How variable input shapes word recognition

单词学习是语言发展的核心部分--语言的所有其他方面都在单词的层面上相交。这是一项令人惊讶的复杂任务。单词学习者面临着两个主要的分类问题--除了确定世界上单词应该映射到的对象集之外，他们还必须确定构成相同单词的标记集。这是一个困难的问题，因为一个单词没有两个实例在声学上是相同的。说话者内部和说话者之间的一些差异不会影响单词的意思。然而，当不同口音的说话者说出单词时，也可能发生沿维度的变化，从而改变单词的含义。例如，当一位讲法语的人说出英语单词“PIG”时，由于“b”和“p”在两种语言中的发音方式不同，所以对英语听众来说，它可能听起来像“Big”。

Word learning is a core part of language development - all other aspects of language intersect at the level of the word. And it is a surprisingly complex task. Word learners face two major categorization problems - in addition to determining the sets of objects in the world to which words should be mapped, they must also determine the set of tokens that constitute the same word. This is a difficult problem because no two instances of a word are acoustically identical. Some differences within and across speakers do not affect the meanings of words. However, when words are produced by speakers with different accents, changes can also occur along dimensions that alter the meanings of words. For example, when a French speaker produces the English word “pig”, it may sound to an English listener like the word “big”, because of differences in how the sounds “b” and “p” are produced in the two languages.