1st South European Workshop on Stress Assignment in Reading

1–2 September 2008

Institute for Language & Speech Processing

Athens, Greece

Venue

The workshop will take place at the Institute for Language & Speech Processing (lecture room, ground floor), “Athena” Research Center. Address: Artemidos 6 & Epidavrou, GR-15125 Maroussi, Greece. Directions to reach ILSP by car or public transportation can be found at the institute web page http://www.ilsp.gr/contact_eng1.html.

Schedule

Monday, September 1st
10:00		Welcome
10:05	Paizi	Stress assignment in Italian young and adult readers: Friendship vs. dominance
10:45	Arduino	The role of stress neighborhood on nonword reading in Italian
11:25	Wilson	Hide and seek: Does stress assignment unpredictability make age of acquisition effects pop up in Italian?
11:55		Coffee
12:25	Colombo	Stress priming in reading nonwords in Italian (I)
13:05	Deguchi	Stress priming in reading nonwords in Italian (II)
13:30		Lunch
14:45	Protopapas	Sources of information for stress assignment in reading Greek: From the onset of children's fluent reading through expert adult performance
15:25	Gerakaki	Is stress specified in the lexicon? Design of a cross-modal priming study in Spanish and Greek
15:50	Grimani	Word endings from productive morphology as a source of stress assignment information in Greek
16:10		Coffee
16:40	Gutiérrez	Stress awareness and lexical stress assignment in Spanish
17:15		General discussion of stress assignment in reading
18:00		End of day 1
Tuesday, September 2nd
10:00	Tzakosta	Multiple Parallel Grammars in the acquisition of stress: implications for stress assignment in reading
10:45	Zevin	Ad hoc functional units in spelling to sound: Grain size issues in cross-linguistic modeling of reading
11:30	Monaghan	Cross-linguistic studies of probabilistic cues to stress position
12:15		Coffee
12:45		Discussion of developmental and modeling issues
13:45		Lunch
15:00		Discussion of cross-linguistic collaborations
17:00		End of meeting

Participants

Arduino, Lisa Saskia

Institute of Psychology

University of Urbino, Italy

lisa.arduino@uniurb.it

Boureux, Magali

Dipartimento di Psicologia Generale

Università di Padova, Italy

magali.boureux@unipd.it

Burani, Cristina

Institute for Cognitive Sciences and Technologies (ISTC-CNR)

Rome, Italy

cristina.burani@istc.cnr.it

Colombo, Lucia

Dipartimento di Psicologia Generale

Università di Padova, Italy

lucia.colombo@unipd.it

Deguchi, Chizuru

Dipartimento di Psicologia Generale

Università di Padova, Italy

chizuru.deguchi@unipd.it

Gerakaki, Svetlana

Graduate Program in “Basic and Applied Cognitive Science”

University of Athens, Greece

sgerak@phs.uoa.gr

Grimani, Katerina

Graduate Program in “Basic and Applied Cognitive Science”

University of Athens, Greece

kathi_32@hotmail.com

Gutiérrez-Palma, Nicolás

Departamento de Psicología

Universidad de Jaén, Spain

ngpalma@ujaen.es

Kappa, Ioanna

Department of Philology, Section Linguistics

University of Crete, Rethimno, Greece

kappa@phl.uoc.gr

Kapnoula, Efi

Graduate Program in “Basic and Applied Cognitive Science”

University of Athens, Greece

mskapnoula@yahoo.gr

Monaghan, Padraic

Department of Psychology

Lancaster University, UK

p.monaghan@lancaster.ac.uk

Outos, Kostas

Graduate Program in “Basic and Applied Cognitive Science”

University of Athens, Greece

outosk@hol.gr

Paizi, Despina

Institute for Cognitive Sciences and Technologies (ISTC-CNR)

Rome, Italy

despina.paizi@istc.cnr.it

Protopapas, Athanassios

Institute for Language & Speech Processing, “Athena” Research Center

Maroussi, Greece

protopap@ilsp.gr

Ševa, Nada

Department of Psychology

University of York, UK

nadaseva@gmail.com

Tzakosta, Marina

Department of Preschool Education

University of Crete, Rethimno, Greece

martzak@gmail.com

Vlahou, Eleni

Institute for Language & Speech Processing, “Athena” Research Center

Maroussi, Greece

evlahou@ilsp.gr

Wilson, Maximiliano

Institute for Cognitive Sciences and Technologies (ISTC-CNR)

Rome, Italy

maximiliano.wilson@istc.cnr.it

Zevin, Jason

Sackler Institute for Developmental Psychobiology

Weill-Cornell Medical College, New York, USA

jdz2001@med.cornell.edu

Zoccolotti, Pierluigi

Department of Psychology

University of Rome “La Sapienza”, Italy

pierluigi.zoccolotti@uniroma1.it

Abstracts

Stress assignment in young and adult Italian readers: Friendship vs. dominance

Despina Paizi^1,2, Pierluigi Zoccolotti^2,3,1, & Cristina Burani¹

¹ Institute for Cognitive Sciences and Technologies (ISTC-CNR), Rome

² Department of Psychology, University of Rome “La Sapienza”

³ Neuropsychological Unit, IRCCS Fondazione Santa Lucia, Rome

Stress assignment to Italian polysyllabic words is unpredictable, because stress is neither marked nor predicted by rule. Stress assignment, especially to low frequency words, could be a function of stress regularity (stress assignment by default) or determined by the number of words that share the same stress pattern and final orthographic/phonological sequence (stress neighbourhood). Six experiments investigate stress assignment in adult and young, skilled and dyslexic, readers. Skilled readers, young and adult, were only affected by word frequency, irrespective of stress pattern, arguing against stress assignment by default. Dyslexic children, although also affected by word frequency, showed a tendency to make more stress regularisation errors on low frequency words. The effect of stress neighbourhood on low frequency word reading was found to be significant and independent of stress pattern for all groups of readers, including developmental dyslexics. Words with many stress friends were read faster and more accurately than words with many stress enemies, whether regularly or irregularly stressed. Young skilled and dyslexic readers were as sensitive as adults to stress neighbourhood effects. It is argued that stress is assigned lexically and stress neighbourhood, rather than stress regularity, determines stress assignment to low frequency words.

The role of stress neighborhood on nonword reading

Lisa S. Arduino^1,2, Despina Paizi^2,3, Simone Sulpizio², & Cristina Burani²

¹ Institute of Psychology, University of Urbino

² Institute for Cognitive Sciences and Technologies (ISTC-CNR), Rome

³ Department of Psychology, University of Rome “La Sapienza”

With four experiments we investigated the role of stress neighbourhood in reading aloud Italian nonwords. Stress neighbourhood in nonwords was manipulated on the basis of the final sequences. Two lists of nonwords were created: A list of nonwords with final sequences more frequent in words with regular stress pattern (on the penultimate syllable) and a list that contained nonwords with final sequences found predominantly in irregularly stressed words (on the antepenultimate syllable). Each list contained an equal number of nonwords matched on several lexical and sublexical variables, in order to exclude reading by analogy to real words. Analyses were conducted on both reaction times and pronunciation errors, whereas stress assignment was coded as coherent or incoherent according to stress neighbourhood. The results showed that stress assignment to nonwords is not governed by a default rule. Final sequences may only partially drive stress assignment. In all four experiments, when nonwords were assigned the irregular stress pattern, they were pronounced faster than when the regular stress pattern was assigned, irrespective of stress neighbourhood. 

Hide and seek: Does stress assignment unpredictability make age of acquisition effects pop up in Italian?

Maximiliano Wilson & Cristina Burani

Institute for Cognitive Sciences and Technologies (ISTC-CNR), Rome

It has been claimed that word age-of-acquisition (AoA) effects are dependent on the orthography- phonology mapping consistency of a language (Ellis & Lambon Ralph, 2000; Monaghan and Ellis, 2002). Then, little or no effects of AoA should be found in reading aloud a transparent orthography as Italian. That has been the case so far with mixed item sets that had both penultimate- and antepenultimate- stressed words (e.g., Burani et al., 2007). We will present preliminary results of a word naming task with 3-syllable words in which AoA (early and late) and stress assignment (penultimate- and antepenultimate- stressed) varied orthogonally. Sets were presented either mixed or blocked by stress. AoA effects were found in all conditions, except for the most predictable one: all penultimate-stressed words. In a second experiment, AoA and length (2 and 3 syllables) were manipulated. The mixed condition (2- and 3-syllable words together) reproduced the results found by Burani et al. (2007): No main effect of AoA. Nevertheless, when only 3-syllable words were read, we found AoA effects. In conclusion, it seems that when stress assignment is unpredictable reading becomes “deeper” and AoA effects pop up—even in a transparent language as Italian. 

Stress priming in reading nonwords in Italian.

Lucia Colombo & Chizuru Deguchi

Dipartimento di Psicologia Generale, Università di Padova

We present some data investigating stress priming in interaction with stress neighborhood. Data from production paradigms (e.g., Schiller et al., 2004) show that stress priming is not easily observed. However, those studies priming was induced through picture naming, a task that necessarily involves lexical activation. Thus, stress patterns may be automatically retrieved. In reading nonwords, stress cannot be assigned through lexical retrieval. In our experiments, both primes and targets were nonwords. To insure that prime stimuli were assigned a certain type of stress by most participants, stress neighbourhood information was manipulated, by constructing nonwords with endings present in strong stress neighbourhoods, that is, neighbourhoods in which a certain stress type dominated, or weak neighbourhoods, in which both stress patterns were present. Priming effects were measured for nonword targets with dominant stress (penultimate syllable – strong neighbourhood), with non-dominant stress (initial syllable – strong neighbourhood), and with weak neighbourhood stress (about 50% probability for each stress pattern). The context was given by dominant and non-dominant stress nonwords.  Priming effects were present for all types of targets, but were stronger in the weak neighbourhood condition, suggesting that priming and stress neighbourhood strength interact.

Sources of information for stress assignment in reading Greek: From the onset of children's fluent reading through expert adult performance

Athanassios Protopapas¹ & Svetlana Gerakaki²

¹ Institute for Language & Speech Processing, “Athena” Research Center

² Graduate Program in “Basic and Applied Cognitive Science”, University of Athens

In Greek, stress is always marked with a special diacritic. Stress information may also be available from the lexicon once words are recognized on the basis of the segmental sequences. We have investigated the relative contribution of these sources of information in children and adults, using pseudowords resembling or not resembling words, presented with or without a diacritic, consistent or inconsistent with the source word. We have consistently found strong lexical influences on stress assignment, as well as a preference for pentultimate stress in the absence of specific information. Recently, we investigated the development of decoding the stress diacritic in children from Grades 2, 3, and 4. Strong and increasing lexical effects were found through the grades. The effect of the diacritic grows more rapidly, approaching but not reaching the lexical effect at higher grades. It seems that decoding of the diacritic is not the preferred option for developing readers, only becoming efficiently utilized at advanced reading stages, despite constant availability and reliability. These findings raise issues regarding the incorporation of metrical patterns in reading models. More challenging issues concern the apparent delay of fluently decoding the stress diacritic, relative to graphophonemic decoding, and the role of sight-vocabulary transparent orthographies.

Is stress specified in the lexicon? Design of a cross-modal priming study in Spanish and Greek

Svetlana Gerakaki¹, Nicolás Gutiérrez-Palma², & Athanassios Protopapas³

¹ Graduate Program in “Basic and Applied Cognitive Science”, University of Athens

² Departamento de Psicología, Universidad de Jaén

³ Institute for Language & Speech Processing, “Athena” Research Center

Recent findings on stress assignment in reading suggest that stress patterns may be derived from the lexicon as words are recognized on the segmental level from the written letter sequences. We present the design of a cross-linguistic study in Greek and Spanish to investigate whether metrical patterns are in fact specified in the mental lexicon and become automatically activated in reading. If metrical patterns are unspecified (e.g., in case a “default” applies), or if they are computed late in the word recognition process, then visual processing of a word will not produce rapidly a corresponding metrical pattern. If, however, metrical patterns are specified in the lexicon, then they might become activated early, with the segmental phonological representation of the word. In a cross-modal priming paradigm, we will use visual word and nonword primes and auditory word and nonword targets either matched or mismatched in stress.  All stimuli are carefully matched in a host of psycholinguistic parameters. If stress is specified in the lexicon then we should observe reduced response times to both word and nonword targets when they follow same-stress word primes (but not nonword primes) regardless of stress position. Alternative predictions follow from the default-rule hypothesis.

Word endings from productive morphology as a source of stress assignment information in Greek

Katerina Grimani¹ & Athanassios Protopapas²

¹ Graduate Program in “Basic and Applied Cognitive Science”, University of Athens

² Institute for Language & Speech Processing, “Athena” Research Center

In Greek, many morphological suffixes constrain stress position.  We investigated whether readers use this knowledge for stress assignment in addition to previously determined stress assignment information sources such as the diacritic and the lexicon.  We used pseudowords ending with suffixes from productive morphology that are heavily biased towards a specific stress pattern. We also used real words with these suffixes as well as real words ending in the same letter sequences but stressed contrary to the suffix. The stimuli were presented without a diacritic as well as with a diacritic on a position consistent or inconsistent with the suffix bias. The results will show the contribution of morphological information for stress assignment relative to the orthographic mark (stress diacritic).

Stress awareness and lexical stress assignment

Nicolás Gutiérrez-Palma

Departamento de Psicología, Universidad de Jaén

Recent studies suggest that prosody may affect reading acquisition. On the one hand, prosodic sensitivity may facilitate the detection of some phonological units, such as the English rime, that are important for reading (Goswami, Thonson, Richardson, Stainthorp, Hughes, Rosen, & Scott, 2002); on the other hand, prosodic skills may also be necessary for learning how to assign prosody in reading aloud (Whalley & Hansen, 2006). The present study focuses on lexical stress, which refers to prosody at the word level. Participants were adults who performed phonological awareness, stress awareness (i.e., to indicate the stressed syllable in a three-syllable word), and reading aloud (pseudowords) tasks. Dependent variables were errors at assigning lexical stress, and reaction times for pseudowords read correctly. Results showed that stress awareness explained a significant amount of the variance of the reaction times, but not of the stress errors. These data are discussed in the context of the role of prosodic skills in the acquisition and automation of reading prosody.

Multiple Parallel Grammars in the acquisition of stress: Implications for stress assignment in reading

Marina Tzakosta

Department of Preschool Education, University of Crete

Stress is one of the major theoretical issues in Greek language development because of the lexical character of the language; the position of stress is unpredictable in Greek, i.e. it falls on one of the last three syllables of the word. However, accentual landing sites are not phonology-dependent only; morphology, and more specifically word derivation, determines stress assignment. This results in infants being confronted with metrically variable forms in Greek a fact which marks ambiguity in the perception and the production of stress. More specifically, children are exposed to forms of variable syllabic length which are being assigned stress on various word positions. Therefore, Greek children do not perceive a clear accentual system and, hence, they produce multiple metrically ambiguous forms. For example, trisyllabic W(eak)S(trong)W(eak) words may be produced as either WS or SW (by syllabic truncation) or SW or WS (by truncation & stress metathesis). Perceptual and/or production ambiguity may last till the first school years resulting in reading problems. In this talk we will make some proposals regarding the solution of such presumptive problems.

Cross-linguistic studies of probabilistic cues to stress position

Padraic Monaghan¹, Nada Seva², and Joanne Arciuli³

¹ Department of Psychology, Lancaster University

² Department of Psychology, University of York

³ Faculty of Health Sciences, University of Sydney, Australia

The first few and the last few letters of words in English provide an enormous amount of information about stress position. But how are these beginnings and endings used by children learning to read? We will present corpus analyses of reading-age appropriate corpora and developmental data indicating the changing reliance on beginnings and endings between ages 5 and 12. We will also present corpus analyses indicating the extent to which word beginnings and endings may be useful in reflecting stress position in other European languages, such as Italian, German, and Dutch.

Ad hoc functional units in spelling to sound: Grain size issues in cross-linguistic modeling of reading

Jason D. Zevin

Sackler Institute for Developmental Psychobiology, Weill-Cornell Medical College

In studies of the development of literacy, there are long-standing debates about the basic functional units of spelling-to-sound translation at the segmental level. Debates have centered on whether spelling-to-sound knowledge is best captured by “small” units – typically one or two letters that map onto a particular speech sound – or “large” units that may encompass most of a word. In computational models of the acquisition of reading, the functional units for mapping from spelling to sound turn out to be ad hoc, driven by the statistics of the mapping itself, and available to the learner to varying degrees from the earliest phases of learning.  In this presentation, I will discuss how PDP models of reading aloud can account for complex, qualitative patterns of performance in adults, as well as in children at the earliest stages of learning to read, and how this approach generalizes to the Chinese writing system. These findings from relatively deep orthographies may inform our thinking about stress assignment in shallower orthographies, as analogous issues – multiple, probabilistic cues emerging at different grain sizes – arise in both contexts.