How Many Rooms in an Octopus Apartment? Effects of Prosody on Conceptual Combination

How Many Rooms in an Octopus Apartment? Effects of Prosody on Conceptual Combination Dermot Lynott (dermot.lynott@northumbria.ac.uk) Louise Connell (louise.connell@northumbria.ac.uk) Cognition & Communication Research Centre, Division of Psychology, Northumbria University Newcastle upon Tyne, NE1 8ST, UK Abstract Previous theories of conceptual combination have failed to address the possible role of suprasegmental factors such as prosodic emphasis patterns. Here, we investigated the effect of emphasising the initial word (modifier emphasis), both words (dual emphasis), or the final word (head emphasis) on people’s interpretations of novel noun-noun compounds (e.g., octopus apartment). We found that dual emphasis alone affects the frequency of production and speed of property- and relation- based interpretations. People produced more relation-based (e.g., an octopus apartment is where an octopus lives) than property-based (e.g., an octopus apartment has eight rooms) interpretations when compounds were presented with dual emphasis than in written form. Additionally, dual emphasis caused people to arrive at property-based interpretations more rapidly than relation-based interpretations. Findings are discussed with reference to existing theories of conceptual combination. Introduction Octopus apartment, latte crowd and snail mail are all examples of coined noun-noun compounds generally referred to as concept combinations. These novel combinations reflect a fundamental aspect of language generativity, accounting for between 30% and 60% of new terms in English (Cannon, 1987; McFedries, 2004). It often happens that such compounds survive to become permanent fixtures of the language and used everyday by the wider language community (e.g., soccer mom, camera phone). Others may only be used in one specific context and not become part of the language as a lexicalised term in its own right (e.g., daisy cup as “a cup with a daisy pattern on it”). With these phrases occurring everywhere from newspaper headlines to advertisements and novels, it is clear that they represent an important aspect of language use and growth. Over the past 25 years, there has been a wealth of research into the comprehension of concept combinations, with a number of different theoretical positions proposed. Most recent research acknowledges that for any given novel compound there can be many possible interpretations (e.g., constraint theory, Costello & Keane, 2000; Competition Among Relations in Nominals: CARIN, Gagne & Shoben, 1997; Producing and Understanding Novel Compounds: PUNC, Lynott et al., 2004; concept specialisation, Murphy, These theories generally hold that the conceptual combination process involves the manipulation of properties and relations between the head (first concept) and the modifier (second concept). For example, the PUNC model (Lynott et al, 2004) interprets the compound cactus beetle by moving the feature “has spikes” from the modifier to the head, giving rise to the property-based meaning “a beetle that has spikes”. Alternatively, the relation “eats” can be placed between the constituent nouns to give rise to the relation- based meaning “a beetle that eats cacti” (e.g., Lynott, et al., 2004; Gagne, 2000). Where conceptual combination theories differ is in the relative importance they place each of the nouns that make up the compound. The CARIN model (Gagne, 2000; Gagne & Shoben, 1997) argues for the primacy of the modifier, showing that the frequency with which a particular relation (e.g., located, made-of) is associated with a modifier noun can predict the speed with which people use that relation in an interpretation (e.g., if the relation [located] is frequently associated with the noun daisy then daisy cup will be easily interpreted as “a cup containing daisies”). In contrast, concept specialisation theory (Murphy, 1990) leans in favour of primacy of the head, contending that the modifying noun acts to specialise the representation of the head noun. On the other hand, constraint theory argues for equal primacy of head and modifier (Costello & Keane, 2000), with both concepts given equal weight in the search for the best interpretation of a given compound. Finally, the PUNC model (Lynott et al., 2004) differs from all the above by arguing that that neither constituent noun has a priori primacy in the interpretation process, and that relative importance is dependent on each individual concept and that concept’s internal structure. Clearly, there is little agreement on the relative importance of the constituent concepts in a noun- noun compound; an issue that needs to be resolved if we are to understand how people perform this complex microcosm of language comprehension. Prosody and Conceptual Combination None of the cognitive theories of conceptual combination in the literature to date have specified a role for suprasegmental information (e.g., prosodic information) in the comprehension of novel noun-noun compounds. This is a curious omission in light of the fact that general linguistic theories of language have noted the meaning-altering effects of prosody for a long time (Bresnan, 1971; Ladd, 1996), including effects on lexicalised compound phrases (e.g., Isel, Gunter & Friederici, 2003). Furthermore, the frequency with which novel noun phrases occur in the English language only adds to mystery of this oversight. Broadly speaking, prosody refers to changes in aspects of speech such as emphasis, pitch, intonation, rhythm and timing. It is commonly recognised that changes in prosody can affect meaning. For example, words such as contract and object change meaning depending on which syllable emphasis is placed. In investigating the possible effects of