Neighbourhood density and frequency effects in speech production: A case for interactivity

Abstract

In three experiments, we explore the effects of phonological properties such as neighbourhood density and frequency on speech production in Spanish. Specifically, we assess the reliability of the recent observation made by Vitevitch and Stamer (2006), according to which the neighbourhood effect in Spanish has a reverse polarity to that observed in other languages. In Experiment 1, we replicate Vitevitch and Stamer's (2006) experiment, this time adding a control group. The same inhibitory neighbourhood effect found for both groups can not corroborate the hypothesis posited by Vitevitch and Stamer. In Experiment 2, our results show that native speakers of Spanish named pictures with words belonging to high density neighbourhoods faster than those belonging to low density neighbourhoods. In Experiment 3, we test for effects of neighbourhood frequency during lexical selection. Again, we find a facilitatory effect for words with a high-frequency neighbourhood. Together, the results of the present experiments suggest that lexical selection is facilitated by the number of neighbours and by neighbourhoods with higher frequency. These findings are consistent with the predictions of interactive models.

Acknowledgements

This research was supported by a fellowship (BES-2005-8295) and two grants from the Ministerio de Ciencia y Tecnología (SEJ2004-07680-C02-02 and SEJ2006-09238/PS) to Cristina Baus and Manuel Carreiras and two grants from the Spanish Government (SEJ200500409/psic; CONSOLIDER) to Albert Costa.

We thank Michael Vitevitch and Melissa Stamer for sending us the materials used in Experiment 1. We thank Margaret Gillon Dowens for her helpful comments on the manuscript. We are very grateful to Begoña Diaz and Anna Mestres for their help with the experiments and Oliver Müller and Markus Conrad for their help with the preparation of the German stimuli.

Notes

¹Previous studies in comprehension have shown that neighbourhood frequency modulates recognition latencies (Carreiras, Perea, & Grainger, 1997; Grainger, 1990), showing an inhibitory neighbourhood frequency effect: those words with high-frequency neighbourhoods were recognised slower than those with low-frequency neighbourhoods.

²From the original materials of Vitevitch and Stamer's study, we removed four pictures in each condition, since there was some dialectical variation in the names of the pictures or the names selected for the pictures in the original experiment were not be the same as those used in Spain to name the same pictures. The name agreement for these words was lower than 60%. Therefore, we decided not to use them as experimental items in Experiment 1. For example, the picture of a ‘hose’ is named as ‘manga’ in Latin America while in Spain it is named as ‘manguera’. The exclusion of these words did not substantially affect the distribution of word frequency and neighbourhood frequency between the two sets of words.

³The same analyses were carried out, considering only those words analysed in Experiment 2 (n = 39); similar results were found: no statistically significant differences were found between sparse (471.5 ms) and dense (481.5 ms) conditions.