructed describes top-down (LL) parsing, not SLR -- every bottom-up/LR-family parser, SLR included, instead produces a rightmost derivation of the input in reverse.

• Claiming SLR relies on backtracking is false -- its table-driven shift/reduce decisions are fully deterministic, so no backtracking over the input is ever needed.

So the FOLLOW-set-based reduce-action rule is the feature that correctly describes SLR parsing.