Morphological Processing of Chinese and English Compound Words: A Masked Priming Investigation

<p dir="ltr">This thesis compares the morphological processing of compound words in Chinese, a logographic language, and English, an alphabetic language to shed light on language-universal and language-specific mechanisms in early visual word recognition. To this end, three studies, each with separate experiments on Chinese and English, were conducted using the masked priming paradigm combined with a lexical decision task.</p><p dir="ltr">Chapter 2 used the masked constituent priming paradigm (e.g., foxhunt-FOX), a wellestablished tool to study the automatic morpheme decomposition process. Following previous studies, we assessed the priming effect of transparent and opaque compound word primes and constituent morpheme targets in the initial position (e.g., foxhunt/foxtrot-FOX) and the final position (e.g., voicemail/blackmail-MAIL) relative to an unrelated prime; additionally, we used an identity prime (e.g., fox-FOX) as another baseline to measure the cost of morpheme decomposition process. Both the English and Chinese experiments yielded compound priming effects that were independent of morpheme position and semantic transparency. Importantly, they were greater in Chinese than in English when assessed against the unrelated prime, and the cost of decomposition was smaller when assessed against the identity prime. These results support a universal morphological decomposition process primarily driven by morpho-orthography, which is impacted by the greater ease of morphological parsing in Chinese than in English.</p><p dir="ltr">Chapter 3 examined whether the morphological structure of a word modulates the masked transposed-constituent (TC) priming effect, i.e., target compounds are recognized faster when preceded by a masked prime generated by reversing its morphemes compared to an unrelated prime (e.g., hornfog-FOGHORN < risepad-FOGHORN). We used transparent compounds (e.g., FOGHORN) and pseudocompounds (e.g., CARPET) as targets and assessed the effects of TC primes (e.g., hornfog; petcar) against the unrelated prime (e.g., risepad; bugson), and the identity prime (e.g., foghorn; carpet). In both English and Chinese experiments, equivalent facilitative masked TC priming effects were observed for pseudocompounds and transparent compounds, indicating that they were not modulated by whether the word possesses a real- or pseudo-compound structure. As with the constituent priming effect studied in Chapter 2, masked TC priming effect (relative to unrelated prime), and the discrepancy relative to identity prime was larger and smaller in Chinese than in English, respectively, consistent with the greater ease of morpheme parsing in compound words written in the morpho-syllabic vs. alphabetic scripts.</p><p dir="ltr">Chapter 4 examined the access to morpheme meaning in compound word processing. We reported three morpho-semantic priming experiments, one in Chinese with masked primes, two in English with masked and visible primes. In each experiment, transparent (e.g., butterknife) and opaque (e.g., butterfly) compounds were used as primes and the semantic associates of their shared first constituent morpheme as targets (e.g., BREAD). Priming effect (assessed against an unrelated prime) is taken as evidence for access of meaning of the embedded morpheme. Morpho-semantic priming effect was observed for transparent and not for opaque compounds in Chinese with masked primes and in English with visible primes, but completely absent in English with masked primes. The results indicate that morpheme meanings are only accessed for transparent but not for opaque compounds; and it occurs earlier in Chinese than in English, which is attributed to the fact that Chinese characters represent morphemes, unlike alphabetic scripts.</p>