<![CDATA[This study analyzes the immediate context of sebenarnya in the first 10,000 concordance lines (about 22% of 44,738 hits) of the Indonesian Web corpus (IndonesianWaC, approximately 109 million tokens), accessed through Sketch Engine. The tokens directly to the left (L1) and right (R1) of the node word were counted, positional enrichment was calculated against a local baseline, and each line was assigned a function label using ordered, cue-based rules. The right-hand context is dominated by predicate-initial items, such as merupakan (18.4×), sudah (12.5×), and adalah (8.1×), which reflects the syntactic position of the adverb more than its pragmatic function. The left-hand context is most often yang (18.4%), a sentence boundary (14.3%), or a comma (8.0%). Cue-based labels show reality assertion (12.7%), correction or counter-expectation (11.8%), attributive or clause-final use (10.7%), interrogative emphasis (7.6%), comma-set-off discourse marking (4.2%), and first-person stance (2.4%), while 50.6% of lines matched no cue. No single function dominates, and the discourse-marker reading is narrower than commonly assumed. These findings are limited by unvalidated automatic labels, a non-random sample, and the written web register.]]>
Copyrights © 2026
