Understanding literacy effects on predictive processing in reading

预测即将到来的语言信息的能力被认为是高效语言处理的标志，其根源在于大脑是一个预测引擎，它利用过去的经验知识来预测未来的事件。这在词汇预测的背景下得到了最广泛的研究，在词汇预测中，读者使用他们的知识来预测句子中的下一个单词。例如，当阅读一个以“The cold drink was serve with a slice of…”开头的句子时，读者可能会预料到“lemon”会是下一个单词，而“cucumber”可能会出乎意料，但却是可能的。大量证据表明，熟练的读者受益于预测性处理。这包括眼动研究的发现，表明读者花更少的时间看更可预测的单词，脑电图（EEG）对大脑活动的研究表明，更可预测的单词引起的电生理反应更小。然而，令人惊讶的是，很少有研究对不太熟练的读者进行预测处理，尽管这可能会揭示他们所经历的困难。因此，拟议项目的目的是更充分地了解识字技能在支持语言预测方面的作用。该项目将采用实验方法来解决这个问题，其中一些研究内容将在本科生中进行（如下所述），其他部分将在16-18岁的中学生中进行。后者是一个研究不足的参与者群体，他们应该掌握阅读的机制，但与本科生相比，他们可能拥有更多样化的读写技能。与这群人一起工作将探索研究如何可能导致与阅读教学相关的影响。确定读写能力是否会影响词汇预测的使用是非常重要的。这将涉及使用眼睛测量来比较单词在高度限制和较少限制的句子语境中的识别速度。为了评估读写能力的影响，我们将招募具有广泛阅读能力的参与者，我们将使用词汇知识和理解能力的标准化测试（例如尼尔森-丹尼阅读测试）和阅读习惯来评估。这将使我们能够确定技能较差的读者是否更少使用词汇预测，以及哪些因素（词汇量、理解能力、阅读习惯）预测了这种效果。根据结果，后续实验可能会检验熟练读者和不熟练读者的预测成本是否存在差异。这是指修改错误预测的难度；例如，当预测“柠檬”是下一个单词，却发现用“黄瓜”代替。一种可能是，较不熟练的读者发现更难以修正错误的预测。迄今为止的研究主要集中在词汇预测上。然而，人们对其他形式的预测越来越感兴趣，包括句法预测。这是读者利用他们的知识和语言线索来预测即将到来的语法结构。举个例子，“either”这个词出现在“顾客决定点披萨…”开头的句子中。在这里，“either”提供了一个语言线索，表明句子的结构是一个元素（“披萨”）与另一个元素形成对比。如果这些知识是预言性的，读者可能会期望句子后面有一个“或”的短语，比如“or a calzone”，这提供了一个合理的对比。对熟练读者的研究为这种句法预测的使用提供了证据，表明当使用“either”时，花在“or”-短语上的时间更少。因此，一个进一步的研究问题是，不熟练的读者是否也受益于句法预测。

The ability to anticipate upcoming linguistic information is considered a hallmark of efficient language processing, rooted in the idea that the brain is a prediction engine that uses its knowledge of past experience to predict future events. This has been most widely investigated in the context of lexical prediction, where readers use their knowledge to predict the next word in sentence. For example, when reading a sentence beginning "The cold drink was served with a slice of...", readers might anticipate that "lemon" will be the next word, whereas "cucumber" might be less expected but possible. Substantial evidence shows that skilled readers benefit from predictive processing. This includes findings from eye movements studies showing that readers spend less time looking at more predictable words, and from electroencephalographic (EEG) studies of brain activity showing that more predictable words elicit smaller electrophysiological responses. However, surprisingly little research has examined predictive processing by less skilled readers, although this might shed light on the difficulties they experience.Accordingly, the aim of the proposed project is to more fully understand the role of literacy skills in supporting linguistic prediction. The project will take an experimental approach to this question, with some elements of the research to be conducted with undergraduate participants (as described below) and other elements conducted using a population of 16-18 years old secondary school students. The latter is an under-research participant group who should have mastered the mechanics of reading but likely to have more diverse literacy skills compared to an undergraduate student population. Work with this population will explore how the research might lead to impact in relation to the teaching of reading.It is fundamentally important to establish whether literacy skill influences the use of lexical prediction. This would involve using measures of eye measures to compare the speed with which words are recognised in highly-constraining versus less-constraining sentence contexts. To assess effects of literacy skill, we would recruit participants with a broad range of reading ability, which we would assess using standardised tests of vocabulary knowledge and comprehension ability (e.g., the Nelson-Denny reading test) and reading habits. This would allow us to determine whether less skilled readers use lexical prediction less, as well as which factors (vocabulary, comprehension ability, reading habits) predict the effects. Depending on the results, a follow-up experiment might examine whether there is a difference in the prediction cost for skilled versus less skilled readers. This refers to difficulty revising an incorrect prediction; for example when predicting that "lemon" is the next word and finding that "cucumber" is used instead. One possibility is that less skilled readers find it more difficult to revise an incorrect prediction.Research to date has focused on lexical prediction. However, there is growing interest in other forms of prediction, including syntactic prediction. This is where a reader uses their knowledge and linguistic cues to predict upcoming grammatical structure. Consider as an example a situation in which the word "either" appears within a sentence beginning "The customer decided to order either a pizza...". Here "either" provides a linguistic cue that the sentence has a structure in which one element ("a pizza") is to be contrasted with another. If this knowledge is used predictively, readers might expect the sentence to be continued by an "or"-phrase, such as "or a calzone", that provides a plausible contrast. Research will skilled readers provides evidence for this use of syntactic prediction by showing that less time is spent looking at the "or"-phrase when "either" is used. Therefore, a further research question would be whether less skilled readers also benefit from syntactic prediction.