Double modals are a well-known non-standard feature of some regional varieties of English in North America, but due to their rareness in spoken language, questions remain as to the inventory of possible combinatorial types and the geographic extent of their use in contemporary naturalistic speech. This study investigates double modals in the Corpus of North American Spoken English (CoNASE), a 1.2-billion-word corpus of time-stamped and geolocated automatic speech recognition (ASR) YouTube transcripts from the United States and Canada. Double modal sequences were identified in the corpus using regular expressions, then verified via manual examination of videos. The study represents the first large-scale, continent-wide analysis of double modals based entirely on recent naturalistic production data, rather than data such as elicited responses or sentence acceptability judgments, and it demonstrates a larger double modal inventory and a broader geographic range of use for the feature than has previously been documented, including in Canada.

中文翻译：

---

北美的自然主义双模态

双模态是北美一些地区英语变体的一个众所周知的非标准特征，但由于它们在口语中的稀有性，关于可能的组合类型的清单以及它们在当代自然主义中使用的地理范围仍然存在疑问演讲。这项研究调查了北美口语语料库 (CoNASE) 中的双模态，这是一个 12 亿字的带有时间戳和地理定位的自动语音识别 (ASR) 来自美国和加拿大的 YouTube 成绩单的语料库。使用正则表达式在语料库中识别双模态序列，然后通过手动检查视频进行验证。该研究代表了第一次完全基于最近的自然生产数据对双模态的大规模、全大陆分析，