Languages and Tools

Languages that we have worked with





Potsdam, Germany

Potsdam, Germany









Brazilian Portuguese




Tools that we work with

Eye Tracking

Large-scale Rating Studies

Self-paced Reading


Truth-value Judgment

Speed-Accuracy Tradeoff (SAT)




Computational Modeling

Many Languages

Many Tools

Language diversity creates a playground for our research.

English is the most convenient language for us, of course. But often our most interesting findings come from studies on other languages. For this, it is invaluable to be doing psychological research in a linguistics setting, with an team drawn from around the world.

Our experimental research has featured Basque, Bengali, Brazilian Portuguese, Chinese, English, German, Hebrew, Hindi, Italian, Japanese, Korean, Russian, Spanish, and Swedish.

Cross-language comparisons allow us to construct tests that would be impossible in a single language. Just a few examples:

  • Locality: we can distinguish linear and structural notions of locality by comparing syntactic dependencies in English, Japanese, and Bengali, e.g., Aoshima et al. 2004; Omaki et al. 2014; Chacón et al. 2014
  • Levels of encoding: we can distinguish phonetic and phonological contributors to brain measures of speech perception by comparing Russian and Korean, e.g., Kazanina et al. 2006
  • Memory: We gain a better understanding of memory access mechanisms by comparing real-time reflexive licensing in English, Hindi, and Chinese, e.g., Dillon et al. 2013; Dillon et al. 2014; Kush & Phillips 2014

Some cross-language studies entail travel. But the Washington DC area provides a rich pool of language diversity. Currently our eye-tracking and EEG studies are limited to those that we can do in the lab, but soon we will have new equipment that will make it possible to take these measures to remote locations.

Our shared lab environment gives us access to more or less any experimental tool that we could wish for. Different methods are useful for addressing different questions, and I hold on to my wallet when somebody claims that their favorite technique is the best.

Our methods range from the very low-tech to the very high-tech, including advanced neuroscientific and computational tools. Some guiding principles:

  • Simple is good. It’s nice to have access to the fanciest measures, but we can often learn more, at a much lower cost in time and money, by using low-tech methods. If a question can be addressed using EEG (hard!) or speeded acceptability judgments (easy!), the choice is clear.
  • Don’t measure height with a stopwatch. The measure needs to fit the question. Millisecond-precision recordings are useful if you’re testing a time-based hypothesis; brain localization is not so useful if you don’t have neuroanatomical hypotheses. This basic guideline is often overlooked, e.g., Phillips & Wagers 2007; Phillips & Parker 2014. Video from MIT@50 talk, 2011.
  • Creative conflicts. It’s nice to find convergence between studies using different measures, but it’s far more interesting when we don’t, as this forces us to re-examine what the different measures are telling us. E.g., Lau et al. 2008; Phillips 2010 (CUNY Conf. talk)

Since I cannot reasonably be accused of shying away from systematic experimentation myself, I have happily taken the role of staunch defender of simple, traditional methods in linguistics, e.g., Phillips 2010. I believe that the field faces important challenges, but it is fantasy to think that this would be solved by simply running large scale experiments in all cases.