Language and Speech

The verb-particle construction in English has two forms when occurring with a nominal direct object. The particle can precede the object, e.g., look up the information, or follow it, e.g., look the information up. Experimental evidence based on 224 subjects is presented showing that the particle tends to be contiguous to the verb when the verb-particle combination is semantically or phonologically cohesive, i.e., when the combination is idiomatic or when the particle begins with a vowel. A model within the framework of variable rules is proposed to handle the data.

Three experiments tested the proposition that reference assignment is delayed, that is, interpretation is not completed, until some point after the reference term has been read. Subjects were presented with sentences containing either pronominal or elliptical references. The point at which pragmatic inferences could be used was manipulated. The results showed that subsequent content influenced both assignment times (experiments one and two) and reading times (experiment three). This result held for unambiguous elliptical references as well as for both ambiguous and unambiguous pronouns. The results support the proposal that assignment is delayed, allowing subsequent content to influence the process. It is suggested that delayed assignment occurs whenever information needed to complete the processing must be retrieved from elsewhere in the sentence. The results are also contrasted with the view that pronoun assignment is based on heuristic strategies.

Use of a new opto-elcctronic device showed that the stressed vowels produced by a native speaker of English could be clearly differentiated according to tongue position. The distance of the tongue from four sensors located along the hard palate was measured in multiple tokens of vowels in a /b_b/ context. There was little token-to-token variability (standard deviations of less than 1.0 mm). With the exception of a reversal between the onglide of/e¹/ and /∊/, the height of the tongue differed according to phonological height as expected. The tongue was lower in back vowels than corresponding front vowels, suggesting a biomechanical linkage between tongue height and frontness-backness. Schwa differed from all other vowels, from the “average” tongue configuration seen in stressed vowels, and from the “neutral” configuration in pauses. The talker, a Southerner, diphthongized /I/ and /∊/, but showed relatively little tongue movement in /a^U/(compared to that in /a^I/ and /O^I/). Relatively little tongue movement was also seen in the /a^U/ spoken by a second talker, who showed a significantly greater change in upper lip position in producing /a^U/ than in /a^I/ or /o^I/. This finding suggests that lip rounding and tongue movement may “trade off” as articulatory means for producing the rapid spectral change in diphthongs.

A group of five male right-hemisphere-damaged and five male left-hemisphere-damaged aphasics were recorded repeating a set of three phonemic tone-contrasted minimal pairs in Norwegian. A narrow-band spectrographic analysis revealed more impaired tone production (less phono-acoustic differentiation between phonemic pairs, and longer maximal productions) on the part of the left-hemisphere group than the right-hemisphere group, although the productions of the right-hemisphere group also seemed to be perturbed. A perceptual test of the patients' recorded productions by a native speaker revealed more lexical errors due to poor tone contrasts on the part of the left-hemisphere-damaged group. These results are discussed and compared to previous studies of linguistic tones in the neurological literature.