Reading without spaces revisited: The role of word identification and sentence-level constraints

Abstract

Reading without spaces revisited: The role of word identification and sentence-level constraints. Acta Psychologica, Elsevier, 2019, 195, pp. A B S T R A C TThe present study examined the relative contribution of bottom-up word identification and top-down sentencelevel constraints in facilitating the reading of text printed without between-word spacing. We compared reading of grammatically correct sentences and shuffled versions of the same words presented both with normal spacing and without spaces. We found that reading was hampered by removing sentence structure as well as by removing spaces. A significantly greater impact of sentence structure when reading unspaced text was found in probe word identification accuracies and total viewing times per word, whereas the impact of sentence structure on the probability of making a regressive eye movement was greater when reading normally spaced text. Crucially, we also found that the length of the currently fixated word determined the amplitude of forward saccades leaving that word during the reading of unspaced text. We conclude that the relative ease with which skilled readers can read unspaced text is due to a combination of an increased use of bottom-up word identification in guiding the timing and targeting of eye movements, plus an increased interactivity between word identification and sentence-level processing.

OB1-reader: A model of word recognition and eye movements in text reading.

“…latsek , 1998;Mirault, Snell, & Grainger, 2018). OB1 would deal with the removal of spaces by activating all words containing present bigrams, regardless of word length.…”

“…Further, whereas low-level visual information (e.g., word length) may not be available at first glance during unspaced reading, it is conceivable that readers nonetheless engage a sentence-level representation. For instance, Mirault et al (2018) found that, even in unspaced reading, saccade amplitudes were influenced by the length of fixated as well as upcoming words, suggesting that readers mentally parse unspaced text into separate words at quite a rapid pace, conceivably driven by lexical identification processes as well as top-down expectations. In any case, accounting for reading of unspaced text remains an interesting challenge for ongoing model development, especially considering the possibility of accommodating alphabetic writing systems that do not use spaces, such as Thai (e.g., Winskel, Radach, & Luksaneeyanawinc, 2009).…”

New Perspectives on Serialism and Parallelism in Oculomotor Control During Reading: The Multi-Constituent Unit Hypothesis

“…In this case, readers might engage in a more serial word-by-word identification strategy rather than the parallel process that occurs during spaced text reading. The question arises regarding how such word identification processes might operate in the absence of word boundaries in reading unspaced text [34]. Finally, when word length information is not immediately obvious in reading unspaced languages such as Chinese, a language in which it has been shown that the length of a word influences word identification and saccadic targeting [35], it is an interesting challenge for OB1 to explain how readers segment the text into words in order to use word length information to keep track of word order and maintain a sentence-level representation.…”

Lesen in der Fremdsprache Französisch. Kompetenzen von Drittklässlerinnen und Drittklässlern mit unterschiedlichen Schrift- und Sprachfähigkeiten in der Erstsprache Deutsch

“…Grainger (2018) sieht in diesen Prozessmechanismen auch eine mögliche Erklärung dafür, weshalb die Worterkennung in Sätzen ohne Wortgrenzen möglich ist und geübte Leserinnen und Leser darin wenig Schwierigkeiten zeigen: <DasjungePferdspringtüberdenZaun>. Zu einem ähnlichen Schluss kommen auch Mirault et al (2019a):…”

“…Morpho-semantische und morpho-syntaktische Leseprozesse Unauffällige Leserinnen und Leser wenden Wortidentifikationsprozesse beim Lesen von Sätzen an. Zum Erkennen von Wortgrenzen müssen einerseits Worte und sublexi kalische Einheiten (Morpheme) erkannt und andererseits bereits erste syntaktische Analysen vorgenommen werden (lokale Kohärenzbildung) (Mirault et al, 2019a). Genau diese Fähigkeiten werden in Aufgabenformaten zur Überprüfung von morpho-semantischen (Wortidentifikation) und morpho-syntaktischen (lokale Kohärenzbildung) Wortlesekompetenzen überprüft.…”