• SILENT, ORAL, L1, L2, FRENCH AND ENGLISH READING THROUGH EYE MOVEMENTS AND MISCUES

      Goodman, Yetta M.; O'Brien de Ramirez, Kathleen; Goodman, Yetta M.; Waugh, Linda R.; Garrett, Merrill F. (The University of Arizona., 2008)
      During 24 silent and oral readings of Guy de Maupassant and Arthur C. Clarke short stories (1294 and 1516 words) by proficient multilinguals, movement of the left eye was tracked and utterances were recorded. Three hypotheses investigate universality in the reading process: reading in English is similar in reading speed, miscues, and eye movements to reading in French (chapter 4); reading in a first, or native language (L1), is similar in reading speed, miscues, and eye movements to reading in a second, or later acquired, language (L2) (chapter 5); silent reading is similar to oral reading in reading speed and eye movements (chapter 6). Hypothesis are partially confirmed; implications are drawn for teaching and research.Silent reading is consistently faster than oral reading, with a mean difference of 28.7%. Reading speed is similar in English and French, but interacts differently with language experience: L2 readers of English read 50% slower than L1 readers, while in French, L2 readers read 13% faster.Retelling scores demonstrate a slight comprehension advantage for oral reading over silent, a wider range after oral than after silent, L1 readers having a slight advantage over L2 readers, and improved scores after second readings. Proscribing rereading to increase oral accuracy may disadvantage some readers: Second oral readings in English (but not in French) produced more miscues than first oral readings. This requires further study with tightly controlled groups. Overall, English readings produced 36% more miscues than French readings.Mean fixation durations are slightly longer during silent than oral reading, and show little variation between English and French reading. Wide variation in reading speed (L1/L2, silent/oral) is not reflected in mean eye fixation durations, although language dominance show an effect in French, where fixations during L1 readings are 18.6% shorter than during L2 readings.Individual variation is a factor. Emotional affect, poetic style, construction of syntax, and attention to metaphor are all observed in this EMMA data. Future analysis of this database may look at anaphoric relations, metaphor, how texts teach; and how readers develop narrative, verb phases, syntactic, semantic, and pragmatic relations in complete textual discourse.
    • The Ink in Our Speech: Influence from Orthographic Complexity in Speech Production

      Bever, Thomas; Grippando, Shannon; Hammond, Michael; Archangeli, Diana (The University of Arizona., 2020)
      In this dissertation, I present three studies that expand our field’s understanding of the role of orthography in speech production, particularly the interaction between the complexity of an orthographic form and speech duration. The studies address this topic in Japanese, English, and novel words. To the author’s knowledge, there has been no previous research that explores the interaction between the complexity of an orthographic form and speech duration in a language with a non-alphabetic writing system. The results of these studies are presented as evidence that orthographic units beyond letters can influence speech duration, although non-alphabetic orthographic units may interact with speech differently than letters. In the first study I build upon previous research that finds a correlation between spelling and speech duration: The more letters in the orthographic representation of a segment, the longer that segment is produced (Brewer, 2008). Until the time of this study, this effect had only been examined in languages with alphabetic writing systems. I further investigate this behavior in Japanese, a language with a logography. Native Japanese-speaking participants were audio-recorded reading pairs of homophonous words that varied in number of pen strokes or number of whole characters. Two-character words were produced significantly longer than 1-character words, however there was no significant effect from number pen strokes on speech duration.The second study directly expands upon the first study by conducting 5 multi-day novel word learning experiments to investigate the effects on speech production from different measures of orthographic complexity. Each experiment used a novel orthography that varied in the orthographic complexity among homophonous triplets by: a) number of pen strokes; b) number of whole graphemes; c) number of pen strokes and whole graphemes; d) number of repeating sub-graphemic components; e) number of non-repeating sub-graphemic components. Significant effects from orthographic complexity on speech duration were observed in reading tasks for c (number of pen strokes and whole graphemes) and e (number of non-repeating sub-graphemic components).The third study addresses the question of how the speech patterns observed in the first two studies and previous research may develop. One possibility is that auditory cues, not orthography, drive the development of this behavior. Five auditory discrimination experiments were conducted to determine participants’ sensitivity to durational contrasts at lengths similar to those observed in the previous studies. The results showed no evidence of sensitivity at the critical levels of speech duration: Participants showed an inconsistent pattern of sensitivity starting at 2-4 times the length of some speech effects found in previous studies. The results of these studies are discussed in relation to two hypotheses that have been proposed to account for how orthography may influence speech: Phonological Restructuring and On-Line Activation (Brewer, 2008; Perre et al., 2009). Phonological Restructuring argues that through off-line processes, orthographic information can alter underlying phonological forms. On-Line Activations argues that surface forms can be influenced by activation of orthographic information just before articulation occurs. I argue that the results of the current dissertation necessitate modifications to these hypotheses in terms of task dependency and compatibility limitations for categories of writing systems. I discuss how orthography may interact with speech production processes under these two accounts to influence speech. Finally, the number of orthographic units that affected duration varied across studies, differing by writing system and category of orthographic units. To account for this variability, I also propose constraints for the minimum and maximum number of orthographic units required to influence speech, specific to categories of writing systems and types of orthographic units.