Why do readers fail to notice word transpositions, omissions, and repetitions? A review of recent evidence and theory

Abstract

Most readers have had the experience of initially failing to notice an omission or repetition of a function word, or a transposition of two adjacent words. In the present article, we review recent research investigating this phenomenon. We emphasize that failure to notice such errors is of substantial theoretical interest, given what we have learned about how systematically and incrementally readers inspect and process text. We endorse the idea that a process of rational inference may play a critical role, while we cast doubt on the idea that failure to notice errors arises from parallel processing of multiple words. We review a number of recent studies from our own laboratory that have investigated the relationship between eye movements during reading and noticing, or failing to notice, an error. While the conclusions from these studies are broadly consistent with a rational inference account, we find that when readers fail to notice an error, their eye movements generally show no indication that the error was registered at all. On its surface, this finding may be viewed as inconsistent with the idea that the rational inference process that enables readers to overlook errors is genuinely post‐perceptual. We suggest a mechanism by which eye movement control models could account for this finding.

A transposed-word effect across space and time: Evidence from Chinese

“…Consequently, this rapid sequential processing of words, which is possible when multiple words are available, might also provide a mechanism by which readers sometimes misanalyze word order (see Rayner et al, 2013). This would be consistent with a framework proposed recently by Huang and Staub (2021a), in which transposed-word effects are attributed to readers not integrating a word (word n) within a sentence representation in memory prior to identifying the following word (word n+1). Our SVP paradigm may provide more time for readers to integrate word n before word n+1 becomes available for processing, which could explain the smaller transposed-word effect for SVP compared to PVP displays in the present experiment.…”

“…Accordingly, while the transposed-word effect can be explained by parallel models, such as OB1-Reader (Snell & Grainger, 2019b), it may be compatible with serial models, especially if a level of post-perceptual word order flexibility is assumed, perhaps in line with "noisy channel" accounts of sentence processing (Gibson et al, 2013; for detailed discussion, see Huang & Staub, 2021a). The present research shows that, although readers sometimes misprocess word order, this need not entail that words are encoded in parallel.…”

Transposed-word effects when reading serially

“…Moreover, as already proposed in related work from our group (e.g., Dufour et al [ 15 ]; Pegado & Grainger [ 29 ]; Wen et al [ 27 , 30 ]), top-down constraints imposed by sentence-level structures (syntactic and semantic) would contribute to transposed-word effects independently of presentation mode (the same holds for post-lexical integration accounts of transposed-word effects as proposed by Huang & Staub). Thus, the combination of top-down sentence-level constraints (or post-lexical integration processes) plus noisy bottom-up processing provides a means to capture transposed-word effects under serial reading conditions (for further discussion of the serial vs. parallel reading debate see Huang & Staub [ 12 ]; Snell & Grainger [ 31 ]).…”

Barriers and enabling factors for error analysis in NLG research

“…Evaluation criterion: Humans are known to miss some errors when reading (Huang and Staub, 2021), and whether their annotations for one criterion might affect their subsequent reading and annotation of the remaining text is unknown. Asking annotators to consider multiple criteria simultaneously could compound this problem, increasing both disagreement and the volume of missed errors.…”