Effects of vowel and syllable position on laterals in bilingual speakers of English and Spanish

Date: 23/11/2022

Being interested in sound systems from the perspective of both production and perception, Niamh Kelly ran a project examining the production of /l/ sounds by bilingual speakers of English and Spanish from the El Paso region, to investigate the effects of language dominance on velarisation patterns. She also ran a pilot study where she looked at the production of /z/ sound of a bilingual across time. Here, we had her give us a presentation of the outputs of her work.

Part 1: A bilingual community on the US-Mexico border: what are they doing with their [l]?

Background information:

Transfer can occur in the productions of multilinguals, where one language influences the other and such effects can go in either direction between the L1 and L2. Sometimes, speakers are found to have productions that are intermediate between the two languages. In some regions, the whole community is bilingual, making it convenient to look at language transfer effects. 

Although similar to each other, the /l/ sounds in American English (AmE) and Spanish are not exactly matched up. While /l/ sound in Spanish is realised as fronted (light/clear) /l/, in AmE it is more velarised overall, especially when it is in codas. 

The participants in this research lived in a city (El Paso) on the US-Mexico border, which is a bilingual community.

Research questions:

This research asks:

  1. To what extent do balanced bilingual speakers show transfer effects in laterals? That is, are there positional effects in just English or in both English and Spanish?
  2. What effects of vowel height and front/backness have on velarisation in laterals in both languages?


Since these speakers are balanced bilinguals, they could be expected to keep their languages separate: English /l/ would be more velarised overall than Spanish /l/, and that in English, coda /l/ would be more velarised than onset /l/, while in Spanish no such difference would occur. 


From the analysis of the participants’ production in both English and Spanish the following results emerged:

  1. English and Spanish were significantly different in both positions. /l/ was more velarised in English than Spanish in both onset and coda position.
  2. /l/ was more velarised in codas than onsets in English while no such positional difference emerged in Spanish.

 Next steps:

Research like this and more further research can help in adding to the description of non-mainstream varieties of English and varieties used by multilinguals. 

Part 2: A bilingual across time: what happened to his /z/? Acquisition of voicing in English /z/ by an L1 Norwegian speaker in a 25-year period.

Background information:

The English /s/ – /z/ contrast has been found to be difficult to acquire for L2 English speakers who do not have this contrast in their L1. Norwegian-accented English has a lack of voicing in /z/ since Norwegian does not have /z/.

Current study:

This study is a longitudinal study of the L2 English of L1 Norwegian speaker Ole Gunnar Solskjær. Ten interviews from two time periods, 1996-8 and 2021 were examined, focusing on his English productions of /s/ and /z/ and how production patterns change over time. There variables were coded: position in word (medial or final), preceding segment (voiced or voiceless), and morphemic status (morphemic, e.g., ‘goals’ vs stem, e.g., ‘please’).


  1. In the Early timeframe, 100% of /s/ tokens were voiceless and 93% of /z/ tokens were voiceless. In the Late timeframe, 98.5% of /s/ tokens were voiceless (no significant effect of timeframe) and 46% of /z/ tokens were voiceless (a significant effect of timeframe).
  2. Duration was longer when voiceless (supporting the auditory categorisation) but not affected by position in word.
  3. No difference based on morphemic status.

Discussion and next steps:

More exposure to and practise with the L2 has led to an increase in L2-like voicing productions. OGS is acquiring a new voicing contrast, but has not acquired it completely as only about half of /z/ tokens were voiced. More work can be done to look at other fricatives and also the intermediate time frame. 

General conclusion:

  1. Here we find transfer of L1 phonetic and phonological patterns to L2 at the individual level, which can continue even after years of exposure and use. 
  2. It also occurs on a larger scale when a community is bilingual.
  3. It is important for linguists to describe non-mainstream varieties. 

Leave a Reply

Your email address will not be published. Required fields are marked *