The thesis is an attempt to apply procedures of descriptive linguistic analysis to a body of material phonetically transcribed by Leo J. Frachtenberg in the Mary's River dialect of the central Kalapuyan language. In 1913-14 Frachtenberg collected thirteen volumes of myth texts in Mary's River, twelve from William Hartless and one from Grace Wheeler; in addition, there were notes to the texts, three volumes of grammatical notes, and some ethnographic material. The phonetic transcription was carried out in the pre- phonemic tradition of recording everything the speaker said as accurately as possible. There was no attempt to elicit forms in a manner which would establish which sounds were "the same" to the speaker, resulting in a considerable proliferation of variations of forms. The main methodological problem, therefore, has been to sort out phonetic variation and phonemic contrast. This task was more difficult because no complete morphemic analysis exists as yet, so that morphophonemic alternations created additional complications. My method consisted first of tallying the forms that occurred, to establish those variants found most frequently (modal forms). The modal forms were examined to discover patterns of contrast and complementary distribution; those variants of forms which were not modally distributed, and single examples, were then compared to help confirm or modify the emerging phonemic patterns. I arrived at a system of 21 consonants, six vowels, four diphthongs, and phonemic stress. This includes two obstruent series, glottalized and unglottalized, five fricatives, three resonants; and two glides. The vowels are high to mid front, low center, and high to mid back, with length distinctive. One pair of diphthongs is also distinguished by length; the others involve non-phonemic length. Consonant clusters are limited to two members initially and finally. A number of uncertainties remain. These include the behavior of glottal stops and aspiration in final and medial position, the variation in vowel length, the distribution of velars, the distribution of diphthongs, and the relationships between /w/ and / u / , /y/ and /i/, and /f/ and /w/. These may be clarified when the morphemic analysis is complete. The phonemic system appears to resemble most closely that of the neighboring Molala, although the lexicon and grammar show only slight similarities; the related Takelma also seems quite similar phonologically. Since phonological features can be borrowed by one language from another, and in the Northwest frequently have been, further examination of such resemblances may shed light on former historical contacts between Kalapuyan speakers and other groups.

Archaeologists grapple with the problematic nature of archaeological discovery. Certain types of sites are difficult to see even in the best environmental conditions (e.g., low-density lithic scatters) and performing traditional archaeological survey is challenging in some environments, such as the dense temperate rain forests of the Pacific Northwest. Archaeologists need another method of survey to assess large areas and overcome environmental and archaeological barriers to site discovery in regions like the Pacific Northwest. LiDAR (light detection and ranging) technology, a method for digitally clearing away swaths of vegetation and surveying the landscape, is one possible solution to some of these archaeological problems. The Calapooia Watershed in the southern Willamette Valley in Oregon is an ideal area to focus LiDAR's unique archaeological capabilities, as the region is heavily wooded and known to contain hundreds of low-lying earthwork features or mounds. Modern Indigenous Communities, such as the Confederated Tribes of Grand Ronde, consider the Willamette Valley mound sites highly sensitive locations, as ethnographic accounts and limited archaeological work indicate that some are burial sites. However, these mounds have received little archaeological study. Land ownership (94 percent privately owned), dense vegetation that obscures mounds, and the sheer expanse of the landscape (234,000 acres) have impeded professional archaeological research. The focus of this thesis is the development and the testing of a LiDAR and remote sensing predictive model to see if this type of model can detect where potential mound sites are located in the Calapooia Watershed, Oregon. The author created a LiDAR and remote sensing predictive model using ArcMap 10.5.1, LiDAR, and publicly available aerial imagery; manipulating data using standard hydrological tools in ArcMap. The resulting model was successful in locating extant previously identified mound sites. The author then conducted field work and determined that the model was also successful in identifying seven new, previously unrecorded mound sites in the watershed. The author also identified several possible patterns in mound location and characteristics through exploratory model analysis and fieldwork; this exploratory analysis highlights areas for future mound research. This project has clearly established a method and a model appropriate for archaeological mound prospection in the Willamette Valley. This project also shows the efficacy of LiDAR predictive models and feature extraction methods for archaeological work, which can be modified for use in other regions of the Pacific Northwest and beyond. Furthermore, by identifying these mounds I have laid the groundwork for future studies that may continue to shed light on why and how people created these mounds, which will add valuable information to a poorly understood site type and cultural practice.