FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2043968327> ?p ?o ?g. }

• W2043968327 endingPage "1000" @default.
• W2043968327 startingPage "953" @default.
• W2043968327 abstract "Abstract The structure of lexical entries and the status of lexical decomposition remain controversial. In the psycholinguistic literature, one aspect of this debate concerns the psychological reality of the morphological complexity difference between compound words (teacup) and single words (crescent). The present study investigates morphological decomposition in compound words using visual lexical decision with simultaneous magnetoencephalography (MEG), comparing compounds, single words, and pseudomorphemic foils. The results support an account of lexical processing which includes early decomposition of morphologically complex words into constituents. The behavioural differences suggest internally structured representations for compound words, and the early effects of constituents in the electrophysiological signal support the hypothesis of early morphological parsing. These findings add to a growing literature suggesting that the lexicon includes structured representations, consistent with previous findings supporting early morphological parsing using other tasks. The results do not favour two putative constraints, word length and lexicalisation, on early morphological-structure based computation. Acknowledgements We thank Jeff Walker for his invaluable help in conducting the MEG experiment, and acknowledge the collaboration of Paul Ferrari in setting up an earlier version of the compound MEG study. Supported by NIH R01 DC05660 to DP. Notes 1See Bertram & Hyönä (Citation2003) and Janssen, Bi, & Caramazza (Citation2006) for some estimates of the number of compounds in Finnish and English respectively, and Bauer (Citation2001), Plag (Citation1999), Hay & Baayen (Citation2002), among others, for various views on how ‘productivity’ might be defined and measured quantitatively. 2See Rayner (Citation1998) for a comprehensive review of eye-tracking methodological issues and results. Some challenges for the linking of linguistic computations to component measures in the eye-movement record include accounting for recognition effects that are often distributed over several fixations (e.g., Inhoff et al., Citation1996) and may be subject to parafoveal preview effects (e.g., Rayner & Pollatsek, Citation1989; but see Deutsch, Frost, Pollatsek, and Rayner, Citation2000 for an interesting use of the parafoveal preview benefit to detect morphological decomposition in Hebrew). 3A related issue is whether effects found in isolated word tasks like lexical decision would extend to studies of reading. For example, Bertram et al. (Citation2000) found that sentence context seems to play a role in whether decompositional processing is evident for inflected Finnish words carrying the ambiguous suffix –jA. Fixation and gaze duration measures, as well as reading time measures in self-paced reading, show base frequency effects (although base frequency effects were typically weak on first fixation durations and reading times on the inflected word, and more robust in gaze duration and in measures on the following word in both methodologies) while in a simple lexical decision task, base frequency effects were not evident. Hyönä, Vainio, and Laine (Citation2002) have found a distinction in the opposite direction – complexity effects in lexical decision but not reading – using Finnish case-marked complex words vs. single words. As regards compounds, constituency effects seem to hold not only in lexical decision tasks but also in eye-tracking studies of compound processing in sentence context (see Pollatsek & Hyönä, Citation2005 [Finnish], Andrews et al., Citation2004 [English], among others). 4Juhasz et al. (Citation2003) found similar constituent effects for English compounds in naming, lexical decision, and eye movements, finding some effect of first constituent frequency in naming, lexical decision, and first fixations, but more robust second constituent effects in the two behavioural measures and in gaze durations. 5As one reviewer notes, priming for opaque forms remains evident in cross-modal tasks in Semitic languages. Indeed, Plaut and Gonnerman (Citation2000) explored an account of this phenomenon within a distributed-connectionist model without abstract morphological structure, attributing the difference across languages to the overall level of complex word processing in the language. Plaut and Gonnerman (Citation2000) conducted simulations among artificial languages including a distinction in terms of a ‘morphologically rich’ language (modelled on Semitic languages), and a ‘morphologically poor’ language (modelled on English). The intuition behind these simulations is that the morphological properties of a language, such as having a high number of morphologically complex forms, leads to parsing of opaque forms in the morphologically rich, but not the morphologically poor language. However, at least two points are relevant here: (1) the ‘morphologically poor’ languages like English show robust morphological priming regardless of semantic transparency in masked priming tasks, contrary to the predictions of the typological explanation (see Feldman, Citation2000; Fiorentino, Citation2006; Rastle et al., Citation2000, Citation2004) and (2) it has been proposed that relatively more persistent priming in Semitic languages may be the result of the relatively lower level of formal overlap compared with English, thus reducing form-level inhibition which might otherwise counter morphological-level facilitation (see Forster, Citation1998). 6Sandra (Citation1990) used what amounts to a longer SOA, as Zwitserlood (Citation1994) notes. 7These priming results are not without controversy. As regards methodological concerns, the view that masked priming is relatively insensitive to overt strategic effects compared with overt priming has been challenged, for example by Masson and Bodner (Citation2003) and Masson and Isaak (Citation1999). The latter presents the argument that nonword priming effects suggest a pre-lexical locus of masked priming effects. However, Forster (Citation1998, Citation1999) and others argue that the findings on nonword priming can be explained in context of masked priming operating at the lexical level. The claim that results from the masked priming of complex words implicates the existence of morphological-level constituency has been challenged from the distributed-connectionist viewpoint (see the Models section below for more discussion). 8Planned contrasts of compound word versus control single word averaged log frequency were significant (p<.001) as were comparisons of the compounds with high frequency first constituents versus control single words, and compounds with low frequency first constituents versus controls (p<.02 and p<.009, respectively). ANOVA on letter length was also significant, [F(3, 56)/ 5.157, p<.004]. A contrast of the compound versus control average length was significant (p<.004) as was a contrast of high first constituent compound versus control length (p<.002). Letter length also significantly differed among the control sets [paired t-test, t(29)/ 2.8, p<.009, two-tailed]. Letter length among low first constituent frequency compounds versus controls did not differ significantly (p<.293). 9One speculation is that how the effects are manifested may be affected by task, item-set, and other factors thought to influence post-lexical processing. 10As regards the issue of when the first contact with the lexicon occurs, it is of the utmost concern to disentangle effects of visual word form properties with lexical properties such as word frequency. Given that the MEG responses from 100–200 ms are sensitive to properties of the visual word form (letter-length, discriminability in low-contrast presentations, etc.) it is important to consider the possible role of these properties when testing for early frequency effects. For example, while the length and frequency controls were reported in Assadollahi and Pulvermüller (Citation2003), there is no mention of orthographic/phonological regularity or probability controls. On the view that these properties are likely to correlate with frequency, this is of concern given the many studies showing the components around 100–200 ms post-onset are sensitive to letter-string encoding (e.g., Cornelissen et al., Citation2003; Tarkiainen et al., Citation1999), making the interpretation of the effects as lexical-level more difficult. Further differences which are potentially relevant, as noted by Assadollahi and Pulvermüller (Citation2003), are the differences in design among the studies showing earlier vs. later effects of frequency, and differences in how the frequency effects that were reported were reflected in the signal (amplitude modulation vs. latency modulation). 11The frequency counts used here are lemma frequency (the frequency of a word form and its inflectional variants). The lemma frequency counts for whole compounds and single words are counts for the whole word form uninterrupted e.g., by hyphenation, that is, in the same form that they are presented in the experiment. 12Note that Cobuild is advantageous due to its extremely large sample size, allowing a good estimate of word frequency among words not at the top of the frequency range, like the current items. However, we consulted the Francis and Kučera (Citation1982) analysis of the Brown Corpus (first published 1961; approximately 1 million tokens) for the sake of comparison. Only 60% of the compound words, and 70% of the single words were represented in that corpus. Nevertheless, measures of frequency (log frequency, raw frequency) and distribution (number of sources in which lemma appears, number of text samples in which lemma appears) showed that the subsets of these items were matched; log frequency differed by only 0.09 across conditions among the extant CW and SW on this count. In the Carroll, Davies, and Richman (Citation1971) American Heritage Intermediate Dictionary corpus (approximately 5 million tokens), approximately 85% of the stimuli were represented in the corpus. The mean log frequency of the extant CW and SW was again matched; log frequency differed by only 0.05 across conditions. 13While the fact that nonword responses were made with the left (non-dominant) hand makes the interpretation of the response time slowdown more complicated since it requires comparing responses across hands (however see Taft & Forster, Citation1976 among many others for the same result), the high accuracy on this condition provides support to the notion that compound responses were not entirely driven by spotting a morpheme. As for the crucial compounds vs. single words comparison, both responses were made on the dominant (right) hand. One way to address concerns about response hand would be to vary response hand by participant or block; this solution was not utilised in the current study in order to avoid across-participants or across-block analyses of the electrophysiological data (MEG), as across-participants comparisons are non-standard, and number of carefully matched samples should be maximised to achieve the highest signal-to-noise ratio in the MEG responses. 14We also report the results of a replication study with 12 additional participants, in Appendix III, Tables 5A–C. The pattern of results is identical to that of the current experiment, both in a by-participants and in a by-items analysis. Further, these data were reanalysed excluding six single-word items which might be analysed as complex (opaque/bound forms); the patterns were the same. These data are reported in Appendix III, Table 5C. 15Compound nonwords with morphemic constituents consistently elicit long response times (Taft & Forster Citation1976, among others). Van Jaarsveld and Rattink (Citation1988) and others have also shown that a novel compound's lexical status and its interpretability affect response times (see also Coolen, van Jaarsveld, & Schreuder, 1992, among others). 16Juhasz et al. (Citation2003) presume that the relative pervasiveness of second position effects in their studies is because the second constituent position is where constituent and whole-word meanings converge in English (i.e., it is the head position). 17Further, showing constituency effects regardless of transparency would seem to be at odds with the supralexical model, which posits constituent access after initial contact with whole-word representations and only for transparent words; such findings would also present challenges to distributed-connectionist approaches to constituency effects, since those models capture such effects as form-meaning overlaps. Elsewhere we report response time facilitation in masked priming both for constituents of transparent and of opaque lexicalised compound primes (Fiorentino, unpubl. data); for further compound-constituency effects independent of semantic transparency, see also Shoolman and Andrews (Citation2003) (masked priming from constituents to compounds), Zwitserlood (Citation1994) (overt visual priming from compounds to constituents), Libben, Gibson, Yoon, and Sandra (Citation2003) (overt visual priming from constituents to whole compounds) and Pollatsek and Hyönä (Citation2005) (morphemic frequency effects on fixation durations in eye tracking regardless of transparency)." @default.
• W2043968327 created "2016-06-24" @default.
• W2043968327 creator A5014123289 @default.
• W2043968327 creator A5033417964 @default.
• W2043968327 date "2007-11-01" @default.
• W2043968327 modified "2023-10-03" @default.
• W2043968327 title "Compound words and structure in the lexicon" @default.
• W2043968327 cites W1509579795 @default.
• W2043968327 cites W1521440848 @default.
• W2043968327 cites W1942631690 @default.
• W2043968327 cites W1963996585 @default.
• W2043968327 cites W1968340308 @default.
• W2043968327 cites W1968957904 @default.
• W2043968327 cites W1969623577 @default.
• W2043968327 cites W1972098972 @default.
• W2043968327 cites W1972773487 @default.
• W2043968327 cites W1975465191 @default.
• W2043968327 cites W1975909437 @default.
• W2043968327 cites W1976191493 @default.
• W2043968327 cites W1978708528 @default.
• W2043968327 cites W1981029968 @default.
• W2043968327 cites W1981220618 @default.
• W2043968327 cites W1981660002 @default.
• W2043968327 cites W1987544230 @default.
• W2043968327 cites W1989666886 @default.
• W2043968327 cites W1990809698 @default.
• W2043968327 cites W1994864097 @default.
• W2043968327 cites W1995849813 @default.
• W2043968327 cites W1997094127 @default.
• W2043968327 cites W2010222632 @default.
• W2043968327 cites W2013112874 @default.
• W2043968327 cites W2017583241 @default.
• W2043968327 cites W2017929721 @default.
• W2043968327 cites W2019140130 @default.
• W2043968327 cites W2027338973 @default.
• W2043968327 cites W2029697825 @default.
• W2043968327 cites W2030367584 @default.
• W2043968327 cites W2031189361 @default.
• W2043968327 cites W2032439779 @default.
• W2043968327 cites W2037822922 @default.
• W2043968327 cites W2038466489 @default.
• W2043968327 cites W2040295565 @default.
• W2043968327 cites W2042251269 @default.
• W2043968327 cites W2042830081 @default.
• W2043968327 cites W2047662473 @default.
• W2043968327 cites W2048834804 @default.
• W2043968327 cites W2049521441 @default.
• W2043968327 cites W2051991014 @default.
• W2043968327 cites W2052294201 @default.
• W2043968327 cites W2057345692 @default.
• W2043968327 cites W2057585112 @default.
• W2043968327 cites W2058237037 @default.
• W2043968327 cites W2060762641 @default.
• W2043968327 cites W2060779471 @default.
• W2043968327 cites W2061430448 @default.
• W2043968327 cites W2061551065 @default.
• W2043968327 cites W2062521259 @default.
• W2043968327 cites W2067348543 @default.
• W2043968327 cites W2069620859 @default.
• W2043968327 cites W2070586582 @default.
• W2043968327 cites W2073376085 @default.
• W2043968327 cites W2086117364 @default.
• W2043968327 cites W2090459516 @default.
• W2043968327 cites W2094875485 @default.
• W2043968327 cites W2101424182 @default.
• W2043968327 cites W2104996750 @default.
• W2043968327 cites W2105866137 @default.
• W2043968327 cites W2113822177 @default.
• W2043968327 cites W2114057300 @default.
• W2043968327 cites W2115942498 @default.
• W2043968327 cites W2116850456 @default.
• W2043968327 cites W2120025651 @default.
• W2043968327 cites W2121297300 @default.
• W2043968327 cites W2126110252 @default.
• W2043968327 cites W2127284451 @default.
• W2043968327 cites W2128361300 @default.
• W2043968327 cites W2135596603 @default.
• W2043968327 cites W2138102878 @default.
• W2043968327 cites W2138824288 @default.
• W2043968327 cites W2139757867 @default.
• W2043968327 cites W2140178265 @default.
• W2043968327 cites W2141216777 @default.
• W2043968327 cites W2141264999 @default.
• W2043968327 cites W2143851413 @default.
• W2043968327 cites W2146861208 @default.
• W2043968327 cites W2150211557 @default.
• W2043968327 cites W2150634763 @default.
• W2043968327 cites W2154240219 @default.
• W2043968327 cites W2154441655 @default.
• W2043968327 cites W2154717930 @default.
• W2043968327 cites W2155796122 @default.
• W2043968327 cites W2159686064 @default.
• W2043968327 cites W2160383933 @default.
• W2043968327 cites W2160384969 @default.
• W2043968327 cites W2164337345 @default.
• W2043968327 cites W2165392682 @default.
• W2043968327 cites W2170886006 @default.
• W2043968327 cites W2224795691 @default.

• next